Reliability of Mercury and Arsenic Sampling and Analysis

Reliability of mercury and arsenic sampling and analysis
in the petrochemical and oil industry

Aurore Méré
To cite this version:

Aurore Méré. Reliability of mercury and arsenic sampling and analysis in the petrochemical and oil
industry. Analytical chemistry. Université de Pau et des Pays de l’Adour, 2022. English. �NNT :
2022PAUU3024�. �tel-04085642�
HAL Id: tel-04085642

https://theses.hal.science/tel-04085642
Submitted on 29 Apr 2023
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
THESE
présentée à
L’Université de Pau et des Pays de l’Adour

Ecole Doctorale des Sciences Exactes et de leurs Applications
par
Aurore MERE
pour l’obtention du grade de
DOCTEUR
Spécialité :
Chimie Analytique
La fiabilité de l’échantillonnage et de l’analyse du mercure et de

l’arsenic dans l’industrie pétrochimique et pétrolière
Date de soutenance : 04 juillet 2022
Composition du jury :
Charles-Philippe Lienemann Rapporteur Ingénieur de recherche, IFPEN-Lyon
José-Luis Todoli Rapporteur
Florence Pannier Présidente du jury Professeur, Université de Pau et des Pays

de l’Adour
Stuart Baker Examinateur
Jorg Feldmann Examinateur
Brice Bouyssiere Directeur de thèse Professeur, Univ. de Pau et des Pays de
l’Adour
Honggang Zhou Responsable industriel Référent Thermodynamique,
TOTALEnergies
Emmanuel Tessier Co-directeur de thèse Ingénieur de recherches, IPREM
Maxime Enrico
Remerciements
3
Table des matières
Table des figures .................................................................................................................................... 9

Table des tableaux ............................................................................................................................... 12
Table des abréviations......................................................................................................................... 13
INTRODUCTION GENERALE ........................................................................................................ 16
CHAPITRE I : ETAT DE L’ART ..................................................................................................... 18
Partie 1 : Total and speciation analyses of mercury in the crude oil industry: a review .............. 19
Abstract ............................................................................................................................................ 19
1. Introduction ............................................................................................................................. 20
2. General Constraints Related to the Sampling and Quantitative Analysis of Mercury ..... 21
2.1. Natural Gas ........................................................................................................................ 21
2.1.1. Hg Adsorption ........................................................................................................... 22
2.1.1.1. Current Knowledge ............................................................................................... 22
2.1.1.2. Thermodynamic Considerations ............................................................................ 22
2.1.2. Implication of Adsorption Problems for New Pipes ................................................. 23
2.1.3. Implication of Adsorption Problems for the Sampling System ................................. 23
2.2. Liquid Petroleum Products ................................................................................................ 24
2.2.1. Volatile Hg Loss ........................................................................................................ 24
2.2.2. Ionic Hg Adsorption on Container Walls .................................................................. 25
2.2.3. Precipitation of Hg-Containing Minerals .................................................................. 25
2.2.4. Sample Homogeneity ................................................................................................ 26
3. Mercury Analysis in Natural Gas .......................................................................................... 27
3.1. Sampling method ............................................................................................................... 27
3.1.1. Bubbling System ....................................................................................................... 27
3.1.2. Hg-Selective Sorption ............................................................................................... 30
3.1.3. Gold Amalgamation .................................................................................................. 30
3.1.4. Gas Samplers: Cylinders and Tedlar Bags ................................................................ 32
3.2. Hg Analysis ....................................................................................................................... 32
3.3. Direct In Situ Analysis ...................................................................................................... 38
4. Mercury Analysis in Hydrocarbon Liquids .......................................................................... 38
4.1. Total Hg Analysis .............................................................................................................. 38
4.1.1. Preparation................................................................................................................. 38
4.1.1.1. Liquid-liquid Extraction ........................................................................................ 38
4.1.1.2. Sample Digestion ................................................................................................... 39
4
4.1.1.3. Thermal Decomposition ........................................................................................ 39
4.1.1.4. Other Methods ....................................................................................................... 40
4.1.2. Analysis ..................................................................................................................... 41
4.2. Hg Speciation .................................................................................................................... 42
4.2.1. Sample Preparation: Operationally Defined Procedures ........................................... 45
4.2.2. Gas Chromatography ................................................................................................. 46
4.2.3. HPLC Separation ....................................................................................................... 47
4.2.4. Gel Permeation Chromatography .............................................................................. 48
5. Challenges and Perspectives ................................................................................................... 49
6. Conclusion ................................................................................................................................ 50
References ........................................................................................................................................ 51
Partie 2: Arsenic analysis in the petroleum industry: a review ...................................................... 57
Abstract ............................................................................................................................................ 57
1. Introduction ............................................................................................................................. 58
2. Sampling and Analysis of As in Natural Gas ........................................................................ 59
2.1. Sampling ............................................................................................................................ 59
2.1.1. As-Selective Sorption ................................................................................................ 59
2.1.1.1. Bubbling systems ................................................................................................... 59
2.1.1.2. Traps ...................................................................................................................... 60
2.1.2. Gas Samplers: Cylinders and Inert Plastic Bags (Tedlar Bags) ................................ 60
2.2. Analysis ............................................................................................................................. 61
2.2.1. Direct Analysis .......................................................................................................... 61
2.2.2. Analysis after preconcentration ................................................................................. 62
3. Sampling and Analysis of As in Hydrocarbon Liquids ........................................................ 63
3.1. Sample Preparation............................................................................................................ 63
3.1.1. Microemulsions ......................................................................................................... 63
3.1.2. Extraction, Mineralization and Digestion .................................................................. 64
3.1.3. Dilution and Direct Injection ..................................................................................... 65
3.2. Analytical Methods ........................................................................................................... 66
3.2.1. Analysis by GFAAS .................................................................................................. 66
3.2.2. Analysis by ICP-MS .................................................................................................. 67
3.2.3. Analysis by other methods ........................................................................................ 67
4. Sampling and Analysis of As in the New Energy Feedstocks Domain................................ 70
5. Conclusion ................................................................................................................................ 70
References ........................................................................................................................................ 71
5
CHAPITRE II : FIABILITE DE L’ECHANTILLONNAGE ET DE L’ANALYSE DU
MERCURE DANS LE GAZ NATUREL .......................................................................................... 76
Partie 1 : Experimental Tests of Natural Gas Samplers Prior to Mercury Concentration
Analysis ............................................................................................................................................ 77
Abstract ............................................................................................................................................ 77
1. Introduction ............................................................................................................................. 78
2. Experimental Section .............................................................................................................. 80
2.1. Samplers ............................................................................................................................ 80
2.2. Generation of Hg-Contaminated Gas ................................................................................ 81
2.3. Subsampling for the Hg Concentration Measurement....................................................... 82
2.4. Thermal Desorption Experiments...................................................................................... 82
2.5. Hg Analysis ....................................................................................................................... 83
3. Results and Discussions........................................................................................................... 83
3.1. Initial Gaseous Hg Concentration and Temporal Stability ................................................ 83
3.2. Adsorption of Gaseous Hg and Equilibrium ..................................................................... 85
3.3. Thermal Desorption Results .............................................................................................. 87
3.4. Mass Balance and Reaction Rates ..................................................................................... 89
4. Conclusion and Recommendations ........................................................................................ 90
5. Complete information ............................................................................................................. 91
5.1. Determination of Hg concentration ................................................................................... 91
5.2. Thermal desorption of PTFE-coated and uncoated cylinders............................................ 92
Acknowledgements .......................................................................................................................... 94
References ........................................................................................................................................ 94
Partie 2 : Essai expérimental d’échantillonnage de gaz naturel contaminé en Hg0 dans un sac
Tedlar ................................................................................................................................................... 96
1. Matériels et Méthodes ............................................................................................................. 96
1.1. Préparation et échantillonnage. ......................................................................................... 96
1.2. Analyse du Hg0 .................................................................................................................. 97
2. Résultats et discussions ........................................................................................................... 97
3. Conclusion ................................................................................................................................ 98
Partie 3 : Recommandations sur l’échantillonnage et de l’analyse du mercure dans le gaz naturel
au cours d’un essai de puits ................................................................................................................ 99
1. Normes existantes et leurs insuffisances ................................................................................ 99
2. Problèmes spécifiques sur les sites industriels ainsi que durant les tests de puits ........... 100
3. Relation entre la durée du test et le seuil de détection ....................................................... 100
4. Traque et élimination des artéfacts liés à la sélection et la préparation du point
d’échantillonnage........................................................................................................................... 104
6
5. Règles depuis l’échantillonnage jusqu’à l’analyse afin d’observer aucune perte et
contamination de l’échantillon ..................................................................................................... 105
CHAPITRE III : FIABILITE DE L’ECHANTILLONNAGE ET DE L’ANALYSE DE
L’ARSENIC DANS LE GAZ NATUREL ....................................................................................... 110
1. Matériels et Méthodes ........................................................................................................... 111
1.1. Echantillonnage ............................................................................................................... 111
1.2. Procédures d’analyse ....................................................................................................... 111
1.2.1. Mesure de la concentration en arsenic total par système de barbotage ................... 111
1.2.2. Analyse des condensats de gaz contenus dans les cylindres ................................... 113
1.2.3. Expérience de désorption thermique ....................................................................... 114
2. Résultats et discussions ......................................................................................................... 114
2.1. Mesure de la concentration en arsenic total par système de barbotage ........................... 114
2.2. Analyse des condensats de gaz contenus dans les cylindres ........................................... 115
2.3. Expérience de désorption thermique ............................................................................... 118
3. Conclusion .............................................................................................................................. 119
CHAPITRE IV : COMPREHENSION DU PHENOMENE D’ADSORPTION DE L’ARSENIC
ET DU MERCURE SUR LA SURFACE INTERNE DES CYLINDRES.................................... 120
1. Matériels et méthodes................................................................................................................ 121
1.1. Détermination de la concentration en arsenic total dans le biogaz du site de Mendixka .... 121
1.2. Détermination de la concentration en mercure total dans le biogaz du site de Mendixka .. 122
1.3. Analyse de surface des plaques représentant expérimentalement les cylindres
d’échantillonnage ............................................................................................................................ 123
1.3.1. Prélèvement du biogaz dans le réacteur ...................................................................... 123
1.3.2. Analyse de surface par LA-ICP-MS ............................................................................ 124
1.3.3. Analyse de surface par SEM-EDX .............................................................................. 125
2. Résultats et discussions ............................................................................................................. 126
2.1. Détermination de la concentration en arsenic total dans le biogaz du site de Mendixka .... 126
2.2. Détermination de la concentration en mercure total dans le biogaz du site de Mendixka .. 127
2.3. Analyse de surface des plaques représentant expérimentalement les cylindres
d’échantillonnage ............................................................................................................................ 128
2.3.1. Analyse de surface par LA-ICP-MS ............................................................................ 128
2.3.2. Analyse de surface par SEM-EDX .............................................................................. 133
3. Conclusion .................................................................................................................................. 136
CONCLUSION GENERALE ET PERSPECTIVES ..................................................................... 137
Références (hors publications) ............................................................................................................ 140
Annexes .............................................................................................................................................. 142
1. Annexe 1 : Analyse par SEM-EDX du site 2 à 4 de la plaque Silconert® ........................ 142
7
2. Annexe 2 : Analyse par SEM-EDX du site 2 à 5 de la plaque Dursan®........................... 145
8
Table des figures
Figure 1. Schematic presenting the different methodologies for Hg quantification in natural

gas, from sampling (top boxes) to sample processing (middle boxes) and detection (right
boxes) ....................................................................................................................................... 31
Figure 2. Summary of the different methodologies for total Hg analysis in liquid petroleum,
from sample treatment (left boxes) to processing .................................................................... 40
Figure 3. Diagram of mercury speciation analysis in liquid hydrocarbon by UOP Method 938-
10 (adapted with permission from ref 5). ................................................................................. 46
Figure 4. Diagram of Arsenic Analysis in Natural Gas ........................................................... 63
Figure 5. Schemes of the systems used for (A) loading of Hg-contaminated gas in the
cylinders and (B) thermal desorption experiments. ................................................................. 79
Figure 6. Initial Hg concentration measured after sampler loading with Hg-contaminated

argon. Are shown only the initial Hg concentration for experiments including a 30 min
preliminary flushing step during the loading experiment. The results for 6 different samplers
are shown, as well as for the different tests conducted on each sampler. ................................ 81
Figure 7. Gaseous Hg concentration variations over time during stability experiments

conducted on (A) a new silicon-coated cylinder, (B) the old silicon-coated cylinders A and
(C) B, (D) the PTFE-coated cylinder and (E) the Tedlar bag. Note the differences in the x-axis
scale, especially for panel A (days unit compared to other panels in hours). Only the long-
term test is shown for the new silicon-coated cylinder (panel A), and the results of short-term
tests (1 and 2) can be found in supporting information Figure S2. .......................................... 85
Figure 8. Evolution of gaseous Hg concentration in the old silicon-coated cylinder A after

loading with uncontaminated argon. This test was conducted after the five tests with
contaminated argon (Figure 6B). The initial gaseous Hg concentration of 1 ng.L-1 is consistent
with the system blank value. ................................................................................................ 86
Figure 9. Hg recovery during (A) emptying and (B) heating the old silicon-coated cylinder B.
Panel A shows the variations in gaseous Hg concentration during the emptying of the silicon-
coated cylinder B, shown as a gaseous Hg concentration as a function of the pressure left in
the cylinder. Panel B illustrates the temperature dependence of Hg desorption rates. ............ 88
Figure 10. Example of a Hg concentration result given by the volumetric gradient approach 92
Figure 11. Stability of gaseous Hg concentration during tests performed on the new silicon-
coated cylinder ......................................................................................................................... 92
9
Figure 12. Thermal desorption results for (A) the PTFE-coated cylinder and (B) the uncoated
cylinder ..................................................................................................................................... 93
Figure 13. Variation de la concentration d'Hg0 gazeux au cours du temps pendant l'expérience
de stabilité menée sur le sac Tedlar .......................................................................................... 97
Figure 14. Temps de "percée du mercure" obtenu en fonction du volume de gaz produit lors
de trois tests de puits .............................................................................................................. 102
Figure 15. Temps de "percée du mercure" obtenu en fonction de la quantité totale produite
estimée lors de trois tests de puits .......................................................................................... 103
Figure 16. Concentrations en mercure analysées dans chaque piège en fonction du volume de
gaz .......................................................................................................................................... 108
Figure 17. Schéma du système de barbotage utilisé pour la détermination de l'arsenic totale
dans les échantillons de gaz naturel ....................................................................................... 112
Figure 18. Schéma du système de désorption thermique pour la détermination de l’arsenic

absorbé dans les cylindres. ..................................................................................................... 114
Figure 19. Chromatogramme du standard de Ph3As à 25 ppb dans le toluène ...................... 116
Figure 20. Chromatogramme de l’arsenic du condensat de gaz de (A.) l’échantillon A – 2 et

de (B.) l’échantillon A – 5. Analyse des échantillons purs. ................................................... 117
Figure 21. Système de barbotage mise en place sur le site de Mendixka pour la détermination
de la concentration en As total dans le biogaz ....................................................................... 121
Figure 22. Vu intérieur du réacteur (A.) utilisé lors de l’échantillonnage du bio gaz de
Mendixka (B.). ....................................................................................................................... 124
Figure 23. Analyse du soufre (A.), de l’arsenic (B.) et du mercure (C.) par LA-ICP-MS de la
surface de la plaque en acier inoxidable non-revêtue neuve (blanc) ainsi que celle utilisée lors
de l’échantillonnage du biogaz du site de Mendixka ............................................................. 129
surface de la plaque ayant le revêtement Silconert® neuve (blanc) ainsi que celle utilisée lors
de l’échantillonnage du biogaz du site de Mendixka ............................................................. 130
surface de la plaque ayant le revêtement Dursan® neuve (blanc) ainsi que celle utilisée lors de
l’échantillonnage du biogaz du site de Mendixka .................................................................. 131
10
Figure 26. Photographie de la surface de la plaque ayant le revêtement Silconert® (A.) et ainsi
que celle ayant le revêtement Dursan® obtenue par microscope optique (Grossisement : 200
µm) ......................................................................................................................................... 132
Figure 27. Analyse par SEM-EDX de l’arsenic (B.), du soufre (C.) et du mercure (D.)
adsorbés sur la surface du site n°1 (A.) de la plaque Silconert® ........................................... 134
adsorbés sur la surface du site n°1 (A.) de la plaque Dursan® .............................................. 135
11
Table des tableaux
Table 1. Commercial Instruments Dedicated to Hg Analysis in Gases ................................... 28
Table 2. Summary of Sample Preparation Methods Used for Total Hg Concentration Analysis
.................................................................................................................................................. 34
Table 3. Commercial Instruments Dedicated to Hg Analysis in Liquid and Solid Samples and
to Hg Speciation in Liquid Samples ......................................................................................... 43
Table 4. Summary of Sample Preparation Methods Used fo Arsenic Concentration Analysis

with their Detection Limits (MDLs) ........................................................................................ 68
Tableau 5. Concentrations en mercure mesurées à l’entrée et à la sortie d’une MRU .......... 107
Tableau 6. Concentrations en Hg obtenus après analyses de blancs des pièges en changeant

leurs positions lors du nettoyage ............................................................................................ 109
Tableau 7. Informations sur les échantillons de gaz naturel analysés .................................... 111
Tableau 8. Paramètres d’analyse par HR-ICP-MS des solutions de barbotage obtenues après
vidage complet des échantillons de gaz naturel contenus dans les cylindres ainsi que lors de la
désorption thermique .............................................................................................................. 112
Tableau 9. Paramètres d’analyse par GC-ICP-MS des condensats de gaz selon Bouyssiere et
al., 2001 .................................................................................................................................. 113
Tableau 10. Limite de détection expérimentale obtenue pour chaque échantillon de gaz naturel
obtenu en fonction du volume de gaz barboté ........................................................................ 115
Tableau 11. Concentration en As désorbé thermiquement des cylindres .............................. 118
Tableau 12. Paramètres d’analyses par ICP-MS des solutions de barbotages obtenues après
barbotage du biogaz de Mendixka ......................................................................................... 122
Tableau 13. Paramètres d’analyse par LA-ICP-MS des plaques utilisés lors de
l’échantillonnage du biogaz sur le site de Mendixka ............................................................. 125
Tableau 14. . Paramètres d’analyse par SEM-EDX des plaques utilisées lors de
l’échantillonnage du biogaz sur le site de Mendixka ............................................................. 126
Tableau 15.Concentrations en Hg total déterminées dans le biogaz du site de Mendixka .... 127
12
Table des abréviations
AAS Spectromètre d’absorption atomique

AFS Spectromètre de fluorescence atomique
AgNO3 Nitrate d’argent
As Arsenic
AsH3 Arsine
AsO43- Arséniate
Au Or
BIC Bordeaux Imaging Center
BrCl Monochorure de brome
BuMgCl Chlorure de butylmagnésium
(C2H5)3As Triéthylarsine
(C6H5)3As Triphénylarsine
(CH3)As Monométhylarsine
(CH3)2As Diméthylarsine
(CH3)3As Triméthylarsine
CH3Hg+ Methymercure
Cl- Chlorure
Cryo Cryogénie
CSTJF Centre Scientifique et Technique Jean Féger
CV Vapeur froide
DDP Polarographie différentielle par implusions
DLs Limites de détection
DMHg Diméthylmercure
EDX Spectroscopie de rayons X énergétiques
ETAAS Spectrométrie d’absorption atomique électrothermique
EtHg Ethylmercure
ETV Vaporisation électrothermique
FAAS Spectrométrie d’absorption atomique à flamme
FD Facteur de dilution
FeO Oxyde ferreux
Fe2O3 Oxyde ferrique
13
FeS2 Pyrite
FIA Analyse par injection en flux continu
Ge Germanium
GFAAS Spectrométrie d’absorption atomique à four graphite
GPC Chromatographie par perméation de gel
HCl Acide chlorhydrique
HClO4 Acide perchlorique
H2O2 Peroxyde d’hydrogène
H2S Sulfure d’hydrogène
H2SO4 Acide sulfurique
HG Génération d’hydrures
Hg Mercure
Hg0 Mercure élémentaire
HgCl2 Dichlorure de mercure
Hg2Cl2 Calomel
HgS Cinnabre
HNO3 Acide nitrique
HPLC Chromatographie en phase liquide à haute performance
HR Haute résolution
HS Espace de tête
HTL Hydroliquéfaction
I2 Diiode
ICP-MS Spectrométrie de masse à plasma à couplage inductif
ID Dilution isotopique
KMnO4 Permanganate de potassium
LA Ablation laser
LA2O3 Oxide de lanthane
LNG Gaz naturel liquéfié
MDLs Limites de détection de méthodes
MeHg Méthylmercure
MgNO3 Nitrate de magnésium
MIP-AES Spectrométrie d'émission atomique à plasma induit par micro-ondes
MMHg Monométhylmercure
14
NaBH4 Tétrahydruroborate de sodium
NaCl Chlorure de sodium
NaOCl Hypochlorite de sodium
NaOH Hydroxyde de sodium
NG Gaz naturel
(NH4)2C2O4 Oxalate d’ammonium
NH4I Iodure d’ammonium
Ni(NO3) Nitrate de nickel
OES Spectrométrie d’émission optique
Pb Plomb
Ph3As Triphénylarsine
Pt Platine
PTFE Polytétrafluoroéthylène
PVG Génération de vapeur photochimique
PVT Pression Volume Température
S0 Sulfate
SEC Chromatographie d’exclusion stérique
SEM Microscopie électronique à balayage
SnCl2 Chlorure d’étain
THT Tétrahydrothiophène
TMA Triméthylarsine
TMAH Hydroxyde de tétraméthylammonium
UV Photooxydation
15
INTRODUCTION GENERALE
Le mercure (Hg) et l’arsenic (As) sont des éléments naturellement présents dans les
produits pétroliers. Le Hg est présent à des concentrations allant de 0,01 à 10 µg.Nm-3 dans
toutes les fractions pétrolières (Wilhelm and Bloom, 2000). L’As, quant à lui, est présent à des
concentrations allant de 10 à 63000 µg.Nm-3 dans le gaz naturel (Irgolic et al., 1991). Dans les
fluides pétroliers, la présence du mercure pourrait résulter de son existence initiale, avant son
enfouissement, dans la matière organique qui deviendra par la suite du pétrole mais également,
de sa mobilisation à partir de roches environnantes, vers le pétrole formé (Filby, 1994;
Lambertsson et al., 2018). Tandis que pour l’arsenic, sa présence pourrait résulter de la capacité
des organismes marins à bioaccumuler cet élément (Liu et al., 2014), leur décomposition ayant
conduit à la génèse de pétrole contenant de l’As (Irgolic et al., 1991).
La présence de mercure et d’arsenic dans les matrices pétrolières est à l’origine de divers
problèmes pour les compagnies pétrolières. Ces éléments, à des concentrations importantes,
représentent un risque pour la santé du personnel en contact avec ces matrices, en raison de leur
toxicité pouvant provoquer, à long terme, différents cancers (Boffetta et al., 1993; Yeh et al.,
1968). Ceux-ci peuvent aussi poser des problèmes environnementaux, notamment par le rejet
du Hg et de l’As dans l’atmosphère, lors de la combustion de gaz naturel par exemple (Krupp
et al., 2007). D’un point de vue industriel, le Hg a la faculté de catalyser la corrosion de
composants en aluminium : ce qui a des conséquences économiques (remplacements des tubes
et autres pièces plus fréquemment qu’à l’ordinaire) mais surtout potentiellement dramatiques
du point de vue de la sécurité, dans les usines de traitements des fluides pétroliers (Leeper,
1980). La présence d’As dans les fluides pétroliers pose deux problèmes majeurs d’un point de
vue industriel. En effet, la précipitation des composés contenant de l'As conduit à l'obstruction
des tuyaux et des vannes dans les « pipes » (Delgado-morales et al., 1994). De plus, la présence
d’As dans les fluides pétroliers est à l’origine de l'empoisonnement des catalyseurs pendant le
vapocraquage et le traitement du gaz naturel : ce qui représente des coûts financiers élevés
(Delgado-morales et al., 1994; Irgolic et al., 1991).
Des systèmes d’élimination du mercure et de l’arsenic efficaces peuvent être mis en
place par les compagnies pétrolières afin de traiter les fluides pétroliers. Le développement et
le dimensionnement de ces unités de traitements sont basés sur la concentration totale de ces
16
deux éléments et, si possible, sur la spéciation de ceux-ci dans les fluides pétroliers à traiter. De
plus, le gaz d’export est très souvent sujet à spécification en ce qui concerne ce type de
composés. Ainsi, un dépassement de spécification peut avoir comme conséquence l’arrêt de
l’export. Il est par conséquent indispensable d’évaluer avec fiabilité les concentrations en
mercure et en arsenic.
La détermination et la spéciation de ces deux éléments dans les effluents gaziers et
pétroliers ont une approche très différente de celle appliquée en salle blanche ou en laboratoire.
Le questionnement sur la fiabilité et la représentativité de l'échantillonnage réalisé sur site est
prédominant par rapport à l'analyse et la spéciation du mercure et de l'arsenic. L'échantillonnage
et l'analyse des éléments traces dans le gaz naturel et le pétrole sont pleins d'artefacts potentiels.
Par exemple, la concentration de Hg en phase gazeuse peut parfois être surestimée de plus d'un
ordre de grandeur, cela étant dû, notamment, à la présence d’H2S (Rumayor et al., 2017). De
plus, des pertes significatives de mercure et d’arsenic par adsorption sur les parois des « pipes »
ou par réaction en présence d’oxygène peuvent être observées et donc fausser les résultats. Pour
finir, il existe un risque de faux positifs dus à une contamination ou de faux négatifs dus à l'effet
de percée retardée de ces composés lors d'un test de puits.
L’objectif de cette thèse est de traquer des artéfacts potentiels et de proposer des
améliorations durant l’échantillonnage du mercure et de l’arsenic afin d’évaluer avec fiabilité
et de façon représentative leurs concentrations dans les fluides pétroliers.
Le chapitre 1 de cette thèse est un état de l’art de méthodes d’échantillonnage, de

préparations d’échantillons et d’analyses afin de déterminer la concentration et la spéciation du
mercure et de l’arsenic dans les flux pétroliers. Les parties 1 et 2 du chapitre 2 présentent ensuite
les travaux expérimentaux menés sur la fiabilité de l’échantillonnage à l’aide notamment de
cylindres et de sacs Tedlars et de l’analyse du mercure dans le gaz naturel. La partie 3 du
chapitre 2 présente les recommandations sur l’échantillonnage et l’analyse du mercure lors d’un
essai de puits. Le chapitre 3 présente ensuite les travaux expérimentaux menés sur la fiabilité
de l’échantillonnage par cylindres et de l’analyse de l’arsenic dans le gaz naturel. Le chapitre 4
présente les travaux expérimentaux menés sur la compréhension du phénomène d’adsorption
du mercure et de l’arsenic à la surface des cylindres d’échantillonnage. Pour finir, ce manuscrit
est clôturé par une conclusion générale et des perspectives sur le travail de thèse ainsi que les
résultats obtenus.
17
CHAPITRE I : ETAT DE L’ART
18
Partie 1 : Total and speciation analyses of mercury
in the crude oil industry: a review
Energy Fuels 2020, 34, 11, 13307–13320 - https://doi.org/10.1021/acs.energyfuels.0c02730
Received: August 13, 2020; Revised: October 7, 2020; Publication Date: October 19, 2020
Maxime Enrico1,2 II, Aurore Mere1,2,3 II, Honggang Zhou3, Hervé Carrier2,4, Emmanuel
Tessier1,2, Brice Bouyssiere1,2*
1
Institut des Sciences Analytiques et de Physico-Chimie Pour L'environnement et les Matériaux, CNRS/ UNIV
PAU & PAYS ADOUR/ E2S UPPA, UMR5254, 64000, Pau, France
2
Joint Laboratory C2MC: Complex Matrices Molecular Characterization, Total Research & Technology,
Gonfreville, BP 27, F-76700 Harfleur, France
3
TOTAL E&P, CSTJF, Av. Larribau, 64018 Pau, France
4
CNRS/TOTAL/ UNIV PAU & PAYS ADOUR / E2S UPPA, LFCR-IPRA UMR 5150, 64000, Pau, France
*Corresponding Author: Brice Bouyssiere – [email protected]
II - M.E. and A.M. contributed equally to this work.
Abstract
Mercury is one of most dangerous contaminants of natural gas and petroleum and
petroleum fractions present in the crude oil industry due to its corrosive properties. The
determination of Hg levels in petroleum products is a prerequisite for risk evaluation and
ensuring an efficient Hg removal. This Review focuses on total and speciation analyses of Hg
in the petroleum industry. We describe the constraints related to sampling and analysis for both
natural gas and liquid petroleum products and the different methodologies developed over the
past decades. Despite the low detection limits reached with various methodologies, the
reliability of measurements might strongly depend on sampling and storage practices. The
preservation of all Hg species from sampling to analysis might require some further evaluation.
We evaluated the speciation methodologies based on reported performances. We however
conclude that the lack of reference material for Hg speciation analysis in petroleum hampers a
robust evaluation of the methods. Future developments might allow a better characterization of
the different Hg species, helpful to understand the partitioning of Hg between the different
petroleum phases.
19
1. Introduction
Mercury is a naturally occurring and problematic component of petroleum products.

This heavy metal is present in all petroleum fractions at levels varying by several orders of
magnitude ranging from 0.01 ppb to 10 ppm.1 Its presence in hydrocarbon fluids might result
from Hg present in organic matter prior to burial, mantle degassing of gaseous Hg0 and
scavenging by petroleum or mobilization of mineral Hg from surrounding rocks.9,77 Hg is a
toxic and volatile heavy metal. In its metallic form (Hg0), it is generally found as a gas, although
it might also occur as liquid drops when the gas phase is saturated. Upon oxidation, Hg0 might
be converted to salts (such as HgCl2) or complexes (thiolates, amines, etc.). The most toxic
forms of Hg are methylated Hg neurotoxins that are soluble (monomethylmercury) or volatile
(dimethylmercury, DMHg).
In addition to being problematic for both health and the environment, Hg catalyzes the
corrosion of aluminum components, with potentially dramatic consequences in petroleum
processing plants.2 Failures of a liquefied natural gas due to Hg-induced corrosion were
responsible for numerous deaths. The implementation of mercury removal units (MRUs)
efficiently prevents such failure, yet some maintenance is necessary. In particular, each MRU
can remove a given amount of Hg that depends on the amount of hydrocarbons treated and their
Hg concentration. Reliable evaluation of Hg concentrations in hydrocarbon streams is therefore
necessary to avoid saturation of the MRU, which could lead to inefficient Hg removal.
Hg partitioning between the different phases of petroleum can be estimated by
thermodynamics. This, however, generally supposes that Hg is present in its volatile form of
Hg0.1,3 The partitioning of Hg between these phases is governed by Henry’s law. Although this
approach sometimes gives satisfactory results, the gas-phase Hg concentration is sometimes
overestimated by more than an order of magnitude.4 It has been noted that this was particularly
the case when H2S was present. The presence of H2S and other sulfur compounds is known to
affect Hg speciation.5 The initial assumption of Hg0 being the only species is therefore
erroneous when sulfur is present at high concentrations, and the presence of soluble Hg species
concentrates Hg in the relatively heavy hydrocarbon fractions. A better understanding of Hg
chemistry in petroleum is therefore required to better anticipate Hg partitioning and implement
appropriate structures and maintenance procedures.
The assessment of Hg levels in hydrocarbon streams is complicated by several
constraints. The exposure of fluids to pipeline inner surfaces allows for some reactions that
20
might affect Hg chemistry. These different issues presented in the first section of this work need
to be taken into account when sampling and analyzing Hg in hydrocarbons streams. Different
methods have been developed over the past few decades to reliably determine the Hg
concentration in both gas and liquid hydrocarbon samples. We summarize these methods in
sections 2 and 3 of this Review.
In addition to Hg concentration, the efficiency of the MRUs depends in part on the Hg
species present. The different Hg species that can be found in liquid hydrocarbons (including
Hg0, ionic Hg, organic Hg and particulate Hg) have different chemical properties.6−8 In natural
gas, it is usual to consider Hg0 to be the only significant species due to its high volatility
compared to those of the other, more soluble species. The study of Hg speciation in liquid
hydrocarbons is of interest because of the need for sufficient treatment procedures. Although
Hg0 might be the dominant form of Hg in deep reservoirs,9 the presence of oxidized Hg,
particulate Hg and alkylated Hg species was reported.10,11 For Hg concentration, various
techniques are used to evaluate the relative abundance of these species. Section 4 is dedicated
to the description of these techniques.
Among the numerous method developments for both Hg concentration and speciation
determination in petroleum streams, only few gave rise to standard methods. The present
standard methods are also relatively old and were only adapted through minor revisions and did
not consider the major recent developments in the field.
2. General Constraints Related to the Sampling and Quantitative

Analysis of Mercury
2.1. Natural Gas
The objective of Hg analysis in petroleum products is to assess any risk to downstream

operations. Therefore, Hg analysis must be representative for such an assessment. Several
constraints need to be considered for reliably estimate Hg levels. We first focus on general
issues that apply to any sampling methodology.
21
2.1.1. Hg Adsorption
2.1.1.1. Current Knowledge
The Hg adsorption process is not yet completely resolved. However, significant losses
of Hg are observed for H2S-containing streams, which probably indicates a reaction involving
both H2S and Hg. Two studies reported the following reactions:12,13
𝐻2 𝑆 + 𝐹𝑒2 𝑂3 → 2𝐹𝑒𝑂 + 𝑆 0 + 𝐻2 𝑂 (1a)
𝐻𝑔0 + 𝑆 0 → 𝐻𝑔𝑆 (1b)
Cinnabar (HgS) is a solid mineral and can therefore accumulate in pipes. These two studies did
not show this reaction but cited an earlier study2 to support this concept.
Larsson et al.13 observed that a gas containing both H2S and Hg had no loss of Hg when
flowing through PTFE tubes, but significant losses were observed in stainless steel tubes. This
suggests that the Hg and H2S gases do not react directly, which would support the mechanism
described by eqs 1a and 1b.
2.1.1.2. Thermodynamic Considerations
The standard free enthalpy of formation of the iron oxide Fe2O3 is −825.5 kJ.mol−1,
while that of FeO is −272.04 kJ.mol−1. Balancing eq 1a (taking into account H2O, S0 and H2S)
gives a net enthalpy of +18 kJ.mol−1. A positive net standard reaction enthalpy indicates that
the reaction should not take place at room temperature. Adsorption of Hg on stainless steel
surfaces in the presence of H2S was observed at 25°C, so the proposed reaction seems unlikely.
However, from our point of view, another reaction between iron oxide and H2S is more
favorable:
4 𝐻2 𝑆 + 𝐹𝑒2 𝑂3 → 2 𝐹𝑒𝑆2 + 𝐻2 + 3 𝐻2 𝑂 (2)
The net enthalpy for this reaction is −155 kJ.mol−1, which is more likely than reaction 1a. Pyrite
(FeS2) formation could explain the adsorption of Hg. The surface of pyrite contains pyritic
groups (S-S) and Fe (hydr)oxides. Sorption of Hg(II) on the pyrite surface occurred at both the
oxide and pyritic sites,14,15 with Hg bound to an additional counterion (as =S-Hg-Cl, =S-Hg-
OH, =O-Hg-Cl, and =O-Hg-OH). In natural gas streams, however, Hg is present in its elemental
22
form (Hg0) and therefore requires an oxidation step for such surface complexation. In these
studies, precipitation of HgS was excluded, but these studies were not conducted on
hydrocarbon samples.
2.1.2. Implication of Adsorption Problems for New Pipes
The concentration of Hg in natural gas can vary over time, making it difficult to
determine precisely. In particular, the analysis of natural gas from a new well may not be
informative of the concentration of Hg in the reservoir. Indeed, Hg adsorption may occur on
the inner surfaces of the reservoirs, and the gas is therefore depleted of Hg. Hg analysis could
therefore underestimate the Hg content of a gas stream by several orders of magnitude.
In addition, gases that first reach the Earth’s surface may be depleted of Hg relative to
that in the reservoir because sorption can occur on the well walls. It is therefore important to
regularly monitor the Hg concentration in natural gas until a stable value is reached.
In addition, Ryzhov et al.16 observed an oscillation in Hg concentration during sampling
campaigns. A 25-hour cycle was attributed to the lunar tidal cycle, while higher frequency
oscillations were observed but not included. These oscillations sometimes do not significantly
affect the Hg concentration range. In some cases, however, the fluctuations could lead to an
error of more than 20% if a single sample is analyzed. Diurnal and annual cycles can also be
anticipated, as temperature affects adsorption/desorption of Hg in pipes and thus increase the
Hg concentration during the summer.17
2.1.3. Implication of Adsorption Problems for the Sampling System
Natural gas sampling requires a bypass of the main gas stream. This sampling system is
not constantly filled with natural gas, so the initial gas stream is not in equilibrium with respect
to Hg adsorption. An equilibration time is therefore necessary before sampling with the chosen
methodology. This time depends on the total internal surface area of the bypass line. A reduced
line is therefore preferable.
Adsorption also depends on temperature. An experimental study on the efficiency of Hg
transport in stainless steel tubes showed that heating from 70 to 100°C hinders Hg adsorption.13
Therefore, a heated bypass line would significantly reduce sampling times.
23
2.2. Liquid Petroleum Products
Sampling procedures must be appropriate to preserve the stability of Hg in liquid

hydrocarbons. The different species of Hg found in liquid hydrocarbons (including Hg0, ionic
Hg, organic Hg, and particulate Hg) have different chemical properties and behaviors.6−8 To
determine the mercury species composition of a hydrocarbon, it is very important that Hg is
preserved in the sample during sampling, storage, and sample preparation prior to analysis.
Several studies have shown that inappropriate sampling and treatment could lead to an
underestimation of the Hg concentration in the condensate.3,20 The different mechanisms to be
avoided are discussed below.
2.2.1. Volatile Hg Loss
Two species of Hg are particularly volatile: Hg0 and DMHg.10,18 DMHg is generally
very low in concentration and often below the limit of detection.19 Therefore, the main issue is
the loss of Hg0. In the study of Hg speciation in condensates in Egypt, Ezzeldin et al. 3 found
that following a pressure drop between the first and second separators, Hg0 was transferred to
the gas phase. A similar process may occur during sampling or sample storage, as condensate
samples are taken at atmospheric pressure. In the same study, the authors compared their
speciation results to the total Hg measurements. For condensate samples containing mainly Hg0
(untreated condensate), the summed concentration of the detected Hg species was only half of
the total Hg concentration measured independently. This unsatisfactory mass balance was
attributed to losses of Hg0.
Wilhelm et al.20 observed that the Hg concentration in crude oil samples decreased with
the number of bottle openings. During sample storage, dissolved Hg0 may also volatilize into
the free space of a container. These results suggest that sample containers should be filled
without free space and that the opening of containers should be avoided to prevent evaporation
and/or transformation of Hg species over time. Snell et al.21 also observed a significant loss of
Hg0 in mixed hydrocarbon solutions over time.
Different sampling procedures can be designed to avoid volatile Hg losses. Pressurized
cylinders can be used to preserve the initial sample pressure conditions.5 The use of a container
with a septum cap could be practical, as a syringe could be used to take an aliquot of the sampled
gas without opening the container. The ASTM protocol for “Sampling, Storage, and Handling
24
of Hydrocarbons for Mercury Analysis” 22 recommends the use of glass vials with PTFE-lined
septum caps. In this way, it is possible to limit the volatilization of Hg0 by filling the sample
vial with no free space.
2.2.2. Ionic Hg Adsorption on Container Walls
Hg is known to bind to the walls of storage and/or sampling containers/containers.

Despite its relative inertness compared to other materials, glass tends to be negatively charged
and can attract cationic forms of Hg that will absorb onto the glass surface. Acidification of the
glass surface can repel Hg species and stabilize them within the sample, thus preventing further
Hg loss.3
The adsorption of Hg depends on the nature of the material but also on the chemical
species of Hg contained in the matrix. Bloom et al.23 showed that Hg0 and DMHg diffuse
through the walls of plastic containers.
The composition of the sample matrix may play a role in this adsorption effect. Bloom10
tested the stability over time of Hg0, inorganic Hg, and dimethyl- and methylmercury chloride
species contained in doped paraffin oil and in composite oil in different containers (borosilicate
glass containers with Teflon-lined caps, Teflon FEP bottles, polyethylene bottles, stainless steel
bottles (No. 316) and glass bottles lined with aluminum). A significant loss of inorganic Hg
was observed in doped paraffin oil in a glass container, whereas no loss of inorganic Hg was
observed in composite oil contained in the same glass container. This shows that the
composition of the matrix, especially its polarity, can influence this adsorption phenomenon.
This adsorption induces an underestimation of the measured Hg concentration. Various
studies have shown that this problem can be solved by using appropriate sample containers.
The ASTM D7482-17 protocol suggests the use of glass vessels with a PTFE-lined cap.22
Recently, Gaulier et al.5 showed that the use of a Swagelok cylinder lined internally with PTFE
could preserve anoxic conditions for Hg(II) preservation.
2.2.3. Precipitation of Hg-Containing Minerals
Particulate Hg is potentially naturally present in condensates. However, no studies have

actually identified this. Only one study showed the existence of metacinnabar (β-HgS, black
25
cinnabar) in crude oil residues.24 However, it has been shown that Hg can precipitate with S to
form HgS minerals when the sample is exposed to ambient air.5 This is particularly true when
H2S is present in the sample. While Hg does not react directly with H2S,19 contact with air can
promote the oxidation of H2S, leading to the generation of elemental sulfur. HgS minerals can
then precipitate from Hg and elemental sulfur. This is an additional reason why contact with
ambient air should be avoided prior to Hg analysis.
Another potential precipitation process may occur. Snell et al.21 observed that when the
two Hg species Hg0 and Hg(II) are present in a condensate sample, they can react and form a
metallic bond, giving rise to insoluble Hg(I) species:
𝐻𝑔0 + 𝐻𝑔𝐶𝑙2 → 𝐻𝑔2 𝐶𝑙2 (3)
The factors that might affect this precipitation reaction are poorly understood (pressure,
temperature, chemical characteristics of the sample, etc.), so it may be advisable to remove Hg0
from the condensate sample after sampling. Such a procedure exists and is part of a speciation
procedure.25
2.2.4. Sample Homogeneity
Any analysis performed on a sample aliquot assumes that the Hg is homogeneously

distributed in the sample. Heterogeneous samples can lead to a poor estimate of Hg
concentration and inconsistent repeat analyses.26 Depending on the Hg species present, the
different mechanisms described above (adsorption on container walls, volatilization of Hg0, as
well as sedimentation of particles) need to be taken into account.
Wilhelm et al.20 tested different homogenization procedures and found that ultrasonic
treatment at 30°C gave Hg recoveries >80%, while the heating and stirring procedure exhibited
only a 30% recovery. Ultrasonic treatment is therefore recommended prior to aliquoting for
total mercury and speciation analysis
26
3. Mercury Analysis in Natural Gas
3.1. Sampling method
3.1.1. Bubbling System
This sampling procedure involves bubbling natural gas into a trapping solution. This
solution is a mixture of potassium permanganate and sulfuric acid.27 Usually, three bubblers are
used in series to monitor the potential inefficiency of Hg trapping in the first bubbler. This
mixture oxidizes the Hg0 contained in the gas into soluble ionic Hg(II). This soluble ionic Hg(II)
is retained in the trapping solution and can then be analyzed. The sampling time must be long
enough to reach a mercury concentrate that is sufficient for quantification. The amount of Hg
recovered in solution (sum of three trapping solutions) divided by the volume of gas bubbled
into the solution gives the Hg concentration in natural gas. This method was recently adapted
to allow for sampling at pressures up to 100 bar with direct connection to distribution and
transportation networks.41 The results showed that more than 90% of Hg was trapped in the
solution of the first bubbler with insignificant Hg breakthrough, showing the efficiency and
reliability of this sampling system.
27
Table 1. Commercial Instruments Dedicated to Hg Analysis in Gases
Flow
DL Accuracy Weight
Brand Name Detector Analysis rate
(ng/m3) (%) (kg)
(L/min)
Brooks Rand Model III AFS lab, gold traps (requires 0.1
amalgamation)
Nippon Instrument WA-5A AAS Au traps, Tedlar bags, gas 1 pg 13 0.1 – 1

Corp. cylinders
WA-5F AFS Au traps, Tedlar bags, gas <0.1 pg 13 0.1 – 1

cylinders
EMP (3 different instruments) AAS in situ 100 – 10 0.3 – 1

50000
AM-4 AAS Air Monitoring 0.01 2 0.1 – 0.5
LUMEX RA-915+ AAS Direct gas analysis 2 7.5
RA-915M AAS Direct gas analysis 2 7
Light 915 AAS Direct gas analysis 100 3.3
RA-915M + RP-91NG AAS in situ (special NG) 10
AMNG AAS Monitoring for NG 1 - 1000
Jerome 431-X gold film in situ 3000 5 3.2 0.75
28
Arizona Jerome J405 gold film in situ 500 5 - 10 2.4 0.75
Instruments
Jerome 451 gold film in situ, monitoring 3000 8 24 0.75
Jerome J505 AFS in situ 50 10 - 15 3 1
PSA PSA 10.54X Sampling System + AFS in situ, monitoring (special

PSA 10.670 Process Analyzer NG)
10.525 Sir Galahad AFS lab, gold traps <0.1 pg
Ion Science MVI mercury vapor detector UV

in situ 100 10
absorption
Mercury VM-3000 AAS in situ, monitoring 100 7

Instruments
3000-IP in situ 100 5
UT-3000 AAS in situ, monitoring 0.1 9
UT-3000 + NG-sampling system

AAS in situ, for NG 0.1 50
3000
MMS AAS in situ, monitoring 100 35-175
MMS-NG AAS in situ, monitoring, for NG 1
TEKRAN 2600 NG AFS in situ, monitoring, for NG 1
2537A and 2537B AFS in situ, monitoring, ambient air 0.1
2500 AFS lab
29
Hg recovery is, however, significantly reduced in the presence of high H2S
concentrations. Brombach and Pichler28 observed that in the presence of H2S, the recovery of
Hg in potassium permanganate dropped by nearly 50% due to the formation of mercury sulfide
(HgS). This issue was overcome by bubbling the gas in a solution of sodium borohydride
upstream of the Hg trapping solutions to trap H2S and avoid any interference in Hg oxidation.
3.1.2. Hg-Selective Sorption
For this sampling method, a stream of natural gas is carried through a glass tube
containing iodine-impregnated silica gel. The Hg is trapped by chemisorption inside this tube.
The adsorbed Hg is then extracted from this tube by dissolving it in an ammonium iodide/iodine
(NH4I/I2) solution.29 This solution is then analyzed to determine the total Hg content of a natural
gas, taking into account the volume of gas sampled. Quantitative recovery of Hg (91−109%)
was achieved using this methodology in gas wells from China.30 However, the existence of a
liquid extraction stage hinders in situ analysis. In addition, the simplest procedures are always
preferred, as the addition of several processing steps can lead to higher background signals and
possibly more potential for contamination.
3.1.3. Gold Amalgamation
The system is based on trapping Hg from natural gas by directing the flow through a
tube of a solid sorbent containing gold.31 However, this process has a major advantage over the
use of iodine impregnated sorbents. Gold amalgamators can indeed be analyzed without any
digestion steps.3 After sampling, the amalgamator is heated, and Hg is released in an atomic
absorbance spectroscopy (AAS) or atomic fluorescence spectroscopy (AFS) detector. The most
widely used metal for trapping Hg is gold, but Hg can form amalgams with the most noble
metals. Different studies have investigated the use of metal alloys and their efficiency for Hg
trapping.13,32 Silver, gold and platinum−gold alloys allow for quantitative Hg retention.
By using the thermal release procedure, the analysis can be performed on the site if a
detector is available (Table 1). This allows a direct evaluation of the sampling efficiency, and
the sampling time and volume can be directly adjusted. A reduction in pressure is necessary,
however, as the amalgamators operate at ambient pressure.
30
Figure 1. Schematic presenting the different methodologies for Hg quantification in natural gas,
from sampling (top boxes) to sample processing (middle boxes) and detection (right boxes)
This method can be used to determine the total Hg content ranging from 0.01 to 100 μg.m3 in
natural gas samples.31 Adapted sampling flow rates are also required to avoid Hg breakthrough.
Frech et al.33,34 suggested that the maximum natural gas flow rate should be 2 L.min−1, and the
maximum sample volume should be 10 L when using a Pt−Au amalgamator. Despite these
results, Larsson et al.13 obtained Hg recoveries greater than 90% with amalgamators similar to
Frech et al.,34 with higher flow rates (up to 10 L.min−1) and higher sampling volumes (up to
100 L). This difference can be explained by the different compositions of natural gas that could
influence the transport efficiency of Hg on these traps, although this has not been proven.13
Despite this, a recent study has shown that the efficiency of Hg trapping on gold traps is much
lower or almost nil when the natural gas sample contains H2S.28
Heating the amalgamator during sampling generally increases Hg recovery. The aim is
to avoid condensation of liquids that could poison the amalgamator and thus reduce the trapping
efficiency. The temperature to which the amalgamator can be heated depends on the alloy used.
Frech et al.34 showed that Pt−Au amalgamators could be heated up to 200°C without any loss
in trapping efficiency, whereas gold amalgamators were ineffective at over 170°C.
31
3.1.4. Gas Samplers: Cylinders and Tedlar Bags
The use of pressurized cylinders is generally very convenient for sampling natural gas.
Because pressurized cylinders can handle high pressures, sampling is fast. Hg measurements
can be made once back in the laboratory by releasing some or all of the sampled gases through
a gold amalgamator, which is then analyzed by AAS or AFS.31 Although sampling is made
easier with such a system, its reliability is of great concern. Hg is known to adsorb to most
surface materials, including the inner walls of most cylinders.2,12,35 Measuring the concentration
of Hg would therefore underestimate the true concentration of natural gas. Solutions have been
developed to overcome this problem. The application of a silicon coating on the walls inside
the cylinders to make it inert to Hg adsorption. A silicon coating (SilcoNert 2000, Siltek, or
Sulfinert) was specifically designed for this purpose. Its effectiveness has been tested by the
manufacturer and has shown stable Hg concentrations for a period of at least 2 months.36 Other
tests showed that the inertness of the coating is compromised by repeated sampling of natural
gas samples.35
Alternatively, natural gas can also be sampled using inert plastic bags (Tedlar bags).37−39
Despite its ease of use, this sampling technique does not allow for natural gas to be sampled
under pressure, and the conservation of compounds in the gas samples over time is not assured
within these bags.40,41 Although the onsite sampling portion can be simplified by the use of
cylinders, the processing of samples in the laboratory may require more time. All disadvantages
associated with the use of amalgamators (discussed in the gold amalgamation section) remain,
as an amalgamation is required prior to analysis. Despite this constraint, the use of such
samplers allows for a reduction in sampling time in the field, which can be advantageous.
3.2. Hg Analysis
In natural gas, mercury exists almost exclusively as Hg0. 1,3

Therefore, the analyses
performed provide only the total amount of Hg in natural gas, i.e., Hg0.
The trapping solutions (KMNO4/H2SO4 solution) obtained after bubbling the gas sample
as well as the ammonium iodide/ iodine (NH4I/I2) solutions obtained after extraction of the Hg
contained in the iodine-impregnated silica gel can be analyzed by cold vapor generation (CV)-
AFS, CV-AAS, and inductively coupled plasma mass spectrometry (ICP-MS).27,29 For the first
32
two analytical techniques, the detector is based on AAS or AFS, respectively (Figure 1). During
analysis, a reducing agent (SnCl2, NaBH4) must be added to the solution to convert Hg2+ to Hg0
42
during CV.
For gold traps (direct sampling and/or after release of natural gas sampled in Tedlar
cylinders/bags), the usual method for Hg quantification is based on thermal desorption. The
sampled gold trap is heated to ∼500°C under an inert gas stream carrying Hg to another gold
trap. This other trap is heated again to ∼500°C, and the released gases are directed to the
detector. The detector can be either an AAS43 or an AFS.44 The AFS typically has a detection
limit (DL) 10 times lower than that of the AAS with the ability to detect up to a few picograms
of Hg (Table 1). For a robust measurement, the Hg trapped in the amalgamator must be above
the quantitation limit of the device because gold amalgamators may contain small amounts of
Hg even after a cleaning procedure.
33
Table 2. Summary of Sample Preparation Methods Used for Total Hg Concentration Analysis
Reference Sample type Sample processing Analysis MDL Sample size
By Liquid – Liquid Extraction
Schickling and Gas condensates Aqua regia extraction CV-AAS 0.01 ng/g NR
Broekaert34
Liang et al.,45 Gasoline, kerosene diesel, BrCl extraction CV-AFS 0.01 ng/g 5 mL
heating oil
Zettlitzer et Gas condensates and brines Aqua regia extraction CV-AAS 100 ng/mL NR
al.,30
Bloom,10 Crude oil, condensates, light BrCl extraction CV-AFS 0.17 ng/g 0.5 – 1.5 g
distillates, fuel oil, asphalt
Liang et al.,51 Crude oil and related products BrCl extraction CV-AFS 0.05-0.12 ng/g 1
Brandão et al.,54 Gasoline BrCl extraction CV-AAS 0.14 µg/L (BrCl 5 mL

extraction)
Conaway et Gasoline and diesel BrCl extraction CV-AFS NR NR

al.,55
Uddin et al.,56 Crude oil and petroleum Dilution in toluene, extraction with BrCl/HCl CV-AFS 0.38 ng/g 5g
products
Vicentino et Oil products (diesel, biodiesel Emulsion (Triton X-100 + HNO3) CV-AAS 0.6 ng/mL 20 mL
al.,46 mineral oil°
By Sample Digestion
Knauer and Crude, fuel and light oil HNO3/H2SO4 digestion, distillation CV-AAS 10 ng/g 5g
Milliman,50
34
Kelly et al.,52 Crude oil and diesel fuel Digestion by closed system combustion with ID-CV-ICP-MS 0.01 ng/g 0.25-.5 g
SRMs HNO3 + isotope dilution
Liang et al.,51 Coke (solid) HNO3/H2SO4 digestion at 95°C CV-AFS 0.15 ng/g 1g
Munoz et al.,49 Crude oil and diesel fuel Microwave digestion Stripping 194 ng/g 0.1 g
voltammetry
Stripping
chronopotentiometry 104 ng/g 0.1 g
Wilhelm et al.,26 Oil Autoclave HNO3 digestion (300°C, 130 bar) CV-AFS
Yun et al.,57 Crude oil and gasoline Microwave digestion with HNO3/H202 CV-AFS < 0.26 ng/g 0.1 – 0.2 g
Pontes et al.,47 Crude oil, oil sludge HNO3 digestion in closed pressurized system CV-AAS 1.3 ng/g 0.25 g
Aqua regia digestion in cold-ﬁnger system 8.6 ng/g 0.5 g
Alumina digestion in thermal decomposition 3.3 ng/g 0.0253 g

technique
Ezzeldin et al.,3 Gas condensates Autoclave HNO3 digestion (90°C) CV-AFS 0.01 to 0.08 ng/g ~ 1 mL
Gajdosechova et Petroleum hydrocarbon Microwaves digestion (90°C) with HNO3 CV-AFS 0.3 g
al.,11
ICP-MS 0.3 g
Pereira et al.,48 oily sludge Microwaves digestion with HN03, HCl and CV-AFS 0.07 ng/g 0.1 g
H2O2
HPLC-ICP-MS 0.01 ng/g
By Thermal Decomposition
Knauer and Crude, fuel and light oil Combustion CV-AAS 5 ng/g 20 g
Milliman,50
35
Osborne,58 C5 to C14 Electrothermal vaporization (graphite furnace, ETV-ICP-MS 3 ng/g
GF)
Tao et al.,59 Crude oil, naphta and Thermal decomposition AAS < 1.5 ng/g
condendates
Shafawi et al.,60 Gas condensates Combustion and amalgamation AFS 0.7 to 2.5 ng/mL 0.25 mL
Ceccarelli et Naphtas Thermal decomposition AFS 32 ng/g

al.,61
ETV (GF) AAS
Liang et al.,62 Crude oil Combustion and amalgamation AFS 0.2 ng/g 0.04 g
Liang et al.,51 Crude oil and related products Combustion and amalgamation AFS 0.05 – 0.12 ng/g 0.04 g
Bouyssiere et Gas condensates Combustion and amalgamation AAS 0.3 ng/mL 0.1 mL
al.,63
Wilhelm et al.,26 Oil Combustion and amalgamation AFS 0.5 ng/g
Torres et al.,64 Gasoline Dilution by ethanol, CV generation and transfer AAS 0.14 ng/mL 1.5 g
in an oven graphite furnace
Dilution by ethanol, CV generation,

amalgamation and transfer in an oven graphite 0.08 ng/mL 1.5 g
furnace
Pontes et al.,47 Crude oil, oil sludge pyrolysis AAS 3.3 ng/g <0.0253 g
Gaulier et al.,5 Crude oil, condensates Adsorption of 20 µL in a pellet, combustion AAS 0.02 mL
Lambertsson et Crude oil Thermal decomposition AAS 0.03 ng/g

al.,9
Other Methods
Olsen et al.,65 Condensates and naphtas Dilution in xylene ICP-MS 0.13 ng/g 2g
36
Kumar and Naphtas Emulsion with Triton X-100 aqueous solution ICP-MS 0.12 ng/mL 2 mL
Gangadharan,66
Bouyssiere et Gas condensate Dilution in 2-propanol ICP-MS (HR) < 9 ng/mL

al.,67
Brandão et al.,54 Gasoline Microemulsion with 2-propanol + HNO3 CV-AAS 0.10 µg/L 20 mL
Bouyssiere et Gas condensate Microflow injection ICP-MS 0.5 ng/mL 0.0025 mL

al.,63
Vorapalawut et Crude oil and asphaltene silica gel plate impregnation LA-ICP-MS 11 ng/g
al.,68
da Silva et al.,69 Bio ethanol Addition of formic or acetic acid, dilution PVG-AAS 0.05 to 0.09 ng/g > 50 mL
de Jesus et al.,70 Naphta and condensates Microemulsion (sample + propan-1-ol + water PVG-AAS 0.6 ng/mL 1 mL
25:24:1)
Lambertsson et Crude oil Isotope spike, heating at 200°C ID-HS-GC-ICP-MS 0.03 ng/g
al.,9
37
3.3. Direct In Situ Analysis
A few companies offer instruments for the determination of Hg gas in the field (Table
1). Among them, some are dedicated to natural gas. Ryzhov et al.16 used a system based on
AAS with Zeeman correction for direct monitoring of Hg concentration in natural gas. The
analysis is fast and can be used to monitor the evolution of Hg concentration over time. This
method is very convenient because sampling and analysis are performed simultaneously. Such
a tool would be very effective for natural gas with high Hg concentrations (a few μg.m-3),
although it may also be useless if there are only a few ng.m−3 of Hg in the natural gas. Ryzhov
et al.16 reported DLs of approximately 10 ng.m−3. This instrument (LUMEX RA915M coupled
to RP-91NG) can be easily transported and is a good alternative to other quantification methods
because no sampling steps are required.
Other automated instruments based on amalgamation and AAS or AFS also exist and
are suitable for natural gas (Table 1). These automated instruments are, however, more
expensive, require some maintenance and cannot be easily moved. Such an instrument could
be recommended for sites where variable Hg concentrations have been detected to monitor any
changes over time and to avoid any downstream problems.
Direct in situ analysis via direct AAS analysis can monitor the concentration of Hg in
the gas over time, even during the equilibration time. This can provide additional information
on the adsorption process of Hg, especially in natural gases containing H2S.12,13
4. Mercury Analysis in Hydrocarbon Liquids
4.1. Total Hg Analysis
4.1.1. Preparation
4.1.1.1. Liquid-liquid Extraction
Mercury can be extracted from liquid hydrocarbons by liquid−liquid extraction. The

oxidant solution of BrCl seems to be the most used (Table 2). This technique is simple and
rapid because the oxidation and recovery of mercury in the BrCl solution is carried out at the
same time.45 BrCl solution converts the mercury present in the samples to its oxidized ionic
form (Hg2+), so a reducer must be used, prior to any analysis, to convert the recovered Hg2+ in
38
Hg0. It has also been shown that the use of an HCl solution does not allow an optimal recovery
of Hg from a NaBH4 solution45 and more recently from a Triton X-100 + HNO3 solution.46
4.1.1.2. Sample Digestion
Unlike the liquid−liquid extraction mentioned above, here, the organic matrix remains
in the sample. Digestion by heating with the addition of HNO3 is the most widespread (Table
2). Pontes et al.47 showed that thermal digestion coupled with this solution made it possible to
obtain a very satisfactory recovery rate as well as a method detection limit (MDL) of 1.3 ng.g−1
and 8.6 ng.g−1, respectively, which were much lower than those obtained with digestion by a
cold system and the addition of aqua regia. This technique is simple, fast and inexpensive.48
Munoz et al.49 showed that when using a closed-tank microwave oven, they did not observe any
loss of Hg, including volatile Hg. Despite this, several studies have shown that wet digestion
by heating is highly vulnerable to contamination and Hg loss.11,50,51 Isotopic dilution, as
mentioned by Kelly et al.,52 corrects for Hg losses by assuming they are the same for the spike
and the hydrocarbon liquid sample. This digestion therefore seems to be the most advantageous
for Hg analysis, and Kelly et al. also noted that isotopic dilution allows for correction of
potential losses (with the addition of a Hg spike) during digestion.52 Digestion combined with
isotope dilution therefore appears to be the most advantageous method for Hg analysis.
4.1.1.3. Thermal Decomposition
The most common method for sample preparation is the use of thermal decomposition
of samples, followed by gold amalgamation. In the first step, Hg is thermally released from the
sample, and the Hg compounds are decomposed so that only Hg0 remains. Produced gases are
transferred to a gold amalgamator by a carrier gas that is then analyzed.26,60,63 Other systems
use the same principle without any amalgamation step. The sample is thermally decomposed in
a graphite furnace (GF), and the gases are analyzed directly in the graphite furnace coupled
with AAS.9,47,64 However, without the fusion stage, the DLs are usually relatively high. In fact,
Pontes et al.47 wanted to compare their method, which did not have an amalgamation stage,
with that of Liang et al.62 who did have an amalgamation stage. Pontes et al.,47 obtained a DL
of 3.3 ng.g−1 with GF-AAS, which was higher than the MDL of 0.2 ng.g−1 reported by Liang et
39
al.,62 when using an amalgamation step (Table 2). Comparisons between thermal decomposition
and liquid− liquid extraction or sample digestion procedures generally show a better recovery
with thermal decomposition than without this step. Thermal decomposition is therefore
preferred for Hg analysis.26,51
4.1.1.4. Other Methods
Several methods of sample preparation for the determination of total Hg may also be
used (Table 2). Sample injection into a carrier solvent was shown to be a practical method that
does not require sample treatment and fast analysis.63,65,67
Figure 2. Summary of the different methodologies for total Hg analysis in liquid petroleum, from
sample treatment (left boxes) to processing
This technique makes it possible to use very small quantities of samples, with a DL comparable
to those of thermal desorption-AAS or AFS.9,63
Several authors investigated the possibility of analyzing microemulsions by CV-AAS
or photochemical vapor generation (PVG)-AAS. This technique uses a very small sample and
reagent volume,70 and the DLs are comparable to those obtained by extraction with a solution
of BrCl: 0.10 ng.g−1 for a microemulsion and 0.14 ng.g−1 for extraction.54
40
Lambertsson et al.9 recently developed a new method for determining Hg in liquid oil
samples: ID-HS-GC-ICP-MS, which is based on the concentration of Hg in the head space (HS)
of a sample vial combined with isotope dilution (ID), GC separation, and ICP-MS detection.
An isotopic compound (198HgCl2) is added to the sample to accurately trace the Hg
concentration in the samples. The procedure includes one heating step that allows for the
complete conversion of all Hg species to Hg0. Hg0 then balances between the gas and the
dissolved phase. An aliquot of the headspace gas is injected into a gas chromatography (GC)-
ICP-MS system. Both 202Hg and 198Hg are monitored to allow for isotope addition correction.
This method provides a DL similar to that of thermal desorption coupled to AAS (0.3 ng.g−1).
Dilution of the condensate may sometimes be necessary to reduce its viscosity or to provide a
reasonably stable background for mercury signals. Benzene, toluene, and xylene are universal
solvents, but some condensates may be dissolved in hexane, isooctane, or isopropanol, which
may be useful in some cases.56,71
4.1.2. Analysis
The detection limit of a method depends on the detector used, the dilution factors, the
sample matrix after processing and the sample size (Table 2). Analyses by AAS, AFS, and ICP-
MS are the most widely used and produce very low MDLs of approximately 0.1 ng.g−1 (Table
1). Several studies have shown that these three analytical systems provide similar MDL,
accuracy and repeatability values.11,52,54,61,63
For liquid−liquid extraction and digestion methods coupled with CV-AAS and CV-AFS
(Figure 2), a reducer must be added to the solution prior to analysis to convert Hg2+ to Hg0. 42
A SnCl2 solution can be used,10,53 but Brandao et al.54 demonstrated that NaBH4 provides better
repeatability and sensitivity and a lower DL than those achieved when using SnCl2. In addition,
Gajdosechova et al.11 reported an underestimation of the Hg concentration when using SnCl2
because of the consumption of SnCl2 for reducing sulfates to sulfides. In addition, Da Silva et
al.69 noted that the reducing agents are often unstable and costly. We concluded that sample
preparations avoiding the use of reducing agents should be favored for Hg analysis. Cold vapor
generation can also be achieved through the photochemical reduction of Hg. The DLs reported
for PVG-AAS are relatively high (1.1 ng.g−1).69,70
41
Gajdosechova et al.11 found that CV-AFS should be used with alkaline samples, i.e., the
use of a 2% SnCl2 solution prepared in NaOH (5 M), to avoid underestimating the amount of
Hg.
Analysis of hydrocarbon samples by ICP-MS after dilution or emulsification was also
found to yield an acceptable MDL (0.2 ng.mL−1).
ICP-MS analysis has advantages over the AAS or AFS analyses because the samples
can be analyzed directly without pretreatment, and the analysis is faster than the other methods
mentioned above.63,65 However, the DL of ICP-MS (0.5 ng.mL−1) is slightly higher than those
of the other analytical procedures, which could interfere with the analysis of very low Hg
concentrations.63
Laser ablation coupled to ICP-MS enables a relatively fast and accurate analysis of
petroleum samples. This procedure eliminates the typical drawbacks of ICP-MS analysis of
organic solutions, such as matrix signal suppression, carbon deposition, and the formation of
interfering ions in plasma.68 However, the DL of this method is too high (11 ng.g−1) for
application to liquid hydrocarbons.
4.2. Hg Speciation
Speciation analysis of Hg in liquid hydrocarbons can be performed by different

methods. At present, there is no method capable of simultaneously detecting all species
contained in liquid hydrocarbon samples because of the different properties of Hg species.1
Different methods have been tested for the determination of Hg speciation in liquid
hydrocarbons. We can categorize Hg separation methods into two classes: chromatographic
techniques and sequential extractions. Since a review published in 2000, 71 few studies have
attempted new methodologies for the speciation of Hg in liquid hydrocarbons. Table 3 shows
the commercial analytical devices dedicated to the analysis of Hg speciation without the use of
liquid samples.
42
Table 3. Commercial Instruments Dedicated to Hg Analysis in Liquid and Solid Samples and to Hg Speciation in Liquid Samples
Determined
Brand Name Hg species Detector Analysis Samples DL Weight (kg)
Leco AMA-254 Total Hg AAS Direct analysis (thermal Solid/liquids 46 kg

decomposition), Lab
Milestone DMA-80 Total Hg AAS Direct analysis (thermal Solid/liquids 0.001 ng 56 kg

decomposition + amalgamation),
DMA-1 Total Hg AAS Lab Solid/liquids 0.001 ng
FMA Total Hg AFS and Lab, liquids 0.01 ng/g

AAS
Nippon instruments MA-3000 Total Hg AAS Direct analysis (thermal Lab 55 kg

Corp. decomposition), Lab
PE-1000 Total Hg AFS Direct analysis (Petro-pyrolysis), Specific liquid 40 kg

Lab hydrocarbons
Perkin Elmer SMS 100 Total Hg AAS Direct analysis (thermal Solid/liquids 0.005 ng 20 kg
decomposition + amalgamation),
Lab
Brooks Rand Model III + MERX Total Hg AFS Direct analysis (thermal Solid/liquids < 0.03 ng/L
Total Hg Purge & decomposition + amalgamation),
Trap Module Lab
PSA 10.025 Millenium Total Hg AFS or Direct analysis (thermal Solid/liquids

Merlin AAS decomposition + amalgamation),
Lab
LUMEX RA-915 M + PYRO- Total Hg AAS Direct analysis (Pyrolyse), Lab Liquid 0.01 ng/g
915
43
Buck Scientific 410 cold vapor Total Hg AAS Lab Solid/liquids 0.0005 ng/g 20 kg
mercury analyzer
HIRANUMA HG-400 Total Hg AAS Lab Solid/liquids 0.01-10 ng/g 4 kg
Mercury LabAnalyzer LA 254 Total Hg AAS Lab Only liquids 5 pg ng/g 10 kg

instruments
AULA-254 Gold Total Hg AAS Direct analysis, Lab Only liquids 5 ng/L 18.6 kg
Teledyne Hydra IIAA Total Hg AAS Direct analysis, Lab Liquids 0.001 g
Hydra IIC Total Hg AAS Direct analysis (thermal Solid and 0.05 ng/L
decomposition + amalgamation), semisolid
Lab
M-8000 Total Hg AFS Lab Liquids 0.5 ng/L
M-7600 Total Hg AAS Lab Sub-pg
TEKRAN 2600 Total Hg AFS Lab
PSA 10.725 GC Speciation GC-AFS Lab Liquids
10.820 Speciation HPLC- Direct analysis (Online UV Liquids

AFS digestion), Lab
Brooks Rand MERX-M Speciation GC-AFS Direct analysis (Online Liquids 0.002 ng/L
ethylation), Lab
TEKRAN 2700 Speciation GC-AFS Lab Liquids 0.002 ng/L
44
4.2.1. Sample Preparation: Operationally Defined Procedures
The UOP938 protocol proposes a methodology for the speciation of Hg in liquid

hydrocarbons (UOP Method 938-10, n.d, n.d.). This technique uses three extraction steps, the
first of which consists of sample filtration to determine the particulate Hg fraction (Figure 3).
The UOP938 method recommends performing pressure filtration with 0.22 μm polyvinylidene
fluoride filters. The second step consists of purging and trapping the volatile Hg species (Hg0
and eventually DMHg) contained in the sample. After this second step, only soluble Hg
(oxidized species) remains in the sample. The third step consists of a liquid−liquid extraction
step to separate the ionic and organic Hg species before their analysis. The UOP938 method
recommends using a 0.01% sodium chloride solution in water with a volume ratio of 1:5 for
this liquid−liquid extraction.
The MDL for each species depends, in part, on the sensitivity of the detector. The
volatile Hg and particulate Hg concentrations also depend on the size of the sample due to the
preconcentration step (if the filters are analyzed after thermal decomposition and
amalgamation). This method has both advantages and disadvantages: it allows for onsite sample
processing for the first steps of the procedure, which is an important advantage because it
prevents Hg losses and species interconversion during storage.10 The purging step allows for
the collection of volatile Hg0. In the second step volatile Hg0 can be removed from the sample
to avoid its reaction with Hg(II) and, above all, the precipitation of insoluble Hg(I) resulting
from this reaction.21 However, it has been found that when one species dominates a hydrocarbon
sample, it can affect the results for other species.1 In addition, this method does not allow for
the identification of organic species of Hg (MMHg, EtHg, etc.), and MMHg is generally the
dominant species.1 Recently, Gaulier et al.5 optimized the three steps of the UOP938 method to
make it more robust and accurate. They showed that the technique could be more robust by
measuring the Hg recovered from the filter instead of quantifying the difference between the
unfiltered and filtered solutions, as suggested in the UOP938 method.
45
Figure 3. Diagram of mercury speciation analysis in liquid hydrocarbon by UOP Method 938-10
(adapted with permission from ref 5).
Similarly, recovering the Hg0 purged using a gold amalgamator with subsequent analysis
provides a better estimate of its concentration. Sodium chloride used for successive
liquid−liquid extraction can damage some analyzers, and an alternative L-cysteine solution was
reported to give similar extraction yields (95%) with fewer analytical issues than those
occurring with the use of sodium chloride.5
Recently, Gajdosechova et al.72 developed a method based on headspace gas sampling.
Hg compounds present in the organic matrix were first extracted by chelation with dithizone
(diphenylthiocarbazone), and then sodium tetrapropyl borate was added to volatilize the
mercury species. These species were then collected and preconcentrated on ITEX tubes (online
preconcentration using in-tube extraction). This sample preparation resulted in a very low DL
for MeHg (0.06 ng.g−1). Nevertheless, these preparation techniques are specific and generally
allow for only certain Hg species to be extracted, contrary to the UOP938 protocol.
4.2.2. Gas Chromatography
GC is a widely used technique for the analysis of Hg speciation in

petroleum.3,11,59,67,72−75 Coupled with suitable detection instruments such as AFS, ICP-MS, or
MIP-AES, this technique is capable of separating all volatile Hg species. Only Hg0 and dialkyl-
46
Hg species are volatile,10,18 while monoalkyl-Hg and ionic Hg remain in solution.3 A
derivatization step is necessary to convert Hg(II) to dibutyl-Hg and methyl-Hg to methylbutyl-
Hg. A Grignard reagent, generally butylmagnesium chloride (BuMgCl), must be used to carry
out this derivatization step.11,74
2 𝐵𝑢𝑀𝑔𝐶𝑙 + 𝐻𝑔2+ = 𝐵𝑢2 𝐻𝑔 + (3)
𝐵𝑢𝑀𝑔𝐶𝑙 + 𝐶𝐻3 𝐻𝑔+ = 𝐵𝑢𝐶𝐻3 𝐻𝑔 + (4)
Analysis of the samples before and after derivatization theoretically allows for complete
speciation of Hg in a condensate sample.
Despite the robustness and sensitivity of this technique, some Hg compounds present in
the samples may interfere with the analysis. Gajdosechova et al.11 showed that the GC input
temperature can affect the Hg0 signal during analysis if particulate HgS is present in the sample.
They hypothesized that particulate Hg could release Hg0 when it underwent thermal
decomposition when samples were introduced into the GC instrument. Gajdosechova et al.11
therefore recommends that these HgS particles be removed from the sample prior to speciation
analysis of the sample.
Although no speciation method can allow for the direct and complete speciation of Hg,
GC coupled to ICP-MS is capable of quantifying all Hg species (except for particulate Hg).
Presently, there is no other method capable of such achievement. GC has been coupled to
different detectors, but to date, the most robust and the most sensitive is ICP-MS.67,74 The use
of MIP-AES was found to be impractical because of background interferences due to organic
compounds; its use is, however, possible if the eluted Hg is first preconcentrated in a gold
amalgamator.75
4.2.3. HPLC Separation
HPLC is less commonly used than GC for the separation of Hg species. Bouyssiere et
al.71 considered HPLC to be less effective than GC for the separation of Hg species. This
method does not require a derivatization step, which allows for a shorter analysis time than that
of GC.59 Schickling and Broekaert53 coupled HPLC separation with CV-AAS detection.
Detection limits of approximately 10 ng.mL−1 were achieved for Hg species, which were well
47
above those found using the GC-ICP-MS methods (<1 ng.mL-1). In addition, transient signals
and baseline instability were observed.
More recent studies have used HPLC for the separation of Hg species.57,76 Both of these
studies must have used a digestion or extraction step before HPLC-ICP-MS detection. Yun et
al.57 tested different extraction/digestion procedures and found the best recovery rates by
microwave-assisted extraction with the addition of tetramethylammonium hydroxide (TMAH).
Despite this, the recovery rates were only 87% and 71% for MeHg and EtHg, respectively. The
addition of an extraction or digestion step before determining the speciation of Hg is impractical
since the initial advantage of HPLC-based systems is to avoid treatment steps and promote the
precision of the method.53 HPLC-ICP-MS analysis might be combined with an isotope addition
procedure to take into account Hg species conversions and losses.76
4.2.4. Gel Permeation Chromatography
Size-exclusion chromatography (SEC) and gel permeation chromatography (GPC) are

not yet widely used for the separation of Hg species. Until now, only one study has investigated
the potential of this chromatographic technique for Hg speciation in liquid hydrocarbons.5 This
method allows for the separation of species by hydrodynamic volume.
This method may have significant advantages over other Hg speciation procedures,71
such as the GC-ICP-MS technique, as it requires no chemicals and does not affect the
association of Hg with functional groups until they reach the plasma. Separation by GPC is
carried out according to the hydrodynamic volume, and the large molecules elute earlier than
small molecules, which can take a longer path in the porous GPC column.
This technique allows for the co-monitoring of various elements, particularly sulfur,
which can provide crucial information on Hg-binding molecules. Indeed, some sulfur
compounds interact with mercury in liquid hydrocarbon samples.9,24 The use of this technique,
coupled with an ICP-MS detector, makes it possible to simultaneously monitor various
elements, including Hg and sulfur, and potentially to identify the important blinding ligands for
oxidized Hg.
However, this method was not evaluated quantitatively. Although very good trends were
observed for Hg0 species, no linear relationship was found between the Hg concentration and
the detection signal for other Hg species, presumably because of interactions of oxidized Hg
48
with the column. Further method developments might make GPC-ICP-MS a quantitative and
powerful speciation technique for Hg analysis in the future.
5. Challenges and Perspectives
Total Hg analysis itself in both natural gas and liquid petroleum products can be
efficiently evaluated using the above-described methods. However, the reliability of Hg
measurements is debatable, with the under-evaluated interspecies conversion and losses by
volatilization during sampling and sample storage prior to analysis. More investigations on the
evolution of Hg species concentrations after sampling are required to assess the reliability of
laboratory measurements. The development of robust storage procedure stabilizing Hg species
is one of the potential next important steps. Alternatively, onsite procedures might be better
representative of Hg concentration and speciation in streams. In particular, the volatile Hg
species were shown to be critical.
Many studies reported a dominance of HgII species in petroleum products, yet
identifying Hg as HgII is not precise enough to anticipate its properties and behavior during
petroleum processing and in MRUs. A better characterization of HgII (Hg salts, Hg thiolates,
etc.) might lead to a better understanding of Hg partitioning between the different petroleum
phases and help the development of more efficient treatment units. We admit that combining
such precise characterization to the onsite analysis we previously recommended might be
challenging, which further reinforces the need for more investigations relative to Hg species
conversion following sampling.
Our effort in reviewing all developed methodologies revealed a clear gap. The
assessment of techniques, in particular for Hg speciation, is often based on Hg species spiking
and detection. Although this can be considered as a good proof of concept, the development of
an oil reference material with known Hg speciation seems to be necessary to evaluate speciation
methods. Presently, we could only compare the methods based on the number of Hg species
detected and DLs, which is clearly insufficient for choosing the appropriate technique. In
addition, the different laboratories involved in developing Hg (and especially speciation)
research in the field also have the responsibility to provide additional evidence for the reliability
of their measurements. For this purpose, we suggest the implementation of laboratory inter-
49
comparison campaigns, consisting in the parallel analysis of several samples. Such campaigns
conducted with seawater samples during the past decade notoriously harmonized the results
from different laboratories.77,78
Another difficulty for upstream sector is to obtain reliable estimation of the mercury
content of the gas from new fields during short well tests. As explained above, due to possible
chemical reactions and adsorption−desorption equilibrium, mercury can be retained on all kind
of surface, including on the surface of the well tubing and of the testing equipment. For this
reason, mercury will take some time before its arrival at the separator gas outlet. The lower the
mercury content of the gas, the higher the amount of the gas to be produced before being able
to measure the actual mercury content.
Sampling or measuring mercury downhole during a well test can be a solution to
significantly increase the sensitivity and reduce the duration of the test. However, unlike surface
sampling, corrective action becomes nearly impossible in this case. Thus, traceability of the
operation and consistency check will be the key challenge in this case.
6. Conclusion
We compiled different sampling constraints and analytical methods for the

determination of Hg concentration and speciation in petroleum hydrocarbons. Generally,
almost all methods developed for the quantification of Hg have sufficiently low detection limits.
Hg speciation methods give variable information depending on the technique used. The
presence of different Hg species in petroleum might require further investigations using novel
methodologies. Generally, abundant oxidized Hg is only quantified as the sum of Hg species in
the +2 oxidation state. Recent developments, such as SEC separation or X-ray absorption fine
structure spectroscopy, might better identify the different salts and/or complexes within a single
oxidation state. Such information might be critical for a better understanding of Hg partitioning
in the different hydrocarbon phases.
Overall, Hg concentration and speciation measurements are no longer a challenge,
although some future developments might provide more detailed information on Hg behavior
from extraction to processing. The main concern in Hg analysis remains in sampling and storage
procedures. Because sampling induces a modification of the conditions (pressure, temperature),
the partitioning coefficient between different Hg species might be affected and might ultimately
50
cause volatilization, oxidation, methylation or precipitation of Hg; thus, rapid onsite analysis is
recommended
References
(1) Wilhelm, S. M.; Bloom, N. Mercury in Petroleum. Fuel Process. Technol. 2000, 63, 1−27.
(2) Leeper, J. Mercury-LNG’s Problem. Hydrocarbon Process. 1980, 59, 237−240.
(3) Ezzeldin, M. F.; Gajdosechova, Z.; Masod, M. B.; Zaki, T.; Feldmann, J.; Krupp, E. M.
Mercury Speciation and Distribution in an Egyptian Natural Gas Processing Plant. Energy
Fuels 2016, 30, 10236−10243.
(4) Rumayor, M.; Gallego, J. R.; Rodríguez-Valdes, E.; Díaz- ́ Somoano, M. An Assessment
of the Environmental Fate of Mercury Species in Highly Polluted Brownfields by Means of
Thermal Desorption. J. Hazard. Mater. 2017, 325, 1−7.
(5) Gaulier, F.; Gibert, A.; Walls, D.; Langford, M.; Baker, S.; Baudot, A.; Porcheron, F.;
Lienemann, C. P. Mercury Speciation in Liquid Petroleum Products: Comparison between on-
Site Approach and Lab Measurement Using Size Exclusion Chromatography with High
Resolution Inductively Coupled Plasma Mass Spectrometric Detection (SEC-ICP-HR MS).
Fuel Process. Technol. 2015, 131, 254− 261.
(6) INRS. Fiche D’information à L’intention Des Medecins: Fiche No. 1: Cannabis; Institut
National de Recherche et de Securite: France, 2014.
(7) Lide, D. R. Handbook of Chemistry and Physics, 79th ed; CRC Press: Boca Raton, FL,
1999.
(8) Strem. Chemicals for Research Metals, Inorganics and Organo-metallic, 2001.
(9) Lambertsson, L.; Lord, C. J.; Frech, W.; Björn, E. Rapid Dissolution of Cinnabar in Crude
Oils at Reservoir Temperatures Facilitated by Reduced Sulfur Ligands. ACS Earth Space Chem.
2018, 2, 1022−1028.
(10) Bloom, N. S. Analysis and Stability of Mercury Speciation in Petroleum Hydrocarbons.
Fresenius' J. Anal. Chem. 2000, 366, 438− 443.
(11) Gajdosechova, Z.; Boskamp, M. S.; Lopez-Linares, F.; Feldmann, J.; Krupp, E. M. Hg
Speciation in Petroleum Hydrocarbons with Emphasis on the Reactivity of Hg Particles. Energy
Fuels 2016, 30, 130−137.
(12) Harfoushian, J. H. Quantification of Low Levels of Mercury in Gas Reservoirs Using
Advanced Sampling and Analysis Techniques. Proceedings of SPE Annual Technical
Conference and Exhibition; Society of Petroleum Engineers, 1995.
(13) Larsson, T.; Frech, W.; Björn, E.; Dybdahl, B. Studies of Transport and Collection
Characteristics of Gaseous Mercury in Natural Gases Using Amalgamation and Isotope
Dilution Analysis. Analyst 2007, 132, 579−586.
(14) Behra, P.; Bonnissel-Gissinger, P.; Alnot, M.; Revel, R.; Ehrhardt, J. J. XPS and XAS
Study of the Sorption of Hg(II) onto Pyrite. Langmuir 2001, 17, 3970−3979.
51
(15) Ehrhardt, J. J.; Behra, P.; Bonnissel-Gissinger, P.; Alnot, M. XPS Study of the Sorption
of Hg(II) onto Pyrite Fes2. Surf. Interface Anal. 2000, 30, 269−272.
(16) Ryzhov, V. V.; Mashyanov, N. R.; Ozerova, N. A.; Pogarev, S. E. Regular Variations of
the Mercury Concentration in Natural Gas. Sci. Total Environ. 2003, 304, 145−152.
(17) Klusman, R. W.; Jaacks, J. A. Environmental Influences Upon Mercury, Radon and
Helium Concentrations in Soil Gases at a Site near Denver, Colorado. J. Geochem. Explor.
1987, 27, 259−280.
(18) Cameron, C. J.; Didilion, B.; Benayoun, D.; Dorbon, M. Mercury: A Trace Contaminant
in Natural Gas and Natural Gas Liquids. Proceedings Eurogas 96, June 1996.
(19) Wilhelm, S. M.; Liang, L.; Kirchgessner, D. Identification and Properties of Mercury
Species in Crude Oil. Energy Fuels 2006, 20, 180−186.
(20) Wilhelm, S. M.; Kirchgessner, D. A.; Liang, L.; Kariher, P. H. Sampling and Analysis of
Mercury in Crude Oil. J. ASTM Int. 2005, 2, 12985.
(21) Snell, J.; Qian, J.; Johansson, M.; Smit, K. Stability and Reactions of Mercury Species in
Organic Solution. Analyst 1998, 123, 905−909.
(22) ASTM D7482-17. Standard Practice for Sampling, Storage, and Handling of
Hydrocarbons for Mercury Analysis; ASTM International: West Conshohocken, PA, 2017.
(23) Bloom, N. S.; Horvat, M.; Watras, C. J. Results of the International Aqueous Mercury
Speciation Intercomparison Exercise. In Mercury as a Global Pollutant: Proceedings of the
Third International Conference Held in Whistler, British Columbia; Porcella, D. B., Huckabee,
J. W., Wheatley, B., Eds.; Springer Netherlands: Dordrecht, 1995; pp 1257−1268.
(24) Avellan, A.; Stegemeier, J. P.; Gai, K.; Dale, J.; Hsu-Kim, H.; Levard, C.; O’Rear, D.;
Hoelen, T. P.; Lowry, G. V. Speciation of Mercury in Selected Areas of the Petroleum Value
Chain. Environ. Sci. Technol. 2018, 52, 1655−1664.
(25) UOP Method 938-10. Total Mercury and Mercury Species in Liquids Hydrocarbon,
Mercury (Element), Solubility; ASTM International: West Conshohocken, PA, 2018.
(26) Wilhelm, S. M.; Liang, L.; Cussen, D.; Kirchgessner, D. A. Mercury in Crude Oil
Processed in the United States (2004). Environ. Sci. Technol. 2007, 41, 4509−4514.
(27) Norme NF EN 13211. Qualité De L’air - Emissions De Sources Fixes - Methode
Manuelle De Determination De La Concentration En Mercure Total (Air Quality - Fixed
Source Emissions - Manual Method for Total Mercury Determination); Association Francaise
De Normalisation (AFNOR): Paris, France, 2001.
(28) Brombach, C. C.; Pichler, T. Determination of Ultra-Low Volatile Mercury
Concentrations in Sulfur-Rich Gases and Liquids. Talanta 2019, 199, 277−284.
(29) ISO 6978-1:2003, Natural Gas Determination of Mercury. Part 1: Sampling of Mercury
by Chemisorption on Iodine, 2019. https:// www.iso.org/standard/38600.html.
(30) Zhenquan, T.; Quiju, W.; Bin, H. Determination of Mercury Content in Natural Gas by
Chemisorption on Iodine-Impregnated Silica Gel. Chem. Eng. Oil Gas 2009, 2009-03.
(http://en.cnki.com. cn/Article_en/CJFDTotal-STQG200903022.htm)
(31) ISO 6978-2:2003, Natural Gas−Determination of Mercury. Part 2: Sampling of Mercury
by Amalgmation on Gold/Platinum Alloy, 2019. https://www.iso.org/standard/38601.html
52
(32) Dumarey, R.; Dams, R.; Hoste, J. Comparison of the Collection and Desorption
Efficiency of Activated Charcoal, Silver, and Gold for the Determination of Vapor-Phase
Atmospheric Mercury. Anal. Chem. 1985, 57, 2638−2643.
(33) Frech, W.; Baxter, D. C.; Bakke, B.; Snell, J.; Thomassen, Y. Highlight. Determination
and Speciation of Mercury in Natural Gases and Gas Condensates. Anal. Commun. 1996, 33,
7H−9H.
(34) Frech, W.; Baxter, D. C.; Dyvik, G.; Dybdahl, B. On the Determination of Total Mercury
in Natural Gases Using the Amalgamation Technique and Cold Vapour Atomic Absorption
Spectrometry. J. Anal. At. Spectrom. 1995, 10, 769−775.
(35) Enrico, M.; Mere, A.; Zhou, H.; Carrier, H.; Tessier, E.; Bouyssiere, B. Experimental
Tests of Natural Gas Samplers Prior to Mercury Concentration Analysis. Energy Fuels 2020,
34, 5205−5212.
(36) SilcoTek. Prevent Mercury Loss During Transport and Storage with Silconert; SilcoTek
Corp.: Bellefonte, PA, 2019.
(37) Hirner, A. V.; Feldmann, J.; Krupp, E.; Grümping, R.; Goguel, R.; Cullen, W. R.
Metal(Loid)Organic Compounds in Geothermal Gases and Waters. Org. Geochem. 1998, 29,
1765−1778.
(38) Maillefer, S.; Lehr, C. R.; Cullen, W. R. The Analysis of Volatile Trace Compounds in
Landfill Gases, Compost Heaps and Forest Air. Appl. Organomet. Chem. 2003, 17, 154−160.
(39) Pandey, S. K.; Kim, K. H. Experimental Bias Involved in the Collection of Gaseous
Elemental Mercury by the Gold Amalgam Method. Environ. Eng. Sci. 2008, 25, 255−264.
(40) Beauchamp, J.; Herbig, J.; Gutmann, R.; Hansel, A. On the Use of Tedlar® Bags for
Breath-Gas Sampling and Analysis. J. Breath Res. 2008, 2, 046001.
(41) Cachia, M.; Bouyssiere, B.; Carrier, H.; Garraud, H.; Caumette, ̀ G.; Le Hecho, I.
Development of a High-Pressure Bubbling Sampler ́ for Trace Element Quantification in
Natural Gas. Energy Fuels 2017, 31, 4294−4300.
(42) Sanchez-Rodas, D.; Corns, W. T.; Chen, B.; Stockwell, P. B. ́ Atomic Fluorescence
Spectrometry: A Suitable Detection Technique in Speciation Studies for Arsenic, Selenium,
Antimony and Mercury. J. Anal. At. Spectrom. 2010, 25, 933−946.
(43) ASTM D5954-98. Standard Test Method for Mercury Sampling and Measurement in
Natural Gas by Atomic Absorption Spectroscopy; ASTM International: West Conshohocken,
PA, 2014.
(44) ASTM D6350-14. Standard Test Method for Mercury Sampling and Analysis in Natural
Gas by Atomic Fluorescence Spectroscopy; ASTM International: West Conshohocken, PA,
2014.
(45) Liang, L.; Horvat, M.; Cernichiari, E.; Gelein, B.; Balogh, S. Simple Solvent Extraction
Technique for Elimination of Matrix Interferences in the Determination of Methylmercury in
Environmental and Biological Samples by Ethylation-Gas ChromatographyCold Vapor Atomic
Fluorescence Spectrometry. Talanta 1996, 43, 1883−1888.
(46) Vicentino, P. O.; Brum, D. M.; Cassella, R. J. Development of a Method for Total Hg
Determination in Oil Samples by Cold Vapor Atomic Absorption Spectrometry after its
Extraction Induced by Emulsion Breaking. Talanta 2015, 132, 733−738.
53
(47) Pontes, F. V. M.; Carneiro, M. C.; Vaitsman, D. S.; Monteiro, M. I. C.; Neto, A. A.;
Tristao, M. L. B.; Guerrante, M. F. Comparative ̃ Study of Sample Decomposition Methods for
the Determination of Total Hg in Crude Oil and Related Products. Fuel Process. Technol. 2013,
106, 122−126.
(48) Pereira, L.; Maranhao, T. A.; Frescura, V. L. A.; Borges, D. L. G. ̃ Multivariate
Assessment of Extraction Conditions for the Fractionation Analysis of Mercury in Oily Sludge
Samples Using Cold Vapor Atomic Fluorescence Spectrometry. J. Anal. At. Spectrom. 2019,
34, 1932−1941.
(49) Munoz, R. A. A.; Correia, P. R. M.; Nascimento, A. N.; Silva, C. S.; Oliveira, P. V.;
Angnes, L. Electroanalysis of Crude Oil and Petroleum-Based Fuel for Trace Metals:
Evaluation of Different Microwave-Assisted Sample Decompositions and Stripping
Techniques. Energy Fuels 2007, 21, 295−302.
(50) Knauer, H. E.; Milliman, G. E. Analysis of Petroleum for Trace Metals. Determination
of Mercury in Petroleum and Petroleum Products. Anal. Chem. 1975, 47, 1263−1268.
(51) Liang, L.; Horvat, M.; Fajon, V.; Prosenc, N.; Li, H.; Pang, P. Comparison of Improved
Combustion/Trap Technique to Wet Extraction Methods for Determination of Mercury in
Crude Oil and Related Products by Atomic Fluorescence. Energy Fuels 2003, 17, 1175−1179.
(52) Kelly, W. R.; Long, S. E.; Mann, J. L. Determination of Mercury in SRM Crude Oils and
Refined Products by Isotope Dilution Cold Vapor ICP-MS Using Closed-System Combustion.
Anal. Bioanal. Chem. 2003, 376, 753−758.
(53) Schickling, C.; Broekaert, J. A. C. Determination of Mercury Species in Gas Condensates
by on-Line Coupled High-Performance Liquid Chromatography and Cold-Vapor Atomic
Absorption Spectrometry. Appl. Organomet. Chem. 1995, 9, 29−36.
(54) Brandao, G. P.; de Campos, R. C.; Luna, A. S. Determination of ̃ Mercury in Gasoline by
Cold Vapor Atomic Absorption Spectrometry with Direct Reduction in Microemulsion Media.
Spectrochim. Acta, Part B 2005, 60, 625−631.
(55) Conaway, C. H.; Mason, R. P.; Steding, D. J.; Russell Flegal, A. Estimate of mercury
emission from gasoline and diesel fuel consumption, San Francisco Bay area, California. Atmos.
Environ. 2005, 39, 101−105.
(56) Uddin, R.; Al-Fahad, M. A.; Al-Rashwan, A. K.; Al-Qarni, M. A. A Simple Extraction
Procedure for Determination of Total Mercury in Crude Oil. Environ. Monit. Assess. 2013, 185,
3681−3685.
(57) Yun, Z.; He, B.; Wang, Z.; Wang, T.; Jiang, G. Evaluation of Different Extraction
Procedures for Determination of Organic Mercury Species in Petroleum by High Performance
Liquid Chromatography Coupled with Cold Vapor Atomic Fluorescence Spectrometry. Talanta
2013, 106, 60−65.
(58) Osborne, S. P. Quantitation of Mercury in Petroleum by ETV-ICP-MS. Appl. Spectrosc.
1990, 44, 1044−1046.
(59) Tao, H.; Murakami, T.; Tominaga, M.; Miyazaki, A. Mercury Speciation in Natural Gas
Condensate by Gas Chromatography-Inductively Coupled Plasma Mass Spectrometry. J. Anal.
At. Spectrom. 1998, 13, 1085−1093.
54
(60) Shafawi, A.; Ebdon, L.; Foulkes, M.; Stockwell, P.; Corns, W. Determination of Total
Mercury in Hydrocarbons and Natural Gas Condensate by Atomic Fluorescence Spectrometry.
Analyst 1999, 124, 185−189.
(61) Ceccarelli, C.; Picon, A. R.; Mariangel, P. P.; Greaves, E. D. ́ Total Mercury
Determination in naphthas by Either Atomic Fluorescence or Absorption Spectroscopy. Pet.
Sci. Technol. 2000, 18, 1055−1075.
(62) Liang, L.; Lazoff, S.; Horvat, M.; Swain, E.; Gilkeson, J. Determination of Mercury in
Crude Oil by In-Situ Thermal Decomposition Using a Simple Lab Built System. Fresenius' J.
Anal. Chem. 2000, 367, 8−11.
(63) Bouyssiere, B.; Ordóñez, Y. N.; Lienemann, C. P.; Schaumlöffel, D.; Łobinski, R.
Determination of Mercury in Organic ́ Solvents and Gas Condensates by μflow-Injection-
Inductively Coupled Plasma Mass Spectrometry Using a Modified Total Consumption
Micronebulizer Fitted with Single Pass Spray Chamber. Spectrochim. Acta, Part B 2006, 61,
1063−1068.
(64) Torres, D. P.; Dittert, I. M.; Höhn, H.; Frescura, V. L. A.; Curtius, A. J. Determination of
Mercury in Gasoline Diluted in Ethanol by GF AAS after Cold Vapor Generation, Pre-
Concentration in Gold Column and Trapping on Graphite Tube. Microchem. J. 2010, 96,
32−36.
(65) Olsen, S. D.; Westerlund, S.; Visser, R. G. Analysis of Metals in Condensates and naphtha
by Inductively Coupled Plasma Mass Spectrometry. Analyst 1997, 122, 1229−1234.
(66) Kumar, S. J.; Gangadharan, S. Determination of Trace Elements in naphtha by
Inductively Coupled Plasma Mass Spectrometry Using Water-in-Oil Emulsions. J. Anal. At.
Spectrom. 1999, 14, 967−971.
(67) Bouyssiere, B.; Szpunar, J.; Lobinski, R. Gas Chromatography with Inductively Coupled
Plasma Mass Spectrometric Detection in Speciation Analysis. Spectrochim. Acta, Part B 2002,
57, 805−828.
(68) Vorapalawut, N.; Pohl, P.; Bouyssiere, B.; Shiowatana, J.; Lobinski, R. Multielement
Analysis of Petroleum Samples by Laser Ablation Double Focusing Sector Field Inductively
Coupled Plasma Mass Spectrometry (LA-ICP MS). J. Anal. At. Spectrom. 2011, 26, 618−622.
(69) da Silva, C. S.; Oreste, E. Q.; Nunes, A. M.; Vieira, M. A.; Ribeiro, A. S. Determination
of Mercury in Ethanol Biofuel by Photochemical Vapor Generation. J. Anal. At. Spectrom.
2012, 27, 689−694.
(70) de Jesus, A.; Zmozinski, A. V.; Vieira, M. A.; Ribeiro, A. S.; da Silva, M. M.
Determination of Mercury in naphtha and Petroleum Condensate by Photochemical Vapor
Generation Atomic Absorption Spectrometry. Microchem. J. 2013, 110, 227−232.
(71) Bouyssiere, B.; Baco, F.; Savary, L.; Lobinski, R. Analytical Methods for Speciation of
Mercury in Gas Condensates: Critical Assessment and Recommendations. Oil Gas Sci.
Technol. 2000, 55, 639−648.
(72) Gajdosechova, Z.; Pagliano, E.; Zborowski, A.; Mester, Z. Headspace in-Tube
Microextraction and GC-ICP-MS Determination of Mercury Species in Petroleum
Hydrocarbons. Energy Fuels 2018, 32, 10493−10501.
(73) Kawasaki, M.; Otsuka, M.; Yamada, J.; Kobayashi, A. Identification of Mercury Species
in Liquid Hydrocarbons. J. Jpn. Pet. Inst. 2016, 59, 235−240.
55
(74) Pontes, F. V. M.; Carneiro, M. C.; Vaitsman, D. S.; Monteiro, M. I. C.; Neto, A. A.;
Tristao, M. L. B. Investigation of the Grignard ̃ Reaction and Experimental Conditions for the
Determination of Inorganic Mercury and Methylmercury in Crude Oils by GC−ICP−MS. Fuel
2014, 116, 421−426.
(75) Snell, J. P.; Frech, W.; Thomassen, Y. Performance Improvements in the Determination
of Mercury Species in Natural Gas Condensate Using an on-Line Amalgamation Trap or Solid-
Phase Micro-Extraction with Capillary Gas Chromatography−Microwave−Induced Plasma
Atomic Emission Spectrometry. Analyst 1996, 121, 1055−1060.
(76) Rahman, G. M. M.; et al. Speciation of Mercury in Crude Oil Using Speciated Isotope
Dilution Mass Spectrometry. Spectroscopy 2010, 25.
(77) Filby, R. H. Origin and nature of trace element species in crude oils, bitumens and
kerogens: implications for correlation and other geochemical studies. Geol. Soc. Spec. Publ.
1994, 78, 203−219.
(78) Lamborg, C. H.; Hammerschmidt, C. R.; Gill, G. A.; Mason, R. P.; Gichuki, S. An
intercomparison of procedures for the determination of total mercury in seawater and
recomandations regarding mercury speciation during GEOTRACES cruises. Limnol.
Oceanogr.: Methods 2012, 10 (2), 90−100
56
Partie 2: Arsenic analysis in the petroleum industry:
a review
ACS Omega 2022, 7, 43, 38150–38157 - https://doi.org/10.1021/acsomega.2c03708
Received: June 14, 2022; Accepted: August 26, 2022; Publication Date: October 18, 2022
Aurore Mere1,2,3, Maxime Enrico1,2, Honggang Zhou3, Emmanuel Tessier1,2, Brice

Bouyssiere1,2*
1
Institut des Sciences Analytiques et de Physico-Chimie Pour L'environnement et les Materiaux, CNRS/ UNIV
2
3
TOTALEnergies, CSTJF, Av. Larribau, 64018 Pau, France
Abstract
The presence of arsenic in natural gas and liquid hydrocarbons is of great concern for
oil companies. In addition to health risks due to its toxicity as well as environmental issues,
arsenic is responsible for irreversible poisoning of catalysts and clogging of pipes via the
accumulation of As-containing precipitates. To address these problems and to better design
treatment units, robust methods for the analysis of arsenic and its compounds in oil streams are
required. In addition, the use of feedstocks as a novel source of energy is becoming increasingly
important. Most biomasses used as feedstocks are contaminated with arsenic. To avoid
problems related to the presence of this element, it is therefore also necessary to have reliable
methods for the analysis of arsenic and its compounds in these new fluids. This review outlines
the sampling techniques, sample preparation methods and arsenic analysis techniques
developed during recent decades and commonly used in the oil industry and in the new
feedstocks energy domain.
57
1. Introduction
Arsenic (As) is the 20th most abundant element in the Earth's crust and is generally
found in its inorganic form in soils, rocks and water1 at concentrations of several mg/kg.2 As is
also present in biota, especially marine organisms, and in crude oil.3 Marine organisms have
the ability to bioaccumulate arsenic, with concentrations reaching up to 2 g kg-1 dry weight.4
Long-term decomposition of marine organisms gives rise to oil with the persistence of As in
the oil matrix,2 with a suspected role of alkylation/dealkylation reactions in the solubilization
of As.2,5
The presence of arsenic in petroleum fluids is the source of various problems in oil
companies. First, As represents a health and safety risk for personnel in contact with these
fluids. Arsenic might be responsible for skin lesions, respiratory diseases, and cardiovascular
diseases and can cause several long-term cancer developments.6-8 Another issue relates to the
environmental impact of arsenic released to the atmosphere by natural gas burning.9 From an
industrial point of view, there are two major problems related to the occurrence of As. First, the
precipitation of As-containing particles leads to the clogging of pipes and valves. This
phenomenon was first observed in 1987 after the blockage of pipes in a US gas distribution
company. Delgado-Morales et al.5 suggested that the formation of these solid deposits occurred
in the presence of a metallic substrate and were formed from arsenic and H2S. Irgolic et al.2
identified volatile alkylarsines as the source of these residues. The second major concern relates
to catalyst poisoning during steam cracking and natural gas processing. Small amounts of As
might cause severe and irreversible damage to catalysts, representing high financial costs.9,10
The total concentration range of arsenic in natural gas was determined to be 0.01 to 63
µg.dm-3.2 It was later demonstrated that different organic and inorganic forms of As are
present,11,12 with a dominance of trimethylarsine (TMA, (CH3)3As), triethylarsine ((C2H5)3As),
triphenylarsine ((C6H5)3As) and arsine (AsH3).4,9,13,14 TMA is the most abundant volatile As
species in natural gas (55 to 80% of the total As in natural gas).5 The dominance of TMA is
thought to be related to its higher stability over time compared than that of other As
compounds.4
Few articles mention the total amount of As in crude oil; only Olsen et al.15 and Stigter
et al.16 were able to identify a concentration range from < 10 to 26.2 µg.kg-1 As in real crude
oil samples. The main forms of arsenic in these matrices are TMA, which is predominant,
dimethylated arsenic ((CH3)2As) and monomethylated arsenic ((CH3)As).3
58
Arsenic removal systems can be set up by oil companies. The development of these
units is based on the total arsenic concentration and, if possible, on the speciation of arsenic in
the oil fluids. Therefore, before they are set up, it is essential to determine the arsenic content
of natural gas and liquid hydrocarbons. This methodology will also be of great importance in
the field of new energy feedstocks because As can be present in biogas (and thus biomethane)17
and in bio-oil during the hydroliquefaction (HTL) of biomass such as swine manure, which may
contain high levels of arsenic.18
2. Sampling and Analysis of As in Natural Gas
2.1. Sampling
2.1.1. As-Selective Sorption
2.1.1.1. Bubbling systems
This sampling technique consists of bubbling natural gas in one bubbler2 or several
bubblers4,9,14,19 connected in series containing a trapping solution. Several trapping solutions
can be used for this technique, but they must be able to oxidize volatile arsenicals for their
pentavalent oxidizing species, such as oxidizing organic arsenic to the arsenate ion (AsO43-).4
Concentrated nitric acid (HNO3) can be used as a trapping solution.2,14,19 However, it
has been shown that a 2% aqueous silver nitrate (AgNO3) solution provides better TMA
trapping efficiency than a concentrated nitric acid solution.4,14 Despite this, it has been shown
that the arsenic trapping efficiency in a 2% AgNO3 solution is only 70%. These results were
obtained by comparing the amount of TMA found in a natural gas sample after bubbling in this
solution and analyzing it directly by cryotrapping/gas chromatography–inductively coupled
plasma–mass spectrometry (Cryo/GC–ICP–MS). This loss of efficiency can be explained by
the potential absorption of arsenic on the walls of the bubblers.9
Furthermore, if the natural gas contains a high concentration of H2S, a black solid is
formed in the 2% AgNO3 solution, which inhibits the analysis of this solution.14
59
2.1.1.2. Traps
This sampling technique involves passing a natural gas stream through a tube containing
a solid sorbent. This technique allows for large volumes of gas to be sampled.20 The sorbent
used is silica gel impregnated with 1%21 or 2%9 AgNO3. When the gas stream is passed through
the trap, the arsenic retained within it is desorbed with concentrated HNO3. The eluate is then
analyzed either by UV irradiation–hybride generation–atomic fluorescence spectrometry (UV–
HG–AFS), by ICP–MS or by flame atomic absorption spectrometry (FAAS) (see Section 3.2.2
Analysis after sampling). Uroic et al.21 obtained a TMA absorption efficiency of 99.8% with
this type of trap. However, Krupp et al.9 obtained a TMA absorption efficiency of only 70%
with this type of trap. This absorption efficiency was obtained by comparing the amount of
TMA found in a natural gas sample after trapping on a silver nitrate-impregnated silica gel trap
and by direct analysis of the sample by Cryo/GC–ICP–MS. This result is considered to be due
to incomplete absorption inside the trap due to the very low polarity of TMA or incomplete
desorption of TMA during its extraction from the trap with nitric acid.9
2.1.2. Gas Samplers: Cylinders and Inert Plastic Bags (Tedlar Bags)
Natural gas sampling can be conducted using pressurized cylinders or Tedlar bags.
Arsenic analysis in natural gas sampled with these two types of samplers can be determined
directly by Cryo/GC–ICP–MS.9 Arsenic can also be determined by releasing some or all of the
sampled gas into a silicon trap impregnated with AgNO3, where the eluate (see Section 3.1.1.2
Traps) from this tube is then analyzed by UV–HG–AFS (see Figure 4).21
Pressurized cylinders can be used to sample gases at high pressures directly on the
production system or in the well under initial pressure.2 Although natural gas sampling with
pressurized cylinders is a fast and easy-to-handle method, arsenic depletion has been observed
in natural gas samples stored in steel cylinders.4,5 To overcome this problem, a special coating
(Silcosteel® or Sulfinert®) has been designed to create an inert internal surface in these
cylinders to limit the interactions between arsenic and the cylinder walls.
Inert plastic bags (Tedlar bags) can also be used to sample natural gas.22 However,
unlike cylinders, natural gas must be depressurized to be sampled in Tedlar bags. It has been
shown that within 24 hours, there is no significant loss in the various volatile arsenic species
contained in these bags. However, it has been shown that over time, there is adsorption of
60
arsenic compounds to the inner surface of the Tedlar bags, which suggests that arsenic analysis
should be carried out no later than 24 hours after sampling.20
2.2. Analysis
2.2.1. Direct Analysis
Total arsenic and its species in natural gas can be analyzed directly without a sample
preparation and preconcentration step. Freije-Carrelo et al.23 developed a GC–ICP–MS method
that allows for the simultaneous analysis of total arsenic and its speciation in natural gas. To
perform this simultaneous analysis, two gas injection valves were interconnected. The first
valve directs the gas sample into the analytical column (GC column) for arsenic speciation
analysis. The second valve allows for the same gas sample to be directed into a transfer line
directly connected to the GC–ICP–MS interface to analyze the total arsenic by continuous flow
injection analysis (FIA). The detection limits obtained with this technique are very low, i.e., 2
ppt for total analysis and 12 ppt for speciation, and matrix effects are negligible.23
Arsenic species in natural gas can also be analyzed directly using the cryotrapping
technique. This technique is widely used in the research laboratory, but there are no publications
mentioning its direct use in the field. This technique does not require any sample preparation
beforehand. This method consists of trapping gas samples in a capillary column 2 or a GC
column22 or in a glass tube filled with a chromatographic packing.24 This trapping is carried out
cryogenically using liquid hydrogen,9 liquid nitrogen20 or an acetone/liquid nitrogen mixture.25
The arsenic species contained in the gas samples are then thermally desorbed from these
columns/tubes using temperature ramps reaching a maximum temperature of 160°C to 290°C.
The carrier gas used is helium.
Arsenic species can be analyzed using different techniques/detectors and can first be
separated by GC. After separation, As species can be analyzed by MS. The detection limit
obtained with this technique is 0.5 ng.dm-3.2 After this separation by GC, the As species can
also be analyzed by ICP–MS. Detection limits are generally on the order of a few ng.m-3 but
can vary depending on the amount of gas analyzed.9,20-22,24 Arsenic species in the analyzed gases
can also be directly analyzed by ICP–MS without a GC separation phase.25
The cryotrapping technique coupled with ICP–MS is an extremely sensitive and
reproducible method, and when coupled with GC–ICP–MS, the speciation of arsenic in natural
61
gas can be analyzed. Furthermore, the trapping efficiency of cryotrapping is 100% compared
to that obtained with bubbling solutions and traps.9 However, the technique is very expensive
and cannot be used directly in the field.
Uroic et al.21 developed the cryotrapping/UV–HG–AFS method, which is a promising
technique for direct field use for the measurement of arsenic compounds in natural gas. The
cryotrapping and coupling technique was equivalent to that used by Cryo/GC–ICP–MS
described in this same publication. The only difference is that the detector used was an AFS
and that a precursor hydride generation step was performed. A detection limit of 1 µg.m -3 for
20 L of natural gas was obtained. Despite the lack of sensitivity and precision of this technique
compared to Cryo/GC–ICP–MS, it is inexpensive and can be used to carry out preliminary
measurements directly in the field before moving on to more precise analyses.
2.2.2. Analysis after preconcentration
The bubbling solutions (see Section 3.1.1.1. Bubbling systems) obtained after bubbling
the gas sample can be analyzed by ICP–MS to obtain the total arsenic concentration .4,14,19 ICP–
MS is a very sensitive and selective analytical method for the analysis of total arsenic in
bubbling/trapping solutions.4 The detection limit found in the trapping solution and with this
type of detector (analysis with reaction cell) is 0.09 ppb.14 To analyze the total arsenic in the
trapping solutions, As species can also be mineralized (addition of nitric acid and concentrated
hydrogen peroxide) by microwave assistance to transform all arsenic compounds into arsenate
and then analyzed by FAAS (see Figure 4). The detection limit obtained with this analytical
method was 5.0 µg.L-1.9 Irgolic et al.2 were able to determine the total arsenic in their trapping
solutions by mineralizing these solutions with sulfuric acid. These authors were also able to
perform speciation by directly analyzing these trapping solutions by GC–MS after hydride
generation without going through the mineralization step.2
The eluate obtained after desorption of arsenic retained on the surface of silica traps can
be analyzed by FAAS or ICP–MS with a collision cell (to eliminate ArCl- interference) for the
determination of total arsenic.9 Total arsenic can also be determined from the eluate by UV–
HG–AFS. UV photooxidation converts organic arsenic compounds to inorganic arsenic, and
the addition of sodium borohydride to the eluate reduces all inorganic arsenic compounds to
arsine. The detection limit obtained with the UV–HG–AFS technique is 1 µg.m-3 for 20 L of
sampled gas.21
62
Figure 4. Diagram of Arsenic Analysis in Natural Gas
3. Sampling and Analysis of As in Hydrocarbon Liquids

3.1. Sample Preparation
3.1.1. Microemulsions
The microemulsion technique consists of stabilizing a liquid hydrocarbon sample in the

form of a microemulsion before analysis. There are two types of microemulsions: those
prepared with surfactants and those prepared without detergents.
Detergent-free microemulsions involve creating a homogeneous solution using three
components. This solution is usually composed of the sample, propan-1-ol and HNO3.10,26
Water can also be added to this solution.27 The volumetric ratios of these components vary
depending on the sample analyzed. However, it was found that when the volume of the sample
was greater than 10% of the final volume of the solution, the solution was not homogeneous.27
It has been shown that there is no loss of arsenic from detergent-free microemulsions over
time.26 Despite this, arsenic recoveries in detergent-free microemulsions are 2.5% lower than
those obtained in surfactant microemulsions.27
63
Surfactant microemulsions consist of mixing the sample with Triton X-100. The
concentration and amount of surfactant added varies from author to author.27-29 When the
sample volume is greater than 20%, the solution obtained with Triton X-100 is biphasic even
after immediate stirring after its addition.27 Microemulsions prepared with Triton X-100 are
stable for at least 2 h.28 The recovery rates of arsenic in microemulsions are 86% with the
addition of 2.5% Triton X-10029 and close to 100% with the addition of 4% Triton X-100.27
Both of these microemulsion techniques do not require any further sample pretreatment, are
rapid techniques and allow for the accurate determination of arsenic in liquid hydrocarbons.27,28
Finally, these preparations reduce background noise and make the signal more stable during
analysis.29
3.1.2. Extraction, Mineralization and Digestion
Liquid hydrocarbons can be mineralized to analyze the arsenic contained in them. This
step destroys the matrix and therefore limits interference with the matrix during the analysis.
Sulfuric acid (H2SO4), HNO3 or hydrogen chloride (HCl) can be used as well as water to carry
out this mineralization.3,30,31 Potassium permanganate (KMnO4) solution can be added to
facilitate mineralization.31 Iodine can also be added to the solution to increase the water
solubility of the arsenic compounds.30
The arsenic compounds can also be extracted from the matrix without destroying the
matrix. In crude oil samples, a liquid–liquid extraction of arsenic compounds can be performed
with HNO3.26 In naphthas, a liquid–liquid extraction of arsenic compounds can be performed
with a solution of sodium hypochlorite (NaOCl), H2SO4 and water followed by digestion of the
obtained aqueous solution (ASTM UOP946-96, n.d.). Arsenic in paraffin or petroleum can be
extracted on a solid surface using a microcrystalline naphthalene or cetyltrimethylammonium
bromide adsorbent. The detection limit obtained with this solid surface extraction is 0.016
ppm.32 Arsenic can be extracted on a solid surface using a Chelex 100 column, but the sample
must first be burnt in an oxygen bomb.33
Liquid hydrocarbons can also be digested to destroy the matrix and analyzed for arsenic.
This digestion is usually an acid digestion where concentrated H2SO4, HNO3 or HCl is added
to the samples to achieve this digestion.16,32-34 Hydrogen peroxide (H2O2) can be added to the
acid solution to promote digestion of the matrix.16
64
Brandão et al.26 determined the concentrations of As found in different liquid
hydrocarbon samples by analyzing them with the microemulsion technique developed by them,
the mineralization technique developed by Aneva and Iancheva,30 the technique EPA SW-846
Test Method 303131 and the technique ASTM UOP946-96.34 The arsenic concentrations found
in these samples using these four sample preparation techniques are equivalent, showing their
accuracy.26 Despite this, arsenic remains one of the most volatile elements, which makes it
difficult to analyze quantitatively, especially when it undergoes sample preparation or matrix
destruction.35
3.1.3. Dilution and Direct Injection
Liquid hydrocarbons can be directly analyzed without sample preparation, which saves
analysis time36 and minimizes sample handling.37
Samples can either be diluted in xylene15,36 or toluene13 or injected directly into the analytical
instrument.37-39 It has been shown that dilution of liquid hydrocarbons in organic solvents
increases sensitivity when these samples are analyzed by ICP–MS.36 However, due to the high
viscosity of liquid hydrocarbons, especially fuel oil,3 the dilution factor must be relatively high,
making it even more difficult to analyze arsenic, which is at trace levels in this type of sample.15
Direct injection of samples without a dilution step can significantly improve the
sensitivity of the method if multiple injections are used.39 If direct injections are carried out to
analyze liquid hydrocarbons by graphite furnace atomic absorption spectrometry (GFAAS), a
chemical modifier must be added to the sample (see Section 4.2.1 Analysis by GFAAS) to
ensure that the pyrolysis temperature is sufficiently high to efficiently destroy the matrix in the
furnace and, above all, that this high pyrolysis temperature does not result in arsenic losses.26
The dilution of liquid hydrocarbon samples followed by their analysis can make it
possible to analyze the speciation of the arsenic contained in them, particularly by capillary
GC–ICP–MS (see Section 4.2.2. Analysis by ICP–MS).13
65
3.2. Analytical Methods
3.2.1. Analysis by GFAAS
Liquid hydrocarbon samples that have undergone sample preparation, such as

microemulsions, extraction or digestion (see Section 4.1. Sampling), can be analyzed by
GFAAS, also known as electrothermal atomic adsorption (ETAAS). This analytical technique
allows for direct analysis of these samples without the need for sample preparation.10,37,39
Atomic absorption spectrometry is one of the most widely used techniques for analyzing liquid
hydrocarbons, but arsenic remains a difficult element to analyze by this technique in complex
matrices.27 When coupled with a graphite furnace, this technique provides good sensitivity for
arsenic analysis in complex matrices and is a relatively affordable technique compared to ICP–
MS.26,37 This technique consists of introducing the sample into a graphite furnace so that it is
first pyrolyzed and then atomized. The arsenic is then analyzed by AAS. Temperatures and
atomization times vary from 2100°C to 2600°C and 2 to 10 seconds, respectively; the carrier
and purge gas used is usually argon.10,26-28,30,31,37,39
To apply such high pyrolysis and atomization temperatures without loss of arsenic, a
chemical modifier must be injected into the furnace during the GFAAS analysis.39 The most
widely used chemical modifier, also known as the "conventional" or "universal" modifier, is
Pb/MgNO3.26,27,31,37 Triton X-100 can be added to this chemical modifier to improve the
analytical signal during analysis.10,36 Other chemical modifiers can be used, such as
Ni(NO3)226,30 or La2O3.39 The latter provides a better detection limit for arsenic analysis in
naphthas of 0.56 µg.L-1 than the detection limit obtained with palladium, rhodium, tungsten,
silver and mixed palladium and magnesium modifiers.39 So-called "permanent" modifiers such
as ruthenium27 or iridium28 can also be used. The iridium permanent modifier has been shown
to improve the sensitivity of the analysis compared to the 'conventional' chemical modifier of
Pb/MgNO3.28
The detection limits obtained for arsenic analysis by GFAAS are between 0.5 µg.L-1 and
5.6 µg.L-1.10,26-28,30,31,37-39 They are presented in Table 4.
66
3.2.2. Analysis by ICP-MS
Microemulsions and diluted liquid hydrocarbons (see Section 4.1 Sampling) can also be
analyzed by ICP–MS. This technique is very sensitive and allows for the analysis of complex
matrices with minimum sample preparation but remains very expensive.26
To introduce the sample into the instrument, Olsen et al.15 used a Meinhardt-Scott
chamber and obtained good analytical precision and a detection limit of 0.2 ng.g -1 for arsenic.
De Albuquerque et al.36 used a concentric micronebulizer coupled to a refrigerated cyclonic
spray chamber and obtained a detection limit of 0.04 µg.kg-1. The carrier and nebulizer gas used
is generally argon.13,15,29,36 To avoid carbon deposition problems and to ensure plasma stability,
O2 should be used as an auxiliary gas in the analysis of liquid hydrocarbons by ICP–MS.29,36
Arsenic can also be analyzed by FI-HG-ICP–MS.36 After acid digestion of the sample,
HCl is added to the digestate to convert As(V) to As(III). The solution is then mixed with
NaBH4 and the hydrides are separated in a liquid–gas separator and introduced directly into the
ICP–MS instrument. The detection limit obtained with this technique is very low, i.e., 0.002
µg.kg-1, but the generation of hydrides lengthens the analysis time compared to analysis by
ICP–MS alone.36
Speciation of arsenic in liquid hydrocarbons, especially in gas condensates, can be
carried out using ICP–MS coupled with capillary GC.13 The absolute detection limit obtained
with this technique is 0.05 pg, but this technique makes it difficult to identify certain peaks of
arsenical compounds that interfere with other peaks of other compounds present in the analyzed
samples.13
3.2.3. Analysis by other methods
Arsenic in liquid hydrocarbon samples can be analyzed by atomic absorption

spectrometry after hydride generation (HG–AAS). Generally, prior to hydride generation,
reagents such as ammonia water and HCl,33 ammonium oxalate ((NH4)2C2O4) (ASTM
UOP946-96, n.d.) or a mixture of hydrochloric acid + hydroxylammonium chloride + potassium
iodide as well as ascorbic acid16 are added to the sample to adjust the pH. To generate the
hydrides, NaBH433 alone or a solution of NaBH4 and sodium hydroxide (NaOH)16,34 are added
to the sample as reducing agents. Generally, HCl is added throughout the analysis, and the
67
measured wavelength for arsenic is 193.7 nm.33,34 The detection limit obtained for arsenic
analysis by HG-AAS is approximately 10 ppb (see Table 1).
After extraction on a solid surface, arsenic in oil can be analyzed by differential pulse
chromatography (DPP).32 HCl-pyridine-NaCl is often used as the supporting electrolyte. This
method is simple, fast and selective and has a detection limit of 0.016 ppm.32
After mineralization, arsenic compounds in crude oil can be analyzed by helium-DC
plasma after hydride generation.3 A buffer of oxalic acid was added to the ore to adjust the pH
to identify and quantify the arsenic compounds in the samples. With this buffer, arsine, arsenate,
methyl arsenate, dimethyl arsenate and trimethyl arsenate compounds were observed in the
crude oil samples.3
Table 4. Summary of Sample Preparation Methods Used fo Arsenic Concentration Analysis with
their Detection Limits (MDLs)
Reference Sample type Sample processing Analysis MDL
Microemulsions
Detergent-free microemulsion
(propan-1-ol + HNO3 + water)
Aucelio and Gasoline and
ETAAS 2 µg.L-1
Curtius27 kerosene
Surfactant emulsion (Triton
X-100 (in HNO3))
Naphtha, gasoline Detergent-free

Becker et al.10 and petroleum microemulsion (propan-1-ol GFAAS 1.9 µg.L-1
condensate + HNO3)
Diesel = 1.8
µg.L-1
Detergent-free
Brandão et al.26 Diesel and naphtha microemulsion (propan-1-ol GFAAS
+ HNO3)
Naphtha = 1.5
µg.L-1
Surfactant emulsion (Triton

Cassella et al.40 Naphtha ETAAS 2.7 µg.L-1
X-100 (in HNO3))
Kumar and Surfactant emulsion (Triton

Naphtha ICP–MS 0.1 µg.L-1
Gangadharan29 X-100 (in water))
Extraction, mineralization and digestion
Aneva and
Gasoline Iode/HNO3 mineralization ETAAS 5.6 µg.L-1
Iancheva30
68
Chemical extraction and
ANSI Webstore34 Naphtha HG–AAS
HNO3/H2SO4/H2O2 Digestion
Brandão et al.26 Gasoline HNO3 extraction GFAAS 1.2 µg.L-1
Dubey et al.32 Oil Extraction in solid surface DDP 0.016 ppm
EPA SW-846 Test

Diesel Chemical mineralization GFAAS
Method 303131
Refined
HG + analysis Absolute
hydrocarbons,
Puri and Irgolic3 HNO3/H2SO4 mineralization helium-DC MDL = 1.0 to
Residual fuel oil
plasma 0.1 ng
and Crude oil
Combustion in an oxygen
Narasaki33 Butter bomb, extraction in solid HG–AAS 5 ppb
surface and HCl digestion
Stigter et al.16 Crude oil cargeos HNO3/H2O2 digestion HG–AAS 10 µg.kg-1
Dilution and direct injection
Absolute
Bouyssiere et al.13 Gas condensate Direct injection GC–ICP–MS MDL = 0.05
pg
ICP–MS 0.04 µg.kg-1

de Albuquerque et
Crude oil Dilution in xylene
al.36
FI-HG-ICPMS 0.002 µg.kg-1
de Jesus et al.38 Crude oil Direct injection GFAAS 5.1 µg.kg-1
Condensate and
Olsen et al.15 Dilution in xylene ICP–MS 0.2 ng.g-1
Naphtha
Naphtha =
0.56 µg.L-1
Naphtha and
Reboucas et al.39 Direct injection ETAAS
Condensate
Condensate =
1.33 µg.L-1
69
4. Sampling and Analysis of As in the New Energy Feedstocks
Domain
Arsenic is present in many biomasses, including plants,41-43 paper,44 pig manure,18

sludge45 and wood.46 The byproducts resulting from this biomass are bio-oils as well as biogas
and can be obtained notably by pyrolysis41-43,45,46 or by coliquefaction.18 These byproducts
formed from these As-contaminated biomasses thus remain contaminated with this element.
To determine the total As concentration in bio-oils, they must first undergo digestion. This
digestion process is usually performed by adding HNO3 and HCl to the bio-oil samples.18,42-
44,46
Digestion can also be performed by adding HNO3 and H2O241 or by adding HNO3, HF and
HClO4.45 Digestion of bio-oil samples can be promoted by performing microwave-assisted
digestion after the addition of reagents.18,42,43,46 Arsenic in these digestates is usually analyzed
by ICP–MS.41-45 Arsenic can also be analyzed in digestate by ICP–emission spectroscopy(ES)46
or by ICP–optical emission spectroscopy (OES), with a detection limit of 25 mg/kg.18 The total
As concentrations found in bio-oils are in the mg.kg-1 range.41,43-45
To analyze total As as well as its speciation in biogas, there are different sampling and
analytical methods. Biogas can bubble in HNO3 and H2O2 solutions contained in 3 bubblers in
series. The total arsenic trapped in these bubbler solutions was then analyzed by ICP–MS.47,48
Arsenic in biogas can also be trapped by passing a biogas stream through a silica trap
impregnated with 1% AgNO3.49 The arsenic retained within the trap is desorbed with boiling
concentrated HNO3. The eluate was then analyzed by UV–HG–AFS to determine total arsenic.
Arsenic speciation can be performed from the eluate of this tube by adding concentrated H2O2
and analyzing this solution by HPLC–IPC–MS.49 With this analytical technique, Mestrot et al.50
was able to quantify TMA, dimethylarsine and arsine in biogas samples. Finally, biogas can be
sampled in Tedlar bags, and the speciation of arsenic contained in it can be directly determined
by Cryo/GC–ICP–MS after transferring the biogas contained in the Tedlar bag to the Cryo-
Catcher column.17
5. Conclusion
Various sampling techniques, sample preparation methods, pretreatment methods, and

analytical methods for As concentration determination in the petroleum industry have been
70
described in this review. However, most articles focus on laboratory methods, but potential
artifacts related to on-site sampling/analysis are not comprehensively discussed. Determining
arsenic compounds in gaseous and petroleum effluents is difficult due to the possibility of
contamination, loss by adsorption, or the occurrence of chemical reactions related to changes
in temperature, pressure, and redox conditions. In the petroleum industry, the difficulty is not
in obtaining samples, performing analyses or even detecting arsenic with a sufficiently low
detection limit but in preparing the sampling campaign while anticipating the validation of the
results by identifying and eliminating all possible artifacts related to the on-site environment.
With the increasing development of new energy feedstocks, most of which contain arsenic, the
determination of arsenic in biogas and bio-oil has become important. The proper consideration
of potential artifacts related to the sampling and analysis of these fluids should be predominant
to validate the concentration of arsenic found in these new fluids. Therefore, additional work is
still needed to identify artifacts and avoid them by using appropriate sampling/analysis
methods.
References
(1) Zheng, L.; Liu, G.; Chou, C.; Gao, L.; Peng, Z. Arsenic in Chinese Coals: Its Abundance,
Distribution, Modes of Occurrence, Enrichment Processes, and Environmental Significance.
Acta Geoscientica Sinica 2006, 27, 355–366.
(2) Irgolic, K. J.; Llger, D.; Zingaro, R. A.; Spall, D.; Puri, B. K. Determination of Arsenic
and Arsenic Compounds in Natural Gas Samples. Appl. Organomet. Chem. 1991, 5, 117–124.
(3) Puri, B. K.; Irgolic, K. J. Determination of Arsenic in Crude Petroleum and Liquid
Hydrocarbons. Environ. Geochem. Health 1989, 11, 95–99.
(4) Liu, Z. X.; Tang, D. Z.; Yan, Q. T.; Xu, H.; Wang, S. Y.; Xu, R.; Wu, S. Characteristic of
the Volatile Arsenic in Natural Gas and its Novel Determination Method. Appl. Mech. Mater.
2014, 522-524, 1515–1521.
(5) Delgado-Morales, W.; Zingaro, R. A.; Mohan, M. S. Arsenic in Natural Gas: Analysis
and Characterization of Pipeline Solids by 1H NMR and Other Methods. Int. J. Environ. Anal.
Chem. 1994, 57, 313–328.
(6) Mazumder, D. G. Chronic Arsenic Toxicity & Human Health. Indian J. Med. Res. 2008,
128, 436–447.
(7) Rahman, M. M.; Ng, J. C.; Naidu, R. Chronic Exposure of Arsenic via Drinking Water
and its Adverse Health Impacts on Humans. Environ. Geochem. Health 2009, 31, 189–200.
71
(8) Yeh, S.; How, S. W.; Lin, C. S. Arsenical Cancer of Skin: Histologic Study with Special
Reference to Bowen's Disease. Cancer 1968, 21, 312–339.
(9) Krupp, E. M.; Johnson, C.; Rechsteiner, C.; Moir, M.; Leong, D.; Feldmann, J.
Investigation into the Determination of Trimethylarsine in Natural Gas and its Partitioning into
Gas and Condensate Phases Using (Cryotrapping)/Gas Chromatography Coupled to Inductively
Coupled Plasma Mass Spectrometry and Liquid/Solid Sorption Techniques. Spectrochim. Acta,
Part B 2007, 62, 970–977.
(10) Becker, E.; Rampazzo, R. T.; Dessuy, M. B.; Vale, M. G. R.; da Silva, M. M.; Welz, B.;
Katskov, D. A. Direct Determination of Arsenic in Petroleum Derivatives by Graphite Furnace
Atomic Absorption Spectrometry: A Comparison Between Filter and Platform Atomizers.
Spectrochim. Acta, Part B 2011, 66, 345–351.
(11) Caumette, G.; Lienemann, C.-P.; Merdrignac, I.; Bouyssiere, B.; Lobinski, R. Element
Speciation Analysis of Petroleum and Related Materials. J. Anal. At. Spectrom. 2009, 24, 263–
276.
(12) Fish, R. H.; Brinckman, F. E.; Jewett, K. L. Fingerprinting Inorganic Arsenic and
Organoarsenic Compounds in In Situ Oil Shale Retort and Process Waters Using a Liquid
Chromatograph Coupled with an Atomic Absorption Spectrometer as a Detector. Environ. Sci.
Technol. 1982, 16, 174–179.
(13) Bouyssiere, B.; Baco, F.; Savary, L.; Garraud, H.; Gallup, D. L.; Lobinski, R.
Investigation of Speciation of Arsenic in Gas Condensates by Capillary Gas Chromatography
with ICP-MS Detection. J. Anal. At. Spectrom. 2001, 16, 1329–1332.
(14) Xu, R.; Tang, D. Z.; Yan, Q. T.; Xu, H.; Wang, S. Y.; Tao, S.; Han, Z. X. Exploration of
Detection Technology About Arsenic Content in Natural Gas and Application. Energy Fuels
2015, 29, 3863–3869.
(15) Olsen, S. D.; Westerlund, S.; Visser, R. G. Analysis of Metals in Condensates and
Naphtha by Inductively Coupled Plasma Mass Spectrometry. Analyst 1997, 122, 1229–1234.
(16) Stigter, J. B.; de Haan, H. P. M.; Guicherit, R.; Dekkers, C. P. A.; Daane, M. L.
Determination of Cadmium, Zinc, Copper, Chromium and Arsenic in Crude Oil Cargoes.
Environ. Pollut. 2000, 107, 451–464.
(17) Pinel-Raffaitin, P.; Le Hecho, I.; Amouroux, D.; Potin-Gautier, M. Distribution and Fate
of Inorganic and Organic Arsenic Species in Landfill Leachates and Biogases. Environ. Sci.
Technol. 2007, 41, 4536–4541.
(18) Xiu, S.; Shahbazi, A.; Shirley, V. B.; Wang, L. Swine Manure/Crude Glycerol Co-
Liquefaction: Physical Properties and Chemical Analysis of Bio-Oil Product. Bioresour.
Technol. 2011, 102, 1928–1932.
(19) Cachia, M.; Bouyssière, B.; Carrier, H.; Garraud, H.; Caumette, G.; Le Hécho, I.
Development of a High-Pressure Bubbling Sampler for Trace Element Quantification in
Natural Gas. Energy Fuels 2017, 31, 4294–4300.
(20) Haas, K.; Feldmann, J. Sampling of Trace Volatile Metal(loid) Compounds in Ambient
Air Using Polymer Bags: A Convenient Method. Anal. Chem. 2000, 72, 4205–4211.
(21) Uroic, M. K.; Krupp, E. M.; Johnson, C.; Feldmann, J. Chemotrapping-Atomic
Fluorescence Spectrometric Method as a Field Method for Volatile Arsenic in Natural Gas. J.
Environ. Monit. 2009, 11, 2222–2230.
72
(22) Hirner, A. V.; Feldmann, J.; Krupp, E.; Grümping, R.; Goguel, R.; Cullen, W. R.
Metal(loid)organic Compounds in Geothermal Gases and Waters. Org. Geochem. 1998, 29,
1765–1778.
(23) Freije-Carrelo, L.; Moldovan, M.; García Alonso, J. I.; Vo, T. D. T.; Encinar, J. R.
Instrumental Setup for Simultaneous Total and Speciation Analysis of Volatile Arsenic
Compounds in Gas and Liquefied Gas Samples. Anal. Chem. 2017, 89, 5719–5724.
(24) Feldmann, J.; Hirner, A. V. Occurrence of Volatile Metal and Metalloid Species in
Landfill and Sewage Gases. Int. J. Environ. Anal. Chem. 1995, 60, 339–359.
(25) Hirner, A. V.; Feldmann, J.; Goguel, R.; Rapsomanikis, S.; Fischer, R.; Andreae, M. O.
Volatile Metal and Metalloid Species in Gases from Municipal Waste Deposits. Appl.
Organomet. Chem. 1994, 8, 65–69.
(26) Brandão, G. P.; de Campos, R. C.; Luna, A. S. Determination of Mercury in Gasoline by
Cold Vapor Atomic Absorption Spectrometry with Direct Reduction in Microemulsion Media.
Spectrochim. Acta, Part B 2005, 60, 625–631.
(27) Aucelio, R. Q.; Curtius, A. J. Evaluation of Electrothermal Atomic Absorption
Spectrometry for Trace Determination of Sb, As and Se in Gasoline and Kerosene using
Microemulsion Sample Introduction and Two Approaches for Chemical Modification. J. Anal.
At. Spectrom. 2002, 17, 242–247.
(28) Cassella, R. J.; Barbosa, B. A. R. S.; Santelli, R. E.; Rangel, A. T. Direct Determination
of Arsenic and Antimony in Naphtha by Electrothermal Atomic Absorption Spectrometry with
Microemulsion Sample Introduction and Iridium Permanent Modifier. Anal. Bioanal. Chem.
2004, 379, 66–71.
(29) Kumar, S. J.; Gangadharan, S. Determination of Trace Elements in Naphtha by
Inductively Coupled Plasma Mass Spectrometry using Water-in-Oil Emulsions. J. Anal. At.
Spectrom. 1999, 14, 967–971.
(30) Aneva, Z.; Iancheva, M. Simultaneous Extraction and Determination of Traces of Lead
and Arsenic in Petrol by Electrothermal Atomic Absorption Spectrometry. Analytica Chimica
Acta 1985, 167, 371–374.
(31) US EPA Method 3031. Acid Digestion of Oils for Metals Analysis by Atomic Absorption
or inductively Coupled Plasma (ICP) Spectrometry; EPA SW-846 Test Method 3031, 1996.
(32) Dubey, R. K.; Puri, B. K.; Hussain, M. F. Determination of Arsenic in Various
Environmental and Oil Samples by Differential Pulse Polarography After Adsorption of its
Morpholine-4-Carbodithioate on to Microcrystalline Naphthalene or Morpholine-4-Dithio-
Carbamate-CTMAB-Naphthalene Adsorbent. Anal. Lett. 1997, 30, 163–172.
(33) Narasaki, H. Determination of Arsenic and Selenium in Fat Materials and Petroleum
Products by Oxygen Bomb Combustion and Automated Atomic Absorption Spectrometry with
Hydride Generation. Anal. Chem. 1985, 57, 2481–2486.
(34) ANSI Webstore. ASTM UOP946-96. Arsenic in Petroleum Naphthas by HG-AAS. n.d.
https://webstore.ansi.org/standards/astm/astmuop94696?gclid=CjwKCAjwxOCRBhA8EiwA0
X8hiwIWrPB2v9pkspCcezK2qfIAEL0bsxF2YXU06sQsN5gvkdraTEGCCBoCycoQAvD_B
wE (accessed 2022–3–21).
73
(35) Hofstader, R. A.; Milner, O. I.; Runnels, J. H. Arsenic. In Analysis of Petroleum for
Trace Metals, Advances in Chemistry, Hofstader, R. A., Milner, O. I., Runnels, J. H. Eds.;
American Chemical Society: Washington, D. C., 1976; pp 55–67.
(36) de Albuquerque, F. I.; Duyck, C. B.; Fonseca, T. C. O.; Saint'Pierre, T. D. Determination
of As and Se in Crude Oil Diluted in Xylene by Inductively Coupled Plasma Mass Spectrometry
Using a Dynamic Reaction Cell for Interference Correction on 80Se. Spectrochim. Acta, Part
B 2012, 71-72, 112–116.
(37) Reboucas, M. V.; Ferreira, S. L. C.; Neto, B. D. B. Arsenic Determination in Naphtha
by Electrothermal Atomic Absorption Spectrometry After Preconcentration Using Multiple
Injections. J. Anal. At. Spectrom. 2003, 18, 1267–1273.
(38) de Jesus, A.; Zmozinski, A. V.; Damin, I. C. F.; Silva, M. M.; Vale, M. G. R.
Determination of Arsenic and Cadmium in Crude Oil by Direct Sampling Graphite Furnace
Atomic Absorption Spectrometry. Spectrochim. Acta, Part B 2012, 71-72, 86–91.
(39) Reboucas, M.; Ferreira, S.; Debarrosneto, B. Behaviour of Chemical Modifiers in the
Determination of Arsenic by Electrothermal Atomic Absorption Spectrometry in Petroleum
Products. Talanta 2005, 67, 195–204.
(40) Cassella, R. J.; de Sant'Ana, O. D.; Santelli, R. E. Determination of Arsenic in Petroleum
Refinery Streams by Electrothermal Atomic Absorption Spectrometry After Multivariate
Optimization Based on Doehlert Design. Spectrochim. Acta, Part B 2002, 57, 1967–1978.
(41) Arnold, S.; Rodriguez-Uribe, A.; Misra, M.; Mohanty, A. K. Slow Pyrolysis of Bio-Oil
and Studies on Chemical and Physical Properties of the Resulting New Bio-Carbon. J. Cleaner
Prod. 2018, 172, 2748–2758.
(42) He, J.; Strezov, V.; Zhou, X.; Kumar, R.; Kan, T. Pyrolysis of Heavy Metal
Contaminated Biomass Pre-Treated with Ferric Salts: Product Characterisation and Heavy
Metal Deportment. Bioresour. Technol. 2020, 313, 123641.
(43) He, J.; Strezov, V.; Kan, T.; Weldekidan, H.; Asumadu-Sarkodie, S.; Kumar, R. Effect
of Temperature on Heavy Metal(loid) Deportment During Pyrolysis of Avicennia marina
Biomass Obtained from Phytoremediation. Bioresour. Technol. 2019, 278, 214–222.
(44) Zhang, Z.; Macquarrie, D. J.; De Bruyn, M.; Budarin, V. L.; Hunt, A. J.; Gronnow, M.
J.; Fan, J.; Shuttleworth, P. S.; Clark, J. H.; Matharu, A. S. Low-Temperature Microwave-
Assisted Pyrolysis of Waste Office Paper and the Application of Bio-oil as an Al Adhesive.
Green Chem. 2015, 17, 260–270.
(45) Chiang, H.-L.; Lin, K.-H.; Lai, N.; Shieh, Z.-X. Element and PAH Constituents in the
Residues and Liquid Oil from Biosludge Pyrolysis in an Electrical Thermal Furnace. Sci. Total
Environ. 2014, 481, 533–541.
(46) Kim, J.-Y.; Kim, T.-S.; Eom, I.-Y.; Kang, S. M.; Cho, T.-S.; Choi, I. G.; Choi, J. W.
Characterization of Pyrolytic Products Obtained from Fast Pyrolysis of Chromated Copper
Arsenate (CCA)- and Alkaline Copper Quaternary Compounds (ACQ)-Treated Wood
Biomasses. J. Hazard. Mater. 2012, 227-228, 445–452.
(47) Cachia, M.; Bouyssiere, B.; Carrier, H.; Garraud, H.; Caumette, G.; Le Hécho, I.
Characterization and Comparison of Trace Metal Compositions in Natural Gas, Biogas, and
Biomethane. Energy Fuels 2018, 32, 6397–6400.
74
(48) Chin, K. F.; Wan, C.; Li, Y.; Alaimo, C. P.; Green, P. G.; Young, T. M.; Kleeman, M. J.
Statistical Analysis of Trace Contaminants Measured in Biogas. Sci. Total Environ. 2020, 729,
138702.
(49) Mestrot, A.; Uroic, M. K.; Plantevin, T.; Islam, M. R.; Krupp, E. M.; Feldmann, J.;
Meharg, A. A. Quantitative and Qualitative Trapping of Arsines Deployed to Assess Loss of
Volatile Arsenic from Paddy Soil. Environ. Sci. Technol. 2009, 43, 8270–8275.
(50) Mestrot, A.; Xie, W.-Y.; Xue, X.; Zhu, Y.-G. Arsenic Volatilization in Model Anaerobic
Biogas Digesters. Appl. Geochem. 2013, 33, 294–297.
75
CHAPITRE II : FIABILITE DE
L’ECHANTILLONNAGE ET DE
L’ANALYSE DU MERCURE
DANS LE GAZ NATUREL
Il existe différentes méthodologies pour la détermination de la concentration de mercure

dans le gaz naturel (cf 3.1.Sampling method, p 27). L’une d'entre elles, repose sur
l'échantillonnage du gaz naturel à pression atmosphérique à l’aide de sacs Tedlar ainsi qu’à
haute pression à l'aide de cylindres en acier inoxydable avant de mesurer le Hg hors site (cf
3.1.4.Gas Samplers: Cylinders and Tedlar Bags, p 32).
Un revêtement interne est présumé empêcher l'adsorption du Hg à l’intérieur de ces
cylindres ainsi que des sacs Tedlar, rendant vraisemblablement l'analyse du Hg fiable. Grâce à
différentes expérimentations développées dans la partie 1 de ce chapitre, cette notion de fiabilité
d’échantillonnage de gaz naturel pour l’analyse du mercure dans des cylindres est remise en
question, en montrant que même les cylindres revêtus de silicium sont inefficaces pour
empêcher l'adsorption du Hg sur les parois internes. Ces expérimentations décrites dans la partie
1, ont été réalisées avec un gaz propre et non sur un échantillon réel de gaz naturel contenant
des concentrations substantielles de soufre par rapport au Hg.
Afin de déterminer si le soufre ainsi que l’oxygène ont un rôle dans le mécanisme
d’adsorption du mercure sur les échantillonneurs, un sac Tedlar a été rempli avec ¼ d’air
ambiant et de ¾ de gaz pétrolier auxquels du mercure gazeux y a été ajouté. La stabilité de
celui-ci à l’intérieur du sac Tedlar a été suivi au cours du temps. Les résultats sont décrits dans
la partie 2 de ce chapitre.
Les résultats de stabilité du mercure dans ces échantillonneurs au cours du temps, décrits
dans les parties 1 et 2 de ce chapitre, ont permis, par la suite, d’élaborer des recommandations
sur l’échantillonnage et l’analyse du mercure dans le gaz naturel au cours d’un essai de puits.
Ces recommandations sont décrites dans la partie 3 de ce chapitre.
76
Partie 1 : Experimental Tests of Natural Gas
Samplers Prior to Mercury Concentration Analysis
Energy Fuels 2020, 34, 5, 5205 – 5212 - https://doi.org/10.1021/acs.energyfuels.9b03540
Received: October 14, 2019; Revised: December 26, 2019; Publication Date: December 26,
2019
Maxime Enrico1,2, Aurore Mere1,2,3, Honggang Zhou3, Hervé Carrier2,4, Emmanuel

Tessier1,2, Brice Bouyssiere1,2*
1
Institut des Sciences Analytiques et de Physico-Chimie Pour L'environnement et les Materiaux, CNRS/ UNIV
2
3
TOTAL E&P, CSTJF, Av. Larribau, 64018 Pau, France
4
CNRS/TOTAL/ UNIV PAU & PAYS ADOUR / E2S UPPA, LFCR-IPRA UMR 5150, 64000, Pau, France
Abstract
Mercury (Hg) is a natural, trace component of natural gas. Corrosion of aluminum heat
exchangers by liquid metallic Hg can lead to dramatic issues. The quantification of the gaseous
Hg concentration in natural gas streams is therefore crucial prior to the implementation of Hg
removal units for preventing the formation of liquid Hg. Different methodologies exist for the
determination of the Hg concentration in natural gas, one of which relies on the sampling of
natural gas at high pressure using stainless-steel cylinders prior to off-site Hg measurement. An
inert internal coating is supposed to hamper Hg adsorption, presumably making the Hg analysis
reliable. Here, we challenge this statement by showing that even silicon-coated cylinders are
inefficient for preventing Hg adsorption on internal walls. Different cylinders were tested for
gaseous Hg concentration stability over time in a clean argon matrix. We find that the gaseous
Hg concentration sharply declines in almost all tested cylinders (uncoated,
polytetrafluoroethylene-coated, and silicon-coated) to reach undetectable levels within a day or
two as a result of adsorption, with the notable exception of a brand-new silicon-coated cylinder.
Heating cylinders up to 190°C allowed for the recovery of most adsorbed Hg and revealed the
77
occurrence of two distinct Hg species with distinct release temperatures. Our results suggest
that Hg0 is first physically adsorbed and further oxidized, presumably in relation to sulfur
compounds covering the internal walls of the cylinders. The newly purchased silicon-coated
cylinder kept a constant gaseous Hg concentration over 6 months because it never interacted
with any real natural gas sample containing substantial sulfur concentrations relative to Hg.
1. Introduction
The occurrence of mercury (Hg) in natural gas is of great concern for the petroleum
industry. Hg occurs naturally in petroleum and natural gas, with reported concentrations
covering several orders of magnitude (from ng Nm−3 to mg Nm−3 ranges in natural gas).1 Hg in
petroleum reservoirs undergoes reduction and is therefore mostly present as Hg0,2 a highly
volatile species. In hydrocarbon extraction and processing, Hg therefore likely goes to the light
natural gas phase, although variations toward lower temperatures might allow for some
oxidation.3
In addition to health and environmental problematics, the petroleum industry adds
another issue concerning Hg. Specifically, the large drop in temperature during liquefied natural
gas (LNG) production tends to lower the gaseous Hg saturation value, potentially leading to the
formation of liquid metallic Hg.4 The action of metallic mercury as a corrosion catalyst for
aluminum heat exchangers is well-illustrated by two accidents that occurred in 1973 and 2004
in Skikda (Algeria). The corrosion of aluminum heat exchangers as a result of liquid Hg
formation led to a dramatic failure, causing numerous deaths. Hg-induced corrosion can occur
via two pathways: amalgam corrosion and liquid-metal embrittlement.5,6 Both pathways require
a breach in the surface aluminum oxide layer to allow for direct contact between metallic
aluminum and liquid Hg.
The implementation of Hg removal units upstream of sensitive operations avoids such
failure. These Hg removal units are, however, efficient up to a maximum amount of Hg treated,
after which they have to be either regenerated or replaced. The knowledge of the Hg
concentration in the natural gas treated is therefore a prerequisite for establishing appropriate
operation procedures. Different methodologies exist for the quantification of the Hg
concentration in natural gas. The sampling step can be performed by preconcentrating Hg on
solid sorbents or gold traps (amalgamation)7 or bubbling in oxidizing solutions.8 All of these
78
sampling techniques have disadvantages related to sampling conditions (flow rate and
pressure)8,9 that can lead to long sampling periods depending upon the Hg concentration.
As an alternative, natural gas sampling at a high pressure can be performed using
cylinders. This efficiently reduces the sampling time, and large amounts of gas can be stored.
Although a subsequent amalgamation step is required prior to Hg quantification, this step is
more conveniently performed off-site in a laboratory. Hg adsorption issues on metallic surfaces
are, however, well-known, including potential amalgamation with other metals,10 precipitation
of Hg sulfides,11,12 or simply physical adsorption. The efficiency of cylinders as natural gas
samplers prior to Hg quantification is therefore doubtful, with the possibility of adsorption
issues. Cylinders internally coated with inert materials [silicon and polytetrafluoroethylene
(PTFE)] were developed to overcome adsorption issues for sulfur and Hg. Stability tests from
the manufacturer13 as well as another independent test11 showed the efficiency of a silicon
coating in preventing Hg losses from the gas phase.
Figure 5. Schemes of the systems used for (A) loading of Hg-contaminated gas in the cylinders and
(B) thermal desorption experiments.
79
To understand some positive and negative results that were highlighted in the laboratory
during recent Hg sampling and measurement campaigns, we decided to perform our own
experiments on different cylinders to confirm their efficiency prior to their use on the field.
Moreover, none of the latter two studies investigated the efficiency of a silicon-coated cylinder
that was previously used on the field (and might already contain adsorbed compounds or have
been modified by the sampling or the transportation). The influence of a hydrocarbon gas with
a complex matrix therefore remains poorly understood. Our tests thus focus on the comparison
of new and used silicon-coated cylinders. We also included a PTFE-coated cylinder, an
uncoated cylinder, and a Tedlar bag to obtain a full overview.
2. Experimental Section
2.1. Samplers
We purchased a new silicon-coated cylinder from Restek (Lisses, France). This cylinder
had therefore never interacted with any hydrocarbon gas, and we could evaluate the initial
inertness of the coating material.
Two other silicon-coated cylinders (hereafter referred to as silicon old A and B)
purchased from Restek (Lisses, France) were used for our tests. These cylinders were both used
approximately 10 times for sampling processed commercial natural gas. The major sulfur
compound that was present in the sampled gas is tetrahydrothiophene (THT), an additive for
detecting natural gas leaks (odorant).
We obtained one PTFE-coated and one uncoated cylinder from the Total Exploration
Production Research Center (CSTJF, Pau, France). Both were used for natural gas sampling at
production sites in the past, but the precise sampling history is unknown. However, the presence
of gas components, such as H2S or thiophenic compounds, is highly probable.
Another sampling option is the use of Tedlar bags (Lisses, France). These bags made of
a fluoropolymer are supposed to be inert, and this material is commonly used in Hg research.
Our tests therefore include experiments with a new Tedlar bag for comparison to cylinders.
80
2.2. Generation of Hg-Contaminated Gas
For loading, a cylinder was first connected to the previously developed system presented
in Figure 5A.8 Argon (5.0 from Linde) was delivered to the system at 70 bar, passing first
through a permeation cell containing liquid Hg and then through the cylinder. Argon is
contaminated in Hg when passing through the permeation cell. The Hg concentration depends
upon the argon flow rate and system pressure. To generate reproducible Hg concentrations in
cylinders, we adopted a procedure that used a 30 min flushing period with Hg-contaminated
argon at 500 mL.min−1 prior to cylinder closure. All experiments, except when stated otherwise,
were conducted using this procedure at a pressure of 70 bar.
A similar method was used for loading the Tedlar bag, except that the flow meter was
installed upstream of the bag (no outlet on Tedlar bags). The Tedlar bag was connected after
the 30 min flushing procedure and filled with 15 L of argon. The system with the permeation
cell was further kept at 70 bar to ensure comparability of the cylinder results.
Figure 6. Initial Hg concentration measured after sampler loading with Hg-contaminated argon.
Are shown only the initial Hg concentration for experiments including a 30 min preliminary
flushing step during the loading experiment. The results for 6 different samplers are shown, as well
as for the different tests conducted on each sampler.
81
For blank evaluation, argon was delivered directly to a gold trap using a bypass (Figure
5A). The system blank was relatively high (1 ng.L−1), representing the repeated uses of this
system for experimental work on Hg over the past decade.
2.3. Subsampling for the Hg Concentration Measurement
The gaseous Hg concentration was determined at several time steps for each test,
including the evaluation of the initial Hg concentration. One valve of the cylinder was opened,
and the stream out was adjusted to 150−200 mL.min−1. A gold trap was then connected to
sample the desired gas volume. The gas flow rate and sampled volume were controlled and
measured by a gas counter (Mass-View MV-102, Bronkhorst).
Because the reliability of a single measurement is doubtful, every sampling used four
distinct gold traps. The four subsampling replicates were performed successively by replacing
gold traps. A volumetric gradient approach was adopted for reliable Hg quantification (see
Figure 10 of the Supporting Information). The slope between the volume sampled and the
amount of Hg recovered was used as the Hg concentration.
2.4. Thermal Desorption Experiments
Following the gaseous Hg stability experiments, thermal desorption was attempted on

several cylinders. The cylinder was placed in a programmable oven under an argon flow of
approximately 200 L.min−1. Desorbed Hg was sampled using a gold trap installed downstream
of the cylinder, outside the oven (Figure 5B).
The initial experimental temperature was 70°C. In addition to the control panel, the
temperature was monitored using two independent sensors. The gold trap was left for 10 min
(approximately 2 L of gas sampled) before being replaced. This was repeated 3 times to provide
three different estimates for desorption rates at each temperature. The oven temperature was
then increased by 10°C, and the procedure was repeated until 190°C. Because we used PTFE
tape on screws to prevent gas leaks, we did not perform experiments at higher temperatures
(PTFE melts at 205°C).
82
2.5. Hg Analysis
We used an automated double amalgamation system with atomic fluorescence

spectrometry (Sir Galahad, PS Analytical, U.K.) for Hg analysis. Briefly, Hg from the gold
traps was thermally desorbed by heating at 500°C for 50 s and carried to an analytical gold trap
by an argon flow. The analytical gold trap is then heated at 500°C for 15 s, and an argon flow
carries the released gaseous Hg to the atomic fluorescence detector. The instrumental detection
limit is 10 pg. Considering that our system blank was 1 ng.L−1, we had no issues related to the
instrumental detection limit.
3. Results and Discussions
3.1. Initial Gaseous Hg Concentration and Temporal Stability
The loading procedure led to different initial Hg concentrations depending upon the
cylinder that we tested. Three successive tests with the Tedlar bag gave consistent values of
approximately 71 ± 3 ng.L−1 (n = 3; Figure 6), which compares well to a test performed on the
new silicon-coated cylinder (76 ± 6 ng.L−1). Two other tests were conducted on the silicon-
coated cylinder, but we adopted a different peanut butter-based loading strategy. For these two
tests (2 and 3), the 30 min flushing step was avoided, which led to initial gaseous Hg
concentrations of 29 ± 1 and 33 ± 1 ng.L−1, respectively. This lower concentration is related to
the variable flow rates through the permeation cell during the loading, leading to different
gas/liquid fractionation values.
The old silicon-coated cylinder A was successively tested 5 times with comparable
initial Hg concentrations (average value of 58 ± 4 ng.L−1; n = 5). This is slightly lower than the
new silicon-coated cylinder and Tedlar bag. We acknowledge that small differences in the
loading procedure (exact flushing flow rate in particular) might be responsible for small
differences in the initial gaseous Hg concentration. The old silicon-coated cylinder B, however,
gave a low initial Hg concentration of 29 ± 1 ng.L−1. This low Hg concentration relative to the
new silicon-coated cylinder is unlikely to be derived from the procedure itself and likely reveals
adsorption already occurred during our loading procedure.
83
The three experiments conducted on the PTFE-coated cylinder gave three different
initial Hg concentrations, with an increasing trend from 6.5 ± 0.3 to 17.3 ± 1.6 ng.L−1. In
comparison to the results obtained for the silicon-coated cylinders, these values are rather low
and likely indicate that adsorption occurred. The results for the uncoated cylinder were even
worse, with initial Hg concentrations evaluated to be an order of magnitude below the system
blank (approximately 0.07 ng.L−1, average of three tests). This clearly shows that uncoated
cylinders efficiently adsorb Hg. In fact, even a test performed during the flushing procedure by
sampling at the outlet of the uncoated cylinder reflected a negligible gaseous Hg concentration.
Further stability tests were not conducted on uncoated cylinders, because all Hg seemed to be
initially rapidly adsorbed. All other cylinders were tested for Hg concentration stability over
time.
The newly purchased silicon-coated cylinder efficiently maintained a constant Hg
concentration during short-term experiments (a few hours; see Figure 11 of the Supporting
Information). A longer-term experiment showed that the gaseous Hg concentration is stable
over at least 6 months, supporting previous tests performed by the manufacturer (Figure 7A).
The old silicon-coated cylinders, however, behaved differently, with a clear decrease in the
gaseous Hg concentration, visible after only a few hours (panels B and C of Figure 7). Despite
the lower initial Hg concentration in the old silicon-coated cylinder B (Figure 6), the decline
seems to be slower than that for cylinder A (panels B and C of Figure 7). The PTFE-coated
cylinder showed a fast decline in the gaseous Hg concentration, reaching undetectable levels
after only 3 h. Despite the differences in the initial Hg concentration among the three tests, we
observe a surprisingly identical declining trend.
84
Figure 7. Gaseous Hg concentration variations over time during stability experiments conducted on
(A) a new silicon-coated cylinder, (B) the old silicon-coated cylinders A and (C) B, (D) the PTFE-
coated cylinder and (E) the Tedlar bag. Note the differences in the x-axis scale, especially for panel
A (days unit compared to other panels in hours). Only the long-term test is shown for the new
silicon-coated cylinder (panel A), and the results of short-term tests (1 and 2) can be found in
supporting information Figure S2.
3.2. Adsorption of Gaseous Hg and Equilibrium
Only the new silicon-coated cylinder and the Tedlar bag were able to maintain a constant
gaseous Hg concentration (panels A and E of Figure 7). From all of these observations, we
85
conclude that gaseous Hg is progressively adsorbed on the internal cylinder walls, at least in
the old cylinders (silicon- and PTFE-coated). The five successive tests on the old silicon-coated
cylinder A revealed different rates of decrease (Figure 7B), with a faster decline for the first
tests (no Hg detected after 20 h) than the last (20% of Hg remaining after 60 h). Following these
experiments, an empty cylinder was filled with uncontaminated argon (bypass shown in Figure
1) at 70 bar without flushing the system.
Figure 8. Evolution of gaseous Hg concentration in the old silicon-coated cylinder A after loading
with uncontaminated argon. This test was conducted after the five tests with contaminated argon
(Figure 7B). The initial gaseous Hg concentration of 1 ng.L-1 is consistent with the system blank
value.
The gaseous Hg concentration was then monitored for 5 h. It increased from an initial Hg
concentration of 1 ng.L−1 (similar to the system blank) to stabilize at approximately 2.9 ng.L−1
(Figure 8). This demonstrates that Hg adsorption is ruled at least partly by an equilibrium
relationship with gaseous Hg.
All of our stability tests consisted of measurements performed on very small gas samples
compared to the amount of gas stored in the cylinder, which is not informative regarding the
behavior of Hg during cylinder purge. The old silicon-coated cylinder B was left filled with
Hg-contaminated gas for 15 more days. The gas was then entirely flushed through gold traps at
150−200 mL.min−1. Each gold trap was sampled and contained 1−3 L of gas. We determined
86
that the gaseous Hg concentration at the beginning of the procedure was approximately 0.5
ng.L−1. It progressively decreased with pressure during the cylinder purge (Figure 9A). This
finding shows that pressure influences the equilibrium relationship between adsorbed Hg and
gaseous Hg.
3.3. Thermal Desorption Results
We tested the possibility of recovering adsorbed Hg by heating the cylinder. The

cylinder was flushed with argon and heated from 70 to 190°C. A gold trap recovered desorbed
Hg for 10 min before being replaced. We found that desorption rates generally increased with
the temperature (Figure 9B). However, we also noted a double-peak shape, indicating two
optimal desorption temperatures. The desorption rates first increased when the temperature was
increased from 70 to 130°C. Desorption rates then stabilized or decreased, followed by a further
increase.
The double-peak shape that we obtained during the thermal desorption experiment is
likely indicative of the occurrence of two distinct forms of adsorbed Hg species, because the
volatilization temperature indicates Hg species.14,17,18 Adsorbed Hg0 is classically released at
low temperatures as a result of its volatility (from 70°C, peak around 80−150°C depending
upon the procedure).17 This is consistent with the first peak that we observe, with desorption
starting around 80°C and increasing until 130°C. Other Hg forms, consisting of oxidized Hg
(either as particulate, Hg salts, or Hg complexes), are released at higher temperatures that vary
depending upon the species.
This is well-illustrated by a study showing that the gas chromatography (GC) injector
temperature is critical for the evaluation of Hg0 levels in liquid hydrocarbon samples. Thermal
decomposition of Hg particles into Hg0 might lead to overestimation of the Hg0 concentration.19
Our procedure does not allow for the identification of this species, but the double peak shape
unambiguously suggests the presence of both physically adsorbed Hg0 and an oxidized Hg form
into the cylinder. The decomposition of Hg compounds to gaseous Hg0 is also the principle of
a Hg quantification method in liquid hydrocarbon samples consisting of heating at 200°C and
measuring headspace gaseous Hg, in equilibrium with dissolved Hg0.20 Similarly, a Hg removal
technique suggests heating liquid hydrocarbons at temperatures higher than 175°C to transform
all Hg species to Hg0, easier to remove from hydrocarbon streams.3
87
During our experiments, only Hg0 and argon were delivered to the cylinder. Therefore,
Hg oxidation obviously occurred in the cylinder. Sulfur species, especially H2S, are known to
indirectly cause depletion in gaseous Hg12,15 by first reacting with metallic oxide surfaces. The
sampling history of this cylinder (old silicon-coated cylinder B) indicated that its use was
limited to the sampling of a natural gas containing very low amounts of Hg (on the ng.Nm−3
level, measured independently by the amalgamation method). Sulfur species should have been
present at very low levels as well, with the notable exception of THT, which is injected
upstream for leak detection. As H2S, THT contains reduced sulfur that might interact with Hg.
The most likely explanation for our observations involves the adsorption of THT or other
reduced sulfur compounds on internal cylinder walls, providing binding sites for Hg0.
Figure 9. Hg recovery during (A) emptying and (B) heating the old silicon-coated cylinder B. Panel
A shows the variations in gaseous Hg concentration during the emptying of the silicon-coated
cylinder B, shown as a gaseous Hg concentration as a function of the pressure left in the cylinder.
Panel B illustrates the temperature dependence of Hg desorption rates.
88
The oxidation of Hg from physically adsorbed Hg0, followed by complexation with reduced
sulfur, was already observed in solution.16 We propose that a similar sorption and oxidation
pathway also affects gaseous Hg.
We note that silicon coating is also expected to be inert to sulfur compounds. This is
probably true to some extent, but the inertness is not perfect, as shown by tests from the
manufacturer itself (from Restek Technical Resources). The main problem here is that typical
sulfur and Hg concentrations differ by several orders of magnitude. In other words, although
less than 1% sulfur adsorption is not problematic for sulfur measurement and speciation, it can
affect Hg adsorption and lead to the complete depletion of Hg from the gas phase. The thermal
desorption procedure was also applied to the uncoated cylinder and the PTFE-coated cylinder
(see Figure 12 of the Supporting Information). In contrast to the silicon-coated cylinder B,
however, the procedure was applied several months after the cylinder purge. Both revealed a
very high content of adsorbed Hg, although exact quantification was not attempted. The results
for the PTFE-coated cylinder also suggest two Hg species, although oxidized Hg seems to be
very high compared to physically adsorbed Hg. This might derive from faster oxidation, which
is supported by the fast depletion of the gaseous Hg concentration during our experiments (all
Hg lost within 3 h). It could also be related to the lag between the gaseous Hg stability and
thermal desorption experiments (2 months).
3.4. Mass Balance and Reaction Rates
The initial gaseous Hg concentration in the old silicon-coated cylinder B suggested that
600 ng of gaseous Hg was present inside the cylinder at the initial stage of the experiment. As
already mentioned, this is quite different from the new silicon-coated cylinder, for which the
same loading procedure led to an initial gaseous Hg amount of about 1500 ng. We already
discussed this difference and attributed it to fast adsorption occurring during the loading
procedure. The initial gaseous Hg amount in the silicon-coated cylinder B does not account for
Hg adsorbed during the loading and, therefore, underestimates the initial Hg budget.
During the purge of silicon-coated cylinder B, we recovered a total of 5 ng of Hg (Figure
9A), which is somewhat negligible compared to the initial Hg budget. The thermal desorption
procedure allowed us to recover 1100 ng of Hg (Figure 9B), almost twice the initial Hg budget.
The 500 ng excess is unlikely to originate from the natural gas previously sampled using this
89
cylinder, because the Hg concentration was 4 orders of magnitude lower than that of the
contaminated gas that we generated. It more likely derives from Hg that was adsorbed during
the loading of the cylinder, prior to the evaluation of the initial gaseous Hg concentration. The
total amount of Hg that we retrieve during the purge and thermal desorption is comparable to
the 1500 ng that is present in the new silicon-coated cylinder when using the same loading
procedure. The slight difference might be caused by minor differences during the loading
procedure or the formation of another Hg species that is released at a higher temperature.
Considering that we have two Hg species in the cylinder, fast adsorption might be related to the
physical adsorption of Hg0. Adsorption rates during the initial stage of the experiments
(including during the loading procedure) are therefore controlled by sorption sites for Hg0. The
reaction scheme with subsequent oxidation, together with the slower adsorption rates observed
during our experiments (Figure 7C), suggests that the oxidation step is the limiting factor when
equilibrium is reached between physically adsorbed and gaseous Hg0.
The reaction scheme that we inferred excludes the direct oxidation of gaseous Hg0,
because we speculated that Hg0 first requires physical adsorption. The observation of Larsson
et al.12 is, however, in good agreement with this hypothesis, because no Hg loss was observed
at 75 and 100°C from a gas containing Hg0 and H2S. At such low temperatures, only the
physical adsorption of Hg0 should be inhibited; thus, direct oxidation of gaseous Hg0 would
induce losses of Hg even at 75−100°C. Additional indirect evidence is derived from a previous
test of a silicon-coated cylinder. The authors showed no evolution of the gaseous Hg
concentration in their cylinder upon heating at 100°C throughout the experiment,11 although
this study did not show any results for a similar cylinder kept at ambient temperature.
4. Conclusion and Recommendations
We provided experimental evidence showing the inefficiency of silicon coatings for

natural gas sampling prior to Hg analysis (as well as PTFE coatings). The inertness of the
coating regarding Hg adsorption is initially efficient but does not last in time. We suggest that
the adsorption of sulfur species on the coated surfaces provides sorption sites for Hg and can
ultimately lead to the transformation of Hg0 into less volatile oxidized forms. Tedlar bags might
be inert to Hg. However, our tests used only a new Tedlar bag, so that the fluoropolymer
surfaces never experienced any contact with sulfur species. Further experiments are required
90
for evaluating Tedlar bags as a sampling alternative prior to Hg analysis in natural gas, but the
fragility and size of this sampler system are also negative points that might be considered when
the sample has to travel.
We acknowledge that appropriate sampling procedures using cylinders might be
imagined (sampling at 100°C to avoid adsorption), but adsorption needs to be considered even
in the case of silicon-coated cylinders. The repeated use of cylinders for sampling different
natural gas samples might lead to both adsorption and desorption, because the partition is ruled
by an equilibrium relation and depends upon the total Hg budget and pressure. Alternative
standard methodologies involving Hg preconcentration by amalgamation or oxidation in acidic
solutions (KMnO4/H2SO4)8 might be preferable, despite flow rate limitations.9
5. Complete information
5.1. Determination of Hg concentration
Every gaseous Hg concentration determination relied on 4 analyses during the stability

tests, by adopting a volumetric gradient approach. Depending on the Hg concentration
expected, sampled volumes were adapted in order to preconcentrated sufficient Hg to insure a
precise analysis.
Figure 10 illustrates the volumetric gradient approach. The slope between sampled
volumes and Hg sampled gives a good estimate of Hg concentration. On Figure 10 a negative
intercept is observed. This is caused by a lag time between system connection and count volume
count start.
The uncertainty on the determined concentration is given by the standard error on the
slope. Uncertainties on yields over time (Figure 7 in the main text, Figure 11) is estimated by
error propagation.
91
Figure 10. Example of a Hg concentration result given by the volumetric gradient approach
Figure 11. Stability of gaseous Hg concentration during tests performed on the new silicon-coated
cylinder
5.2. Thermal desorption of PTFE-coated and uncoated cylinders
Thermal desorption was performed on the PTFE-coated cylinder and the uncoated
cylinder up to 180°C (Figure 12 A and B). Results for the PTFE-coated cylinder suggest a
92
pattern similar to the one observed for the old silicon-coated cylinder, with an increase in
desorption rate starting before 100°C and a second increase at higher temperatures. In contrast
to the silicon-coated cylinder B, here the second increase is very high compared to the first
peak. The total amount of Hg desorbed during the experiment was 7.7 µg. However, Hg
desorption still occurred and we maintained a high temperature (190°C) during a few hours
before we re-evaluate Hg desorption. We found a low desorption rate, suggesting the cylinder
can be cleaned.
The uncoated cylinder however only revealed one peak, with a consistent increase in
Hg desorption rate with temperature, possibly because physically-adsorbed Hg0 is absent or
insignificant compared to oxidized Hg. At least 4.1 µg of Hg could be desorbed from the
uncoated cylinder, and some is still adsorbed inside.
Figure 12. Thermal desorption results for (A) the PTFE-coated cylinder and (B) the uncoated
cylinder
93
Acknowledgements
This work was supported by the Conseil Regional d ́ ’Aquitaine (20071303002PFM)

and the European Regional Development Fund (FEDER, 31486/08011464). Total is thanked
for both funding this research study and providing cylinders.
References
(1) Wilhelm, S. M.; Bloom, N. Mercury in petroleum. Fuel Process. Technol. 2000, 63 (1),
1−27.
(2) Lambertsson, L.; Lord, C. J.; Frech, W.; Bjorn, E. Rapid dissolution of cinnabar in crude
oils at reservoir temperatures facilitated by reduced sulfur ligands. ACS Earth Space Chem.
2018, 2 (10), 1022−1028.
(3) Lord, C. J.; Lambertsson, L. T.; Björn, E. L.; Frech, W.; Thomas, S. A. Removing Mercury
from Crude Oil. WO Patent 2014143457 A1, Sept 18, 2014.
(4) Huber, M. L.; Laesecke, A.; Friend, D. G. Correlation for the vapor pressure of mercury.
Ind. Eng. Chem. Res. 2006, 45 (21), 7351− 7361.
(5) Bingham, M. D. Field detection and implications of mercury in natural gas. SPE Prod.
Eng. 1990, 5 (02), 120−124.
(6) Wilhelm, S. M. Risk analysis for operation of aluminum heat exchangers contaminated by
mercury. Process Saf. Prog. 2009, 28 (3), 259−266.
(7) International Organization for Standardization (ISO). ISO 6978- 1:2003, Natural Gas
Determination of Mercury Part 1: Sampling of Mercury by Chemisorption on Iodine; ISO:
Geneva, Switzerland, 2003; http://www.iso.org/cms/render/live/en/sites/isoorg/contents/data/
standard/03/86/38600.html.
(8) Cachia, M.; Bouyssiere, B.; Carrier, H.; Garraud, H.; Caumette, ̀ G.; Le Hecho, I.
Development of a high-pressure bubbling sampler for trace element quantification in natural
gas. Energy Fuels 2017, 31 (4), 4294−4300.
(9) Frech, W.; Baxter, D. C.; Dyvik, G.; Dybdahl, B. On the determination of total mercury in
natural gases using the amalgamation technique and cold vapour atomic absorption
spectrometry. J. Anal. At. Spectrom. 1995, 10 (10), 769−775.
(10) Leeper, J. Mercury−LNG’s problem. Hydrocarbon Process. 1980, 59 (11), 237−240.
(11) Harfoushian, J. H. Quantification of low levels of mercury in gas reservoirs using
advanced sampling and analysis techniques. Proceedings of the SPE Annual Technical
Conference and Exhibition; New Orleans, LA, Sept 30−Oct 2, 2013; DOI: 10.2118/166220-
MS.
(12) Larsson, T.; Frech, W.; Björn, E.; Dybdahl, B. Studies of transport and collection
characteristics of gaseous mercury in natural gases using amalgamation and isotope dilution
analysis. Analyst 2007, 132 (6), 579−586.
94
(13) SilcoTek Corporation. Prevent Mercury Loss during Transport and Storage with
SilcoNert 2000; SilcoTek Corporation: Bellefonte, PA, 2019;
https://www.silcotek.com/prevent-mercury-loss.
(14) Rumayor, M.; Diaz-Somoano, M.; Lopez-Anton, M. A.; Martinez-Tarazona, M. R.
Mercury compounds characterization by thermal desorption. Talanta 2013, 114, 318−322.
(15) Edmonds, B.; Moorwood, R.; Szczepanski, R. Mercury partitioning in natural gases and
condensates. Proceedings of the GPA European Chapter Meeting, London, U.K., March 21,
1996.
(16) Zheng, W.; Lin, H.; Mann, B. F.; Liang, L.; Gu, B. Oxidation of dissolved elemental
mercury by thiol compounds under anoxic conditions. Environ. Sci. Technol. 2013, 47 (22),
12827−12834.
(17) Coufalik, P.; Komarek, J. The use of thermal desorption in ́ speciation analysis of mercury
in soil, sediments and tailings. J. Anal. Chem. 2014, 69, 1123−1129.
(18) Biester, H.; Scholz, C. Determination of Mercury Binging Forms in Contaminated Soils:
Mercury Pyrolysis versus Sequential Extractions. Environ. Sci. Technol. 1997, 31 (1), 233−239.
(19) Gajdosechova, Z.; Boskamp, M. S.; Lopez-Linares, F.; Feldmann, J.; Krupp, E. M. Hg
Speciation in Petroleum Hydrocarbons with Emphasis on the Reactivity of Hg Particles. Energy
Fuels 2016, 30 (1), 130−137.
(20) Lambertsson, L.; Lord, C. J.; Frech, W.; Björn, E. Rapid Dissolution of Cinnabar in Crude
Oils at Reservoir Temperatures Facilitated by Reduced Sulfur Ligands. ACS Earth Space Chem.
2018, 2 (10), 1022−1028
95
Partie 2 : Essai expérimental d’échantillonnage de
gaz naturel contaminé en Hg0 dans un sac Tedlar
Afin de déterminer si le phénomène d’adsorption du mercure est observé dans les
échantillonneurs notamment les sacs Tedlar en contact avec des espèces soufrées et de
l’oxygène, un échantillon de gaz pétrolier soufré réel a été introduit dans un sac Tedlar rempli,
au préalable, avec de l’air ambiant. Le remplissage du sac Tedlar avec l’air ambiant et le gaz
soufré a été effectué au sein du pôle PVT au CSTJF de TotalEnergies. Le gaz contenu dans le
sac Tedlar a ensuite été enrichi en Hg0 gazeux à l’IPREM et sa concentration a été analysée au
cours du temps pendant une période d’environ 2 mois et demi.
1. Matériels et Méthodes
1.1. Préparation et échantillonnage.
Un sac Tedlar d’un volume de 3 L contenant un peu moins d’1L d’air a été
complétement rempli avec un gaz pétrolier contenant 120 ppm d’H2S. Ce « prétraitement » par
l’air et l’H2S a eu pour but de simuler le « vieillissement » des sacs Tedlar par des prélèvements
de gaz contenant des composés soufrés. Le Hg0 gazeux saturé a été prélevé dans l’espace de
tête d’un récipient contenant du Hg0 liquide, puis, a été introduit directement à l’aide d’une
seringue étanche au gaz dans ce sac Tedlar rempli. La concentration en Hg0 gazeux obtenu était
de 5 ng.L-1. Le suivi de la concentration du mercure en fonction du temps a été réalisé en
échantillonnant directement 100 mL de gaz dans le sac Tedlar à l’aide d’une seringue étanche
au gaz. Le Hg0 gazeux a été analysé toutes les heures pendant les premières 12h, ensuite, chaque
jour (hors week-end) pendant 1 semaine et 1 fois par semaine pendant 2 mois. Afin d’assurer
la fiabilité de la mesure, l’échantillonnage ainsi que l’analyse ont été réalisée successivement
trois fois pour chaque temps.
96
1.2. Analyse du Hg0
Le Hg0 gazeux a été analysé par spectrométrie de fluorescence atomique (Sir Galahad, PS
Analytical, U.K.). Le gaz pétrolier, échantillonné dans la seringue, a été directement injecté
dans un piège en or analytique à l’intérieur de l’appareil et le Hg a été piégé sur celui-ci. Le
piège en or analytique a ensuite été chauffé à 500 °C pendant 15 s et un flux d'argon a transporté
le Hg gazeux libéré, vers le détecteur de fluorescence atomique. La concentration en Hg0 a été
estimée en fonction de la pente obtenue par gradient volumétrique (cf 5.Complete information,
p 91). La limite de détection instrumentale obtenue était de 10 pg.
2. Résultats et discussions
La concentration en Hg0 gazeux contenue dans le sac Tedlar est restée constante pendant
toute la durée du test, soit environ 2 mois et demi (Figure 13).
100%
80%
[Hg]t/[Hg]t0
60%
40%
20%
0%
0 200 400 600 800 1000 1200 1400 1600
Temps (h)
Figure 13. Variation de la concentration d'Hg0 gazeux au cours du temps pendant l'expérience de
stabilité menée sur le sac Tedlar
Ce résultat montre que lorsque les surfaces en fluoropolymère du sac Tedlar sont en
contact avec des espèces soufrées et de l’oxygène, on n’observe pas de perte significative d’Hg.
Les sacs Tedlar ont donc une bonne fiabilité pour échantillonner le mercure, du moins quand la
teneur de ce dernier est de l’ordre de quelques microgrammes par mètre cube.
97
3. Conclusion
La fiabilité des sacs Tedlar ne semble pas être altérée par leur contact avec les composés
soufrés durant les campagnes d’échantillonnage précédentes : ce qui permet de valider leur
utilisation. Toutefois, ce type d’échantillonneurs ne sont pas conçus pour échantillonner des gaz
sous pression et une dépressurisation ainsi qu’une dérivation du flux principal du gaz naturel
doivent être mise en place afin d’échantillonner celui-ci dans les sacs Tedlar. Ces interventions
peuvent être la source de contamination et le changement brutal de conditions induira un point
froid, ce qui pourrait fausser la concentration en Hg si toutes les précautions ne sont pas prises
concernant la sélection, la préparation et la purge du point d’échantillonnage.
98
Partie 3 : Recommandations sur l’échantillonnage et
de l’analyse du mercure dans le gaz naturel au cours
d’un essai de puits
Lors d'un essai de puits, la mesure du mercure peut facilement être mal déterminée et
conduire à la prise de décisions incorrectes. La principale incertitude ne provient pas de la
mesure elle-même ou de l'étalonnage de l'appareil, mais de la méthode de collecte.
L’objectif lors d’un essai de puits est donc d’assurer la traçabilité et la réduction des
artefacts qui peuvent survenir lors des opérations d'échantillonnage et d'assurer la
représentativité des échantillons pendant un essai de puits.
Afin de répondre à ces objectifs, cette partie 3 qui est tirée d’un « General Specifiation »
(Document confidentiel interne au sein de TotalEnergies), rédigée pendant cette période de
thèse, documente les événements clés pendant l'opération d'échantillonnage.
L’échantillonnage et l’analyse du mercure dans le gaz naturel au cours d’un essai de
puits, permet de prendre la décision concernant la nécessité de la mise en place d’une unité
d'élimination du mercure (MRU) au sein des compagnies pétrolières.
1. Normes existantes et leurs insuffisances
Deux méthodes d’échantillonnage sont généralement recommandées par les normes :

accumulation de mercure par amalgamation (ISO 6978.2-2003, 2003) et le barbotage de gaz
dans une solution acide (ISO 6978.1-2003, 2003, p. 69; Norme NF EN 13211, 2001). La
première méthode est recommandée uniquement car elle permet une plus grande sensibilité
(ISO 6978.2-2003, 2003) et permet d’éviter ou de limiter la manipulation des produits
chimiques sur site.
La spectrométrie en absorption atomique ou en fluorescence atomique est la méthode
recommandée. Leur sensibilité est largement suffisante et après la préconcentration via des
pièges d’amalgamation, toutes les interférences possibles sont éliminées, car les composants
dans le gaz ayant une longueur d’onde de fluorescence proche de celle du mercure sont très
largement éliminés et la détection par les spectromètres sont spécifiques aux longueurs d’onde
99
de mercure, ce qui rend les résultats très fiables (ASTM D5954-98, 2014; ASTM D6350-14,
2014).
Le respect strict de la norme ne permet pas cependant de garantir le succès d’une
campagne d’échantillonnage. Les artefacts sont nombreux et dans un grand nombre de cas, les
compagnies de services rendent des résultats ne permettant pas de les identifier : cela ne
permettant pas de valider ou d’invalider les résultats obtenus.
L’objectif principal de ce travail est donc d’établir une procédure permettant, d’une part,
d’éviter les erreurs et artefacts connus durant l’échantillonnage et d’autre part, de rendre
traçables les problèmes et les artefacts rencontrés pendant la campagne d’échantillonnage.
2. Problèmes spécifiques sur les sites industriels ainsi que durant

les tests de puits
Un analyste se contente parfois de déterminer correctement la teneur en mercure d’un

échantillon que l’on lui donne. Quand il s’agit d’une campagne de prélèvements pendant un test
de puits, la problématique est très différente. En dehors du fait qu’il faut garantir le dosage de
la bonne quantité de mercure dans un échantillon, il est également nécessaire de garantir :
1. qu’il n’y ait ni perte, ni contamination depuis la prise d’échantillon.
2. que la teneur en mercure de l’échantillon est bien représentative de celle dans le

fluide circulant dans la ligne pendant le test.
3. que la teneur en mercure du fluide circulant dans la ligne est bien représentative de
celle dans le fluide de réservoir.
3. Relation entre la durée du test et le seuil de détection
Il a été souvent constaté que la teneur en mercure dans le fluide, au début d’un test de
puits, n’est pas représentative de celle dans le fluide de réservoir. Plusieurs raisons peuvent
expliquer ce phénomène :
100
• La boue utilisée par le forage peut affecter cette mesure soit parce qu’elle peut
contenir des composés contenant du mercure, soit parce qu’elle peut contenir des
composés pouvant capturer des espèces de mercure. Il est alors indispensable
d’évacuer complètement cette boue avant de pouvoir obtenir des mesures non
perturbées par celle-ci.
• L’équilibre adsorption/désorption de mercure sur la paroi interne des « pipes » et

des équipements de test est fréquemment évoqué pour expliquer la « percée »
souvent tardive du mercure.
Le temps de « percée du mercure » a été calculé sur trois tests de puits différents (Figure
14). Pour les trois tests de puits, le moment où on observe la percée du mercure est assez inégale.
Dans deux cas (Puits de gaz A et C), la production d’au moins 1-2 millions de mètres cubes de
gaz a été nécessaire pour observer une stabilisation de la teneur en mercure alors que pour les
puits de gaz B, la production de plus de 5-6 millions de mètres cubes de gaz a été nécessaire
avant d’observer une stabilité du Hg.
101
Percée du mercure
Figure 14. Temps de "percée du mercure" obtenu en fonction du volume de gaz produit lors de
trois tests de puits
Cependant, la connaissance de la teneur en mercure du gaz, après stabilisation, permet

de calculer l’ordre de grandeur de la quantité totale de mercure retenue par l’équipement de test
avant la percée. En représentant la concentration en mercure du gaz mesurée pendant les tests,
non pas en fonction du temps ou du cumul du volume de gaz produit, mais en fonction de la
quantité totale de mercure censée être produite, il est observé cette fois-ci que la quantité de
mercure retenue avant la percée est de l’ordre de quelques dizaines de grammes (Figure 15).
102
Percée
Percée dude mercury
mercure
1000.0 µg/Nm3
100.0 µg/Nm3
puits de gaz A
10.0 µg/Nm3
puits de gaz B
puits de gaz C
1.0 µg/Nm3
0.1 µg/Nm3
0g 50 g 100 g 150 g 200 g 250 g 300 g
Figure 15. Temps de "percée du mercure" obtenu en fonction de la quantité totale produite estimée
lors de trois tests de puits
Cette observation permet d’affirmer que la limite basse de détection du mercure dans le
gaz, pendant un test de puits, dépend du volume total de gaz produit par le puits. Si un test est
planifié pour produire 5 millions de mètres cubes de gaz, il ne serait pas raisonnable d’espérer
déterminer une concentration précise de mercure dans le gaz si la teneur en mercure est de
l’ordre 1 µg.Nm-3, car, pour mesurer une telle valeur, il faudrait débiter 50 à 100 millions de
mètres cubes de gaz au préalable. Si après production de 5 millions de mètres cubes de gaz, la
teneur en mercure mesurée reste faible, la teneur en mercure du gaz dans le réservoir ne devrait
pas dépasser 10-20 µg.Nm-3 car sinon, la percée devrait avoir déjà eu lieu.
L’adsorption d’une couche monoatomique très dense de mercure sur une surface est de
l’ordre de 1 mg.m-2. La surface macroscopique totale d’un équipement de test est typiquement
de l’ordre de quelques centaines jusqu’à un millier de mètres carrés. Par conséquent, il est
difficile d’expliquer la rétention de quelques dizaines de grammes de mercure par l’adsorption
seule, à moins de prendre en compte la surface des produits de corrosion et des microparticules.
103
Une passivation de surface pourrait améliorer la vitesse de percée, mais sans prise en compte
des microparticules naturellement présentes dans les pipelines et celles introduites par la boue
de forage.
Il n’est donc pas judicieux lors d’un test de puits de mettre un point d’honneur à
déterminer précisément la concentration en mercure dans le gaz car s’il est important de
connaître la teneur en mercure du fluide pour prévoir les problèmes et dimensionner le
traitement, le surcoût lié à l’augmentation de la durée de test empêche de descendre plus bas en
sensibilité. Le but d’une campagne d’échantillonnage pendant un test de puits n’est donc pas
de rechercher le mercure dans tous les ordres de grandeurs de concentration possible, mais de
chercher à savoir si la teneur est en dessous de quelques microgrammes par mètre cube de gaz
et quelle est la valeur de cette teneur si elle dépasse ce seuil.
4. Traque et élimination des artéfacts liés à la sélection et la

préparation du point d’échantillonnage
Le mercure est un élément qui est fractionné entre la phase gazeuse et la phase liquide
lorsque les deux phases sont présentes. Le coefficient de répartition entre les deux phases
dépend de la pression et de la température, et plus modérément, des caractéristiques du gaz et
du liquide. Pour cette raison, un entraînement de liquides ou de particules dans l’échantillon de
gaz peut fausser notablement son analyse de la teneur en mercure. Il est donc important
d’échantillonner un gaz propre, exempté de liquides et des particules.
De plus, une fois que le gaz et le liquide sont analysés, afin de remonter à la teneur en
mercure du fluide global, il faut que les deux phases soient assez proches d’un état d’équilibre.
Pour ces raisons, les points principaux d’échantillonnage pour la détermination de mercure
doivent être les sortie gaz et sortie liquide du séparateur de test. Si le séparateur ou les
connexions de celui-ci sont suspectés d’être contaminés par le mercure, un point
d’échantillonnage alternatif peut être choisi pour vérifier cette suspicion. Dans ce cas, il faut
veiller à ce que le volume mort entre la valve d’échantillonnage et la conduite principale soit
inférieur à 0,5 L afin d’éviter que l’échantillon contienne un fluide de stagnation.
Lorsque l’échantillonnage de gaz naturel s’effectue par sacs Tedlar, le volume purgé
doit être au moins 10 fois supérieur à la quantité de gaz stagnant dans le volume mort. Il est
104
recommandé que la valve de détente pour l’échantillonnage soit chauffée afin d’éviter la
formation de liquide ou l’augmentation d’adsorption liée au froid créé par la détente.
Lorsque le mercure est échantillonné par amalgamation en connectant directement
l’échantillonneur au point d’échantillonnage, celui-ci doit être placé au moins 10 minutes avec
un débit de purge de l’ordre de 20 L (conditions atmosphériques) par minute à travers une ligne
de by-pass. Cette purge doit être maintenue lors de la prise d’échantillon. Le débit recommandé,
qui traverse les pièges, doit être compris entre 0,2 et 1 L par minute afin d’éviter un temps de
résidence insuffisant du gaz dans les pièges.
5. Règles depuis l’échantillonnage jusqu’à l’analyse afin

d’observer aucune perte et contamination de l’échantillon
Il est connu que les bouteilles d’échantillonnage sous pression préservent mal le mercure
(cf Partie 1 : Experimental Tests of Natural Gas Samplers Prior to Mercury Concentration
Analysis, p 77). Depuis, différents types d’inertage ont été inventés dans le but de préserver le
mercure dans l’échantillon prélevé (SilcoTek Corporation, 2019). Ces bouteilles sont largement
utilisées aujourd’hui par beaucoup de laboratoires de services. Cependant, il a été démontré que
lorsque les bouteilles ne sont plus « neuves », il est non seulement possible de perdre à nouveau
le mercure dans ces bouteilles, mais aussi, le mercure « stocké » dans ces vieilles bouteilles
peut potentiellement contaminer un gaz qui n’en contenait pas ou peu (cf 3.4. Mass Balance
and Reaction Rates, p 89).
Par conséquent, l’utilisation de ces bouteilles lors de l’échantillonnage de gaz naturel
pour l’analyse du Hg est à éviter. L’utilisation des sacs Tedlar ou la prise d’échantillon par
amalgamation doivent être privilégiées. L’analyse directe sur site, si elle est possible, reste la
méthode la plus fiable pour de l’analyse du mercure lors d’un essai de puits.
En effet, lors des analyses sur site, de nombreux artefacts peuvent subvenir et ne sont
pas toujours prévisibles. Il est donc important d’essayer de les déterminer immédiatement lors
d’analyse directe sur site car dans le cas d’une analyse hors site, à posteriori, les dégâts
occasionnés sur les résultats ne pourront pas être corrigés mais juste constatés.
La vérification croisée de la teneur en mercure dans le gaz et dans l’hydrocarbure liquide
hors mercure particulaire pouvant être éliminées par filtration ou sédimentation, doit être
effectuée. Ces valeurs doivent être compatibles à celle d’équilibre dans les conditions de
105
séparateur. La vérification de la teneur en mercure de fluide de forage et fluide de
complétion est également utile afin d’identifier et éliminer certaines sources d’artefacts.
De plus, la contamination de l’équipement d’échantillonnage est aussi un souci réel.
L’absence de mercure dans le système d’échantillonnage (vannes, tubes et connexions) utilisé
en amont des pièges à mercure, ainsi que les bouteilles utilisées pour la prise d’échantillons de
liquides doivent être vérifiées avant la campagne d’échantillonnage.
La valeur de blanc des pièges d’amalgamation est une question clé. La pratique la plus
commune pour obtenir un blanc satisfaisant pour chaque piège est de les nettoyer par des cycles
de chauffage répétés dans un four lié à un spectromètre, jusqu’à ce que la valeur résiduelle
trouvée soit en dessous d’une limite qu’on considère comme satisfaisante. Il arrive cependant
qu’en dessous de certaines valeurs, la teneur résiduelle en mercure d’un piège ne continue pas
à diminuer malgré les cycles de chauffe, mais persiste à un niveau non satisfaisant ou devient
erratique sans qu’on ne comprenne son origine. Face à cette difficulté, certains augmentent le
volume d’échantillons qui accroît la quantité totale de mercure piégée, hissant celle-ci bien au-
dessus de la zone de blanc. D’autres déduisent la quantité de mercure trouvée dans le piège de
la valeur du dernier blanc trouvée.
Ces pratiques sont cependant problématiques car le mercure ne peut être généré
spontanément dans le piège. Le fait de ne pas arriver à abaisser suffisamment les valeurs de
blanc signifie simplement qu’il existe des sources de contamination non identifiées.
L’augmentation du volume de gaz échantillonné peut avoir deux effets négatifs. La quantité en
mercure passant dans le système d’échantillonnage accroît donc, augmentant de ce fait son
degré de contamination. Cela a aussi pour effet, d’accumuler également plus de composés
pouvant interférer avec l’efficacité de piégeage comme l’eau, les hydrocarbures lourds ou les
composés soufrés.
Quant à la déduction de la dernière valeur de blanc, cela revient à supposer que la source
de contamination contamine le piège de façon répétable, ce qui n’a pas de sens.
Dans le cas d’une opération d’échantillonnage, il est déjà arrivé que les bouchons des
pièges de mercure ont été identifiés que la source de contamination. Cette contamination est
arrivée probablement après que les bouchons aient touché une surface de travail contaminée.
Dans ce cas, la contamination était d’une centaine de fois supérieure à la quantité de mercure
capturée lors de la prise d’échantillon. La simple déduction de la valeur du dernier blanc ne
permet nullement la prise en compte de ce problème.
106
Les pièges de mercure peuvent être aussi contaminés par les particules présentes dans
ceux-ci. Le Tableau 5 présente les résultats de mesures à l’entrée et à la sortie d’une unité de
démercurisation (MRU).
Tableau 5. Concentrations en mercure mesurées à l’entrée et à la sortie d’une MRU
L’ordre de grandeur de concentrations en Hg retrouvées en entrée et en sortie ne change pas en

traversant l’unité de traitement et reste au-dessus de la spécification soit 10 ng.Nm-3 : ce qui
pourrait signifier que l’MRU ne fonctionne plus correctement. Mais une analyse plus fine de
ces résultats nous conduit à une conclusion radicalement différente : ces valeurs sont
principalement dues aux valeurs de blanc des pièges utilisés. En effet, en oubliant la
concentration affichée et en représentant la quantité de mercure trouvée dans chacun de ces
pièges en fonction du volume de gaz les ayant traversés, la teneur en mercure des pièges ne
varie quasiment pas avec le volume d’échantillon (Figure 16.
107
Figure 16. Concentrations en mercure analysées dans chaque piège en fonction du volume de gaz
La quantité de mercure trouvée dans les pièges reste dans la même zone de concentration
soit de 0,9 à 1,7 ng pour les mesures à l’entrée de l’unité alors qu’en extrapolant la relation
quantité de mercure par piège versus volume de gaz échantillonnée, on obtient des valeurs de
1,4 à 1,8 ng qui correspondent aux blancs statistiques.
Il a été démontré que le mercure n’est pas seulement piégé par la partie du piège
contenant de l’or, mais une quantité importante peut être stockée dans le tube de quartz,
probablement associée à des particules (Tableau 6. Le niveau de blanc d’un piège peut être
abaissé au picogramme avec 2 cycles de chauffe. Mais le mercure a été éliminé lors d’un cycle
de chauffe, seulement sur la partie centrale du piège. Le mercure retenu par les particules restant
dans le tube de quartz loin de la partie centrale n’a quant à lui pas pu être éliminé. De ce fait,
lorsqu’il y a un déplacement de la position du piège, ou lorsqu’un flux de gaz déplace ces
particules, une quantité de mercure de l’ordre de 1000 fois de celle du dernier blanc pourrait
être observée.
108
Tableau 6. Concentrations en Hg obtenues après analyses de blancs des pièges en changeant leurs
positions lors du nettoyage
La difficulté de garantir des valeurs de blanc a été mise en évidence à travers ces
exemples. La vérification des blancs avant la prise d’échantillons est nécessaire, mais ne
garantit nullement l’absence de mercure dans les pièges. D’où la recommandation de ne pas se
fier aux mesures effectuées sur des pièges isolés, mais faire des mesures par groupes contenant
3 à 6 pièges ayant vu de volume différent de gaz variant dans une proportion d’au moins 1 à 4.
109
CHAPITRE III : FIABILITE DE
L’ECHANTILLONNAGE ET DE
L’ANALYSE DE L’ARSENIC DANS
LE GAZ NATUREL
Il existe différentes méthodologies pour la détermination de la concentration de l’arsenic

dans le gaz naturel (cf 2.1.2.Gas Samplers: Cylinders and Inert Plastic Bags (Tedlar Bags), p
60). Comme pour le mercure, l'échantillonnage du gaz naturel à haute pression peut être effectué
à l'aide de cylindres en acier inoxydable avant de mesurer l’As hors site.
Le revêtement intérieur de ces cylindres est supposé être inerte et donc empêcher
l'adsorption de l’arsenic à l’intérieur de ceux-ci, rendant l'analyse de l’As fiable. Afin de
déterminer si l’échantillonnage de gaz naturel pour l’analyse de l’arsenic dans ces cylindres est
fiable, différentes expérimentations ont été développées.
Plusieurs échantillons de gaz naturel provenant de différents puits d’un même pays ont
été prélevés dans des cylindres en acier inoxydable en 2009, 2010 et 2013. Ces campagnes
d’échantillonnage de gaz naturel ont été réalisées afin de déterminer si les fluides provenant des
puits étaient contaminés en As. Les échantillons de gaz naturel ont résidé dans ces cylindres
jusqu’aux expérimentations réalisées en 2019. La concentration en arsenic contenue dans les
échantillons de gaz naturel a été déterminée par HR-ICP-MS de solutions de barbotage obtenues
après vidage complet des cylindres dans celles-ci. Une chauffe des cylindres a été réalisée afin
de déterminer si de l’arsenic s’était absorbé sur leurs surfaces internes.
Deux cylindres contenaient du gaz naturel ainsi que du condensat de gaz : cette matrice
organique a été aussi analysée afin de déterminer la concentration en arsenic total ainsi que sa
spéciation.
110
1. Matériels et Méthodes
1.1. Echantillonnage
Six échantillons de gaz naturel ont été prélevés dans un même pays, sur différents puits
et à des dates différentes (cf Tableau 7). Les échantillons de gaz naturel ont été prélevés dans
des cylindres Swagelok® en acier inoxydable à doubles extrémités de 300 cc. Les cylindres
remplis ont été stockés au sein du pôle PVT du Centre Scientifique et Technique de
TotalEnergies jusqu’en 2019.
Tableau 7. Informations sur les échantillons de gaz naturel analysés
Code Numéro Date de Volume

Type d’échantillon
échantillon du puits prélèvement échantillon (m3)
A–1 1 2009 0,032 Gaz naturel
Gaz naturel +
A–2 1 2009 0,027
condensats de gaz
A–3 2 2010 0,0315 Gaz naturel
A–4 3 2013 0,0325 Gaz naturel
(Gaz naturel) +
A–5 3 2013 0
condensats de gaz
A–6 3 2013 0,025 Gaz humide
L’échantillon A – 5 devait contenir du gaz naturel, après ouverture, il y avait uniquement

présence de condensat de gaz. Il semble donc que le cylindre se soit vidé au cours du temps.
Les condensats de gaz des échantillons A – 2 et A – 5 ont été récupérés dans des flacons en
verre, fermés hermétiquement avec des septums et ont été placés au noir à température ambiante
avant leurs analyses.
1.2. Procédures d’analyse
1.2.1. Mesure de la concentration en arsenic total par système de barbotage
La concentration en arsenic totale de chaque échantillon de gaz naturel a été déterminée en

faisant barboter la totalité des gaz contenus dans chaque cylindre dans un système de barbotage
(Figure 17).
111
Figure 17. Schéma du système de barbotage utilisé pour la détermination de l'arsenic totale dans les
échantillons de gaz naturel
Une valve du cylindre a été ouverte et le flux de gaz sortant a été ajusté à un débit de 200
mL.min-1. Le gaz a barboté dans 3 barboteurs montés en série contenant chacun 10 mL d’acide
nitrique à 10% (Cachia, 2017). Une fois les cylindres vidés, les solutions de barbotage ont été
récupérées puis diluées avec un FD = 5 avec de l’eau MQ supra-pure et du germanium y a été
ajouté comme étalon interne. Les solutions de barbotage ont ensuite été analysées par HR-ICP-
MS selon les paramètres présentés Tableau 8.
Tableau 8. Paramètres d’analyse par HR-ICP-MS des solutions de barbotage obtenues après vidage
complet des échantillons de gaz naturel contenus dans les cylindres ainsi que lors de la désorption
thermique
ICP-MS Thermo Element XR

Condition plasma Humide
Puissance générateur 1145 W
Gaz plasma Ar
Ar auxiliaire 1,00 L.min-1
Résolution Haute
Cônes Nickel
Réplicats 20
Type scan EScan
Temps échantillon 0,200 s
Mode détection Triple
72
Eléments (isotopes) Ge et 75As
Afin d’assurer la fiabilité de la mesure, les solutions de barbotage ont été analysées en
triplicat. La concentration en arsenic total a été déterminée en fonction de la pente obtenue à
112
partir d’une courbe d’étalonnage en As élémentaire. Les limites de détection et de quantification
instrumentale obtenues étaient respectivement de 32 ppt et de 96 ppt.
1.2.2. Analyse des condensats de gaz contenus dans les cylindres
Les condensats de gaz provenant de l’échantillon A – 5 et A – 3 (Tableau 7) ont été

dilués (FD = 2 et 10) dans du toluène et le xylène. Les échantillons ont été dilués en triplicatas
dans chacun des deux solvants. Ceux-ci ont été analysés par GC-ICP-MS selon les paramètres
optimisés par (Bouyssiere et al., 2001) (Tableau 9).
Tableau 9. Paramètres d’analyse par GC-ICP-MS des condensats de gaz selon Bouyssiere et al.,
2001
GC-MS Thermo ISQ QD

Colonne chromatographique HP – 5 (30 m × 0,32 mm × 0,25 µm)
Injection Splitless, temps de purge : 1 min
Température d’injection 150°C
Volume d’injection 1 µL
Gaz porteur Helium (2 mL/min)
Programme de température du four 60°C (1min) à 280°C de 50°C/min
Température ligne de transfert 250°C
ICP-MS Thermo X Series 2
Gaz plasma Argon
Ar auxiliaire 0.9 L.min-1
Cônes (Sampler et Skimmer) Nickel
Réplicats 5
13
Eléments (Isotopes) C,33S,35Cl, 75As, 118Sn, 121Sb et 199,202Hg
En plus des condensats dilués, les condensats purs ont pu être injectés et analysés par
cette même technique d’analyse. La concentration en As total a été semi-quantifiée en fonction
de la pente obtenue à partir d’une courbe d’étalonnage en Ph3As préparée dans le toluène et le
xylène. La limite de détection et de quantification instrumentale était respectivement de 81 ppt
et 243 ppt.
113
1.2.3. Expérience de désorption thermique
Après l’analyse de la concentration en As total des cylindres, une désorption thermique

a été réalisée sur les cylindres A-3, A-6 et A-3 (Tableau 7). Les cylindres ont été placés dans
un four programmable sous un flux d'argon à un débit d’environ 200 mL.min-1 (Figure 18). La
température expérimentale initiale était d’environ 20°C (T°C ambiante). La température du four
a été augmentée de 10 °C toutes les 10 minutes et la procédure a été répétée jusqu'à 200°C. Puis
les cylindres ont été laissés dans le four pendant 2h à 200°C.
Figure 18. Schéma du système de désorption thermique pour la détermination de l’arsenic absorbé
dans les cylindres.
L’arsenic désorbé a été piégé dans des solutions de 10% d’HNO3 de la même façon que
pour la « Partie 1.2.1, p 111 ». Les solutions une fois diluées (FD = 5) ont ensuite été analysées
par HR-ICP-MS selon les paramètres présentés Tableau 8.
2.1. Mesure de la concentration en arsenic total par système de barbotage
Suite au barbotage de la totalité des gaz contenus dans chaque cylindre dans le système
de barbotage, la présence d’arsenic dans les solutions de barbotage n’a pas été mise en évidence
après leurs analyses par HR-ICP-MS. La limite de détection instrumentale de l’appareil était de
32 ppt. Sachant que le volume de gaz qui a barboté dans ces solutions était compris entre 0,025
114
et 0,0325 m3 (Tableau 7), la concentration en arsenic dans celles-ci était donc inférieure à la
limite de détection expérimentale qui était de ~ 50 ng.m-3 (Tableau 10).
Tableau 10. Limite de détection expérimentale obtenue pour chaque échantillon de gaz naturel en
fonction du volume de gaz barboté
Code Limite de détection

Echantillon (ng.m-3)
A–1 50
A–2 59
A–3 51
A–4 49
A–6 64
La désorption thermique de ces cinq cylindres a donc été nécessaire afin de déterminer
si l’arsenic potentiellement présent dans ces échantillons de gaz naturel s’était adsorbé à la
surface interne de ces cylindres.
2.2. Analyse des condensats de gaz contenus dans les cylindres
Suite à l’analyse de la courbe d’étalonnage de Ph3As réalisée dans le toluène et dans le

xylène par GC-ICP-MS, le temps de rétention de cette espèce arséniée avec cette technique
d’analyse était de 374 mS (Figure 19).
115
Figure 19. Chromatogramme du standard de Ph3As à 25 ppb dans le toluène
L’analyse des condensats de gaz retrouvés dans les cylindres des échantillons A – 2 et A – 5 dilués et purs n’ont pas pu mettre en évidence la
présence de Ph3As et de d’autres composés arséniés (Figure 20).
116
Figure 20. Chromatogramme de l’arsenic du condensat de gaz de (A.) l’échantillon A – 2 et de (B.) l’échantillon A – 5. Analyse des échantillons purs.
117
La désorption thermique du cylindre ayant contenu l’échantillon A – 5 semble tout de
même nécessaire afin de déterminer si l’arsenic potentiellement présent dans le condensat de
gaz ainsi que dans le gaz naturel initialement présent dans le cylindre, ne s’est pas adsorbé à la
surface interne de ce cylindre.
2.3. Expérience de désorption thermique
Suite à la chauffe des cylindres à 200°C pendant 2 h, l’arsenic potentiellement désorbé

de la surface interne de ceux-ci a été analysé dans les solutions de piégeage du système de
barbotage Figure 18. De l’arsenic a été retrouvé dans les solutions de barbotage de trois
cylindres désorbés thermiquement (Tableau 11).
Tableau 11. Concentration en As désorbé thermiquement des cylindres
Code Echantillon Concentration (µg.m-3)

A–1 < LD
A–2 < LD
A–3 0,24
A–4 0,74
A–5 0,23
A–6 < LD
La concentration en arsenic désorbé la plus importante était pour l’échantillon A – 4 où 0,74

µg.m-3 d’As, a été retrouvée. Ces résultats montrent que l’arsenic contenu dans les échantillons
de gaz naturel s’était effectivement adsorbé à la surface interne des cylindres.
De plus, l’échantillon de gaz naturel A – 4, A – 5 (présence uniquement de condensats
de gaz) et A – 6 ont été échantillonnés dans le même puits , la même année (Tableau 7) : ce qui
montre que le gaz naturel provenant de ce puits était effectivement contaminé en arsenic. Suite
à ces résultats, la compagnie TotalEnergies a donc décidé de renouveler une campagne
d’échantillonnage de gaz naturel en août 2019 sur ce puits, afin de pousser leur investigation
quant à cette contamination du gaz naturel par l’As.
118
3. Conclusion
L’analyse des condensats de gaz et des solutions de piégeage obtenues suite au

barbotage de la totalité des gaz contenus dans chaque cylindre, ainsi, n’ont pas permis de mettre
en évidence la présence d’arsenic dans les échantillons de gaz naturel. La concentration en
arsenic dans ceux-ci était inférieure à ~ 50 ng.m-3. Malgré ces résultats, la désorption thermique
des cylindres ayant contenu ces échantillons de gaz naturel a permis de mettre en évidence que
l’arsenic contenu dans ceux-ci s’était adsorbé à la surface interne des cylindres.
Ces résultats expérimentaux ont donc montré que les cylindres en acier inoxydable
n’étaient pas inertes à l’adsorption de l’arsenic et étaient donc inefficaces pour
l’échantillonnage de gaz naturel avant l'analyse de l'As.
D’autres expérimentations sont nécessaires afin de déterminer si cette adsorption est liée
à l'adsorption des espèces de soufre sur les surfaces internes des cylindres en acier inoxydable,
devenant des sites d’adsorption pour l’arsenic comme pour le mercure (cf 4.Conclusion and
Recommendations, p 90). De plus, l’évaluation de l’utilisation d’échantillonneurs alternatifs de
gaz naturel comme les cylindres revêtus de silicium et les sacs Tedlar doit être effectuée afin
de déterminer si ceux-ci sont inertes à l’adsorption de l’arsenic.
119
CHAPITRE IV :
COMPREHENTION DU
PHENOMENE D’ADSORPTION DE
L’ARSENIC ET DU MERCURE
SUR LA SURFACE INTERNE DES
CYLINDRES
Afin d’essayer de mieux comprendre le phénomène d’adsorption de l’arsenic et du

mercure sur la surface interne de cylindres de différents types lors de prélèvement de gaz
naturel, du biogaz a été prélevé sur le site de Mendixka à Charrite-le-bas (site de valorisation
de déchets verts et ménagers) à l’aide d’un réacteur contenant des plaques (en inox, recouvertes
du revêtement Silconert® et Dursan®) représentant expérimentalement les cylindres
d’échantillonnage. Ce biogaz est le produit résultant de la décomposition de déchets ménagers
enfouis sur le site. Une analyse de surface des plaques a ensuite été réalisée par ablation laser
couplée à un ICP-MS et par Scanning Electron Microscope (SEM) couplée à la spectroscopie
de rayons X à dispersion d’énergie (EDX). Cette dernière analyse a été réalisée par le Bordeaux
Imaging Center (BIC) afin de mieux observer et comprendre ce phénomène d’adsorption.
120
1. Matériels et méthodes
1.1. Détermination de la concentration en arsenic total dans le biogaz du site

de Mendixka
La concentration en arsenic total contenue dans le biogaz a été déterminée en faisant

barboter ce dernier dans un système de barbotage (Figure 21). Le système de barbotage à
directement été connecté au flux primaire de biogaz. Le débit de ce flux était ~ 35 ml.min-1. Le
biogaz a barboté dans 3 barboteurs montés en série contenant chacun 10 mL d’acide nitrique à
10% (Cachia, 2017) pendant 120h. La quantité totale de biogaz qui a barboté dans ces solutions
était ~ 252 L.
Figure 21. Système de barbotage mise en place sur le site de Mendixka pour la détermination de la
concentration en As total dans le biogaz.
Les solutions de barbotage ont été récupérées puis diluées avec un FD = 5 avec de l’eau
MQ supra-pure et du germanium y a été ajouté comme étalon interne. Les solutions de
barbotage ont ensuite été analysées par ICP-MS avec utilisation de la cellule de collision selon
les paramètres présentés Tableau 12. La cellule de collision a été utilisée afin de limiter les
interférences ArCl-. Un mélange d’argon et d’hélium a été utilisé comme gaz auxiliaire lors de
l’utilisation de la cellule de collision.
121
Tableau 12. Paramètres d’analyse par ICP-MS des solutions de barbotage obtenues après
barbotage du biogaz de Mendixka
ICP-MS Agilent 7700 s

Gaz plasma Ar
Cônes Nickel
Réplicats 25
Cellule de collision N2 = 5.2 mL.min-1
72
Eléments (Isotopes) Ge et 75As
Afin d’assurer la fiabilité de la mesure, les solutions de barbotage ont été analysées en
triplicat. La concentration en arsenic total a été déterminée en fonction de la pente obtenue à
partir d’une courbe d’étalonnage en As élémentaire. Les limites de détection et de quantification
instrumentale obtenues étaient respectivement de 53 ppt et de 156 ppt.
1.2. Détermination de la concentration en mercure total dans le biogaz du site

de Mendixka
La concentration en Hg total a été déterminée en échantillonnant le biogaz à l’aide de

pièges Amasil®. Les pièges ont été directement connectés sur le flux de gaz primaire. Différents
volumes de biogaz ont été prélevés : 5, 10, 20, 30 et 40 mL. Chaque prélèvement de gaz aux
différents volumes a été effectué en duplicat.
Le mercure a ensuite été analysé par spectrométrie de fluorescence atomique (Sir
Galahad, PS Analytical, U.K.). La procédure d’analyse était la même que pour la partie « 2.5.
Hg Analysis, p 83 ». La concentration en Hg0 a été estimée en fonction de la pente obtenue par
gradient volumétrique (cf 5.Complete information, p 91). La limite de détection instrumentale
obtenue était de 10 pg.
122
1.3. Analyse de surface des plaques représentant expérimentalement les
cylindres d’échantillonnage
1.3.1. Prélèvement du biogaz dans le réacteur
Les plaques représentant expérimentalement les cylindres d’échantillonnage ont été

obtenues auprès de Silcotek®. Elles mesuraient toutes 4 cm2 et avaient une épaisseur de 0,1
cm. Il y avait trois catégories de plaques. La première catégorie était en acier inoxydable sans
revêtement mais avec un traitement de surface. Le deuxième était en acier inoxydable
recouverte par le revêtement Silconert®. La troisième était en acier inoxydable recouverte du
revêtement Dursan®.
Le réacteur a été conçu au sein de UPPATech à l’université de Pau. Il a été fabriqué en
téflon chargé de 20% de verre afin de limiter toute adsorption à l’intérieur de celui-ci. Il était
composé d’une cavité centrale où deux portoirs à plaques ont été placés (cf Figure 22. Chaque
portoir pouvait contenir 4 plaques.
Lors de l’échantillonnage, les 3 catégories de plaques ont été placées sur le portoir dans
le réacteur soient 3 plaques recouvertes par le revêtement Silconert®, 3 plaques recouvertes du
revêtement Dursan® et 2 en acier inoxydable sans revêtement.
Le réacteur a été connecté directement sur le flux primaire de biogaz du site de
valorisation de déchets de Mendixka (Figure 22.). Le biogaz n’a pas été retenu dans le réacteur
mais est passé à l’intérieur de celui-ci sous flux constant.
123
Figure 22. Vue intérieure du réacteur (A.) utilisé lors de l’échantillonnage du bio gaz de Mendixka
(B.).
Le débit du flux primaire était ~ 5,5 mL.min-1. Le biogaz a traversé le réacteur pendant
120h. Les plaques ont donc été en contact avec ~ 39,6 L de biogaz. Après l’échantillonnage, le
réacteur a été fermé hermétiquement afin d’éliminer toute source de contamination avant
l’analyse de surface des plaques.
1.3.2. Analyse de surface par LA-ICP-MS
Les plaques ont été récupérées des portoirs du réacteur et analysées par LA-ICP-MS
selon les paramètres présentés dans le Tableau 13. La cellule de collision de l’ICP-MS a été
utilisée lors de l’analyse de l’arsenic afin de limiter les interférences ArCl-. L’hélium a été
utilisé comme gaz auxiliaire lors de l’utilisation de la cellule de collision.
124
Tableau 13. Paramètres d’analyse par LA-ICP-MS des plaques utilisées lors de l’échantillonnage
du biogaz sur le site de Mendixka
LA ESI NWR – 213

Energie 30 %
Fluence 1,1 J.cm-2
Fréquence 20 Hz
Diamètre faisceau laser 200 µm
Vitesse déplacement 500 µm.s-1
Gaz vecteur He = 800 mL.min-1
ICP-MS Agilent 7700 s
Gaz plasma Ar
Gaz secondaire 02
Cônes Platine
Réplicats 15
Cellule de collision He = 9,6 mL.min-1
32
Eléments (Isotopes) S,75As et 202Hg
Une plaque de chaque catégorie, n’ayant pas été utilisée lors de l’échantillonnage du
biogaz sur le terrain, a été analysée par LA-ICP-MS en tant que blanc. La quantification de
l’arsenic, du mercure et du soufre n’a pas été réalisée lors de ces analyses. Les blancs ainsi que
les plaques en inox non-revêtues et les plaques avec le revêtement Silconert® ont été scannées
une seule fois. Les plaques avec le revêtement Dursan® ont été scannées 4 fois.
1.3.3. Analyse de surface par SEM-EDX
Suite à l’analyse par LA-ICP-MS, une analyse de surface de chaque plaque a été réalisée
par SEM-EDX par Mme Svahn au Bordeaux Imaging Center (BIC). Les paramètres d’analyse
sont présentés dans le Tableau 14.
125
Tableau 14. . Paramètres d’analyse par SEM-EDX des plaques utilisées lors de l’échantillonnage
du biogaz sur le site de Mendixka
SEM Zeiss GeminiSEM300 – EDX Bruker XFlash 6/60

Tension d’accélération 15 kV
WD Dursan = 7,7 mm ; Silconert = 8,5 mm
Détecteur électrons secondaires – topographie SE2
Détecteur électrons rétrodiffusés – contraste
BSD1
chimique
Vacuum mode Vide élevé
Ces analyses ont permis d’obtenir un spectre des éléments contenus sur la surface de
chaque plaque, plus particulièrement sur les « taches » visualisées par microscope optique
(Figure 26). Ces analyses ont aussi pu déterminer l’abondance de chaque élément en
pourcentage sur ces mêmes taches.
2.1. Détermination de la concentration en arsenic total dans le biogaz du site

de Mendixka
Suite au barbotage du biogaz dans le système de barbotage, de l’arsenic a été retrouvé

dans le barboteur n°1. La pente de la gamme étalon en arsenic dont l’équation était la suivante :
y = 5555,2 x + 2789,1 ; a permis de déterminer que la concentration en As retrouvé dans le
barboteur était de 74 ± 0,03 ppt. Sachant que le volume de biogaz qui a barboté dans les
solutions de piégeage était de 252 L soit 0,252 m3, la concentration en As retrouvée dans le
biogaz de Mendixka était de 932,4 ng.m-3. Le réacteur a donc pu être mis en place sur le flux
primaire de biogaz afin de mieux observer et comprendre l’adsorption de l’arsenic sur la surface
des plaques.
126
2.2. Détermination de la concentration en mercure total dans le biogaz du site
de Mendixka
La concentration en Hg total a été analysée dans chaque piège Amasil® ayant été utilisés
pour échantillonner différents volumes en duplicat de biogaz. La pente obtenue par gradient
volumétrique était de 7,4 et a permis d’estimer la concentration en Hg total contenu dans le
biogaz produit sur le site de Mendixka (Tableau 15.
Tableau 15.Concentrations en Hg total déterminées dans le biogaz du site de Mendixka
Volume Concentration Hg
Quantité Hg dans
Réplicats prélevé Aire Biogaz Mendixka
piège Amasil (pg)
(L) (ng.m-3)
1 0,005 187,6 25,5 5092,1
2 0,005 182,2 24,7 4945,5
1 0,01 359,7 48,8 4881,7
2 0,01 361,2 49,0 4902,1
1 0,02 728,1 98,8 4940,8
2 0,02 719,7 97,7 4883,8
1 0,03 1102,7 149,7 4988,5
2 0,03 1093,4 148,4 4946,4
1 0,04 1478,08 200,6 5015,0
2 0,04 1467,9 199,2 4980,5
1 0,05 1850,5 251,1 5022,9
2 0,05 1797,9 244,0 4880,1
Concentration moyenne en Hg total
4956,6
(ng.m-3)
Erreur (ng.m-3) 63
La concentration en Hg total a d’abord été effectué en déterminant la quantité de celui-

ci retenue dans les pièges Amasil®. La concentration a été estimée en fonction du volume de
biogaz prélevé, ce qui a permis de déterminer que la concentration moyenne de Hg total contenu
dans le biogaz produit à Mendixka est égale à 4956 ± 63 ng.m-3. Le réacteur a donc pu être mis
en place sur le flux primaire de biogaz afin de mieux observer et comprendre l’adsorption du
mercure sur la surface des plaques.
127
2.3. Analyse de surface des plaques représentant expérimentalement les
cylindres d’échantillonnage
2.3.1. Analyse de surface par LA-ICP-MS
Suite à l’analyse de surface par LA-ICP-MS de chaque catégorie de plaques ayant été
utilisées lors de l’échantillonnage du biogaz de Mendixka, une adsorption du soufre, de
l’arsenic ainsi que du mercure présents dans le biogaz a été observée sur chaque catégorie de
plaques. Pour toutes les analyses, il n’y a pas eu de détection de ces trois éléments pendant les
20 premières secondes d’analyse. Ce résultat s’explique par le fait, que le laser chauffe pendant
ces vingt premières secondes et ne scanne donc pas l’échantillon. Le signal obtenu est l’analyse
de ces trois éléments dans le gaz vecteur.
Pour la plaque en acier inoxydable non-revêtue, la plaque en inox sans revêtement, avec
le revêtement Silconert® ainsi que celle avec le revêtement Dursan®, non mises en contact
avec des échantillons de gaz (blancs), contenaient toutes dans leur composition initiale une
infime présence de soufre, une petite quantité d’arsenic (intensité ~ 1000 cps) ainsi que du
mercure (intensité ~ 4000 cps pour la plaque non-revêtue ainsi que la plaque Silconert® et
intensité ~ 1500 cps pour la plaque Dursan®). Ces résultats ont été obtenus après analyse d’une
plaque neuve (blanc) de chaque catégorie plaque.
Lors de l’analyse de la plaque en inox non-revêtue ayant été utilisée lors de
l’échantillonnage du biogaz du Mendixka, l’intensité pour le soufre était ~ 800 fois supérieure
pour la plaque utilisée lors de l’échantillonnage que celle utilisée comme blanc (Figure 23 A.).
Pour l’arsenic, l’intensité était ~ 22,5 fois supérieure pour la plaque utilisée lors de
l’échantillonnage que celle utilisée comme blanc (Figure 23 B.). Enfin pour le mercure, cette
augmentation d’intensité, qui était ~ 3 fois supérieure par rapport au blanc, a aussi été observée.
Ces augmentations d’intensité par rapport aux blancs ont aussi été observées sur la
plaque ayant le revêtement Silconert® ainsi que sur celle ayant le revêtement Dursan®. Pour
le soufre, l’intensité était ~ 1800 fois supérieure pour la plaque Silconert® (Figure 24 A.) et ~
6000 fois supérieure pour la plaque Dursan® (Figure 25 A.). Pour l’arsenic, l’intensité était ~
20 fois supérieure pour la plaque Silconert® (Figure 24 B.) ainsi que pour la plaque Dursan®
(Figure 25 B.). Pour finir, pour le mercure, l’intensité était ~ 3,5 fois supérieure pour la plaque
Silconert® (Figure 24 C.) et ~ 3,3 fois supérieure pour la plaque Dursan® (Figure 25 C.).
128
surface de la plaque en acier inoxidable non-revêtue neuve (blanc) ainsi que celle utilisée lors de
l’échantillonnage du biogaz du site de Mendixka
Ces augmentations d’intensité entre le blanc et les échantillons ont montré que quelle que soit
la catégorie de plaques, il y a une adsorption du soufre, de l’arsenic ainsi que du mercure à la
surface de celles-ci. L’adsorption du mercure et de l’arsenic est un peu moins importante sur la
129
plaque ayant le revêtement Silconert® et ainsi que sur celle ayant le revêtement Dursan® par
rapport à la plaque en inox n’ayant pas de revêtement mais un traitement de surface. Ces
résultats montrent que ces revêtements spécialement conçus pour être inertes à l’adsorption
d’éléments notamment le mercure et l’arsenic, permettent la diminution de ce phénomène mais
ne permettent pas de l’éliminer.
surface de la plaque ayant le revêtement Silconert® neuve (blanc) ainsi que celle utilisée lors de
130
A l’inverse, l’adsorption du soufre était beaucoup plus importante sur les deux plaques
revêtues que sur la plaque sans revêtement. La spéciation du soufre contenue dans le biogaz de
Mendixka n’a pas été effectuée, ne permettant donc pas de déterminer quelle espèce de soufre
s’est adsorbée à la surface de ces plaques. Ces expérimentations ne permettent pas de conclure
sur l’utilisation de cylindres contenant ces revêtements ou non, lors de la recherche de la
présence d’espèces soufrées dans le gaz naturel et le biogaz.
surface de la plaque ayant le revêtement Dursan® neuve (blanc) ainsi que celle utilisée lors de
131
Ces résultats montrent donc que les cylindres en acier inoxydable ou ayant le revêtement
Silcornert® et Dursan® ne sont pas inertes à l’absorption du soufre, de l’arsenic et du mercure.
L’utilisation de ces échantillonneurs ne sont donc pas recommandés pour l’analyse de ces
composés.
Pour finir, lors du premier scan de la plaque Dursan® pour les trois éléments (courbe en
orange sur la Figure 25), un pic élevé d’intensité est observé à ~ 60 secondes. Pour cette même
plaque, le troisième scan de la plaque montre une forte variation de l’intensité du soufre (courbe
en jaune sur la Figure 25 A.).
Figure 26. Photographie de la surface de la plaque ayant le revêtement Silconert® (A.) ainsi que
celle ayant le revêtement Dursan® obtenue par microscope optique (Grossissement : 200 µm)(B.).
132
De manière générale, même si les variations d’intensité sont plus importantes pour la plaque
Dursan®, lors du scan de la plaque non-revêtue ainsi que la plaque Silconert®, il y a tout au
long de l’analyse de plaques, quel que soit l’élément analysé, des variations de l’intensité ont
été observées. Ce phénomène peut être expliqué par la présence de petites « taches » noires
(Figure 26 A.) ou rosées (Figure 26 B.), où les éléments, quelle que soit la catégorie de plaques
semblent s’être davantage adsorbés que sur le reste de la plaque.
Ces petites « taches » observées par microscope optique semblent être des impuretés ou
des microfissures (Figure 26) n’ayant donc pas de revêtement ou de traitement de surface
expliquant la raison de l’adsorption plus importante sur celles-ci que sur le reste de la plaque.
L’analyse par SEM-EDX va donc permettre de mieux comprendre la répartition de l’adsorption
des 3 éléments sur ces plaques et sur ces taches.
2.3.2. Analyse de surface par SEM-EDX
L’analyse par SEM-EDX, réalisée par Mme Svahn au Bordeaux Imaging Center (BIC),
de la plaque ayant le revêtement Silconert ainsi que la plaque ayant le revêtement Dursan
utilisée lors de l’échantillonnage du biogaz du site de Mendixka, a permis de cartographier la
répartition de l’adsorption du soufre, de l’arsenic et du mercure sur plusieurs sites d’analyses
de ces deux plaques.
L’analyse de la plaque Silconert® (Figure 27, Annexe 1 p 142) et de la plaque Dursan®
(Figure 28, Annexe 2 p 145) a permis de mettre en évidence la présence de soufre, d’arsenic et
de mercure sur la surface des sites analysés de ces deux plaques. Ces résultats montrent bien
qu’une adsorption de ces trois éléments s’est opérée à la surface de la plaque. De plus, il y a
présence de soufre, d’arsenic et de mercure sur toute la surface des sites analysés. Ce résultat
suggère donc que l’adsorption de ces trois éléments est diffuse sur la surface interne des
cylindres.
133
Figure 27. Analyse par SEM-EDX de l’arsenic (B.), du soufre (C.) et du mercure (D.) adsorbé sur
la surface du site n°1 (A.) de la plaque Silconert®
Toutefois, l’intensité de l’arsenic analysé sur un des sites de la plaque Silconert®

(Figure 27 A.) est plus importante sur un point précis de ce site analysé (Figure 27 B.). Cet
endroit semble vraisemblablement être une tache similaire à celles observées par microscope
optique (Figure 26 A.). L’analyse d’un des sites de la plaque Dursan® (Figure 28 A.) a été
focalisée sur une de ces « taches ». L’intensité de l’arsenic est plus intense sur celle-ci (Figure
28 B.) que sur le reste du site analysé. Pour le reste des sites analysés, que ce soit pour la plaque
Silconert® (Annexe) ou la plaque Dursan® (Annexe), ce même phénomène est observé.
Pour le mercure ainsi que le soufre, l’intensité de ceux-ci sont également plus importante
sur ces « taches » que sur le reste de la surface des sites analysés que ce soit pour la plaque
Silconert® (Figure 27 C. et D., Annexe) que pour la plaque Dursan® Figure 28 C. et D.,
Annexe).
134
Figure 28. Analyse par SEM-EDX de l’arsenic (B.), du soufre (C.) et du mercure (D.) adsorbés sur
la surface du site n°1 (A.) de la plaque Dursan®
Ces résultats montrent donc que même si ces trois éléments se sont adsorbés sur toute
la surface des deux plaques, il y eu une adsorption plus importante à la surface des « taches »
comme supposé lors de l’analyse de la surface des plaques par LA-ICP-MS ( cf 2.3.1.Analyse
de surface par LA-ICP-MS, p 128). L’adsorption du soufre, de l’arsenic et du mercure est donc
plus importante lorsqu’il y a présence d’impuretés et/ou de microfissures à la surface interne de
ces cylindres.
Pour finir, l’arsenic ne s’est pas adsorbé sur les mêmes impuretés/microfissures que le
mercure et le soufre. A l’inverse, ces deux derniers éléments se sont tous les deux absorbés sur
les mêmes impuretés/microfissures. Ces résultats valident donc l’hypothèse émise dans la partie
« 4.Conclusion and Recommendations, p 90 » sur la capacité des espèces soufrées à s’adsorber
sur la surface interne des cylindres et à devenir des sites d’adsorption pour le mercure.
135
3. Conclusion
L’analyse de la surface par LA-ICP-MS et par SEM-EDX des plaques non-revêtues, des
plaques ayant le revêtement Siclonert® ainsi que des plaques ayant le revêtement Dursan®
utilisées lors de l’échantillonnage de biogaz sur le site de valorisation des déchêts de Mendixka
a permis de confirmer le phénomène d’adsorption du soufre, du mercure et de l’arsenic à leur
surface et par conséquent à la surface interne des cylindres utilisés lors de l’échantillonnage de
gaz naturel. Ces analyses ont permis de déterminer que l’adsorption de ces trois éléments était
diffuse sur toute la surface des cylindres.
De plus, l’analyse des plaques par EDX-SEM a permis de confirmer que l’adsorption
de ces élements était plus importante lorsqu’il y avait présence d’impuretés et/ou de
microfissures sur la surface interne des plaques et donc des cylindres. Pour finir, il a été montré
que le mercure ne s’adsorbe pas directement sur la surface mais c’est le soufre, qui, une fois
adsorbé à la surface des cylindres, devient un site d’adsorption pour le mercure.
136
CONCLUSION GENERALE ET
PERSPECTIVES
Le travail réalisé sur la fiabilité de l’échantillonnage et de l’analyse du mercure et de

l’arsenic dans l’industrie pétrochimique et pétrolière a permis d’identifier que les cylindres
haute pression, avec ou sans revêtement spécial, utilisés pour la prise d’échantillons de gaz
naturel n’étaient pas fiables, la plupart du temps, pour l’analyse de mercure et d’arsenic en
laboratoire, malgré leurs revêtements supposés diminuer l’adsorption de ces deux éléments.
L'inertie des cylindres ayant un revêtement spécial (Silconert® ou Dursan®) était effectivement
initialement efficace concernant l’adsorption du Hg et de l’As, mais cette efficacité est
rapidement compromise après leur première utilisation rendant l’analyse de ces deux éléments
non fiables, lorsque ceux-ci sont échantillonnés dans du gaz naturel à l’aide de ces
échantillonneurs, notamment lors d’un test de puit.
L’adsorption de ces deux éléments à la surface internes des cylindres a été déterminée
comme étant diffuse sur toute la surface de ceux-ci mais plus notablement importante lorsqu’il
y a présence d’impuretés et/ou de microfissures à l’intérieur de ces cylindres. Il a été montré
que le soufre après s’être adsorbé à la surface interne de ces cylindres devient un site
d’adsorption pour le mercure.
L’utilisation de sacs Tedlar lors d’échantillonnages de gaz naturel avant l’analyse du
mercure a été déterminée comme fiable et inerte à l’adsorption du Hg. Malgré cela, cette
utilisation semble problématique lors d’échantillonnages de gaz naturel sur le terrain car ceux-
ci ne sont pas conçus pour échantillonner des gaz sous pression. Une dépressurisation ainsi
qu’une dérivation du flux principal du gaz naturel doit être mise en place, afin de pouvoir
échantillonner du gaz naturel à l’aide de ceux-ci. Ces deux opérations peuvent être la source de
pertes (points froids) ou de contamination et peuvent fausser la concentration en Hg. Les sacs
Tedlar sont également très fragiles et très volumineux une fois remplis, rendant leur transport
très difficile. L’analyse directe sur le terrain du mercure mais aussi de l’arsenic semble donc le
plus recommandé pour limiter les artéfacts et évaluer avec fiabilité leur concentration dans les
fluides pétroliers. L’analyse sur site est d’autant plus recommandée, qu’elle permet
137
l’identification immédiate des artéfacts alors qu’une analyse off-site ne permet pas de constater
les dégâts, dans beaucoup de cas, si des artéfacts sont effectivement observés.
La recherche de nouveaux revêtements inertes de cylindres pour l’échantillonnage de
gaz naturel sous pression lors d’analyses du mercure et de l’arsenic semble donc nécessaire si
ces deux éléments traces ne peuvent pas être déterminés directement sur le terrain. De plus,
l’évaluation de la fiabilité de l’échantillonnage de gaz naturel pour l’analyse de d’autres
éléments, notamment l’oxygène et le soufre peut être aussi intéressante à réaliser.
Les travaux réalisés ont aussi permis de déterminer le lien entre la durée d’un test de
puit, la quantité d’hydrocarbures à produire et le seuil de détection minimal des composés
traces. La spécification concernant le mercure dans un gaz naturel est souvent de l’ordre de 0,01
µg.Nm-3. Il a été démontré dans cette étude, l’impossibilité de déterminer la concentration en
mercure dans un gaz naturel de l’ordre des µg.m-3, sans un test de puit extrêmement long. En
effet, la recherche de mercure à un niveau de quelques µg.Nm-3 peut être réalisable au cours
d’un test mais exigerait une production et donc le brûlage de plusieurs dizaines de millions de
mètres cubes de gaz naturel : ce qui correspondrait au rejet de plus d’une dizaine de milliers de
tonnes de CO2, sans parler du coût financier correspondant au rallongement de la durée du test.
La recherche de la présence de mercure dans le gaz naturel à un niveau plus bas que le µg.Nm-
3
est donc non nécessaire pendant les tests de puits
La recherche d’une teneur en mercure très basse dans un fluide de réservoir, sans passer
par un test de puits très long, serait donc un grand progrès pour les compagnies pétrolières.
Certains laboratoires et compagnies de services développent et testent des échantillonneurs de
fond, à parois inerte, afin d’éviter l’étape de saturation de la surface des équipements, réduisant
ainsi considérablement la durée d’un test de puit. Cependant, pour valider une telle méthode, il
faut d’une part prouver que la rétention de mercure ou de d’autres composés en ultra-traces est
due à l’adsorption sur des surfaces d’équipement seul et que la boue et les particules utilisées
et générées par le forage, ont une capacité de rétention bien inférieures. D’autre part, il faut
trouver une méthode de validation des résultats, ce qui n’est pas chose aisée comme il a pu être
constaté à travers les travaux présentés dans cette thèse.
Pour finir, la production de bio-huiles et de biogaz à partir de biomasses, obtenues par
exemple après pyrolyse ou coliquefaction de celles-ci, semble prendre de plus en plus
d’ampleur afin de remplacer les énergies fossiles. Le mercure et l’arsenic sont présents dans la
plupart des biomasses utilisées pour la production de leurs sous-produits. Ceux-ci restent donc
contaminés en ces deux éléments après leur production. Au vu des problèmes liés à leur
138
présence dans les fluides pétroliers et à la difficulté de déterminer la teneur du Hg et de l’As
dans ceux-ci, il ne paraît pas aberrant de penser que des problématiques similaires seront
observés dans la production de ces nouvelles énergies. Il sera donc primordial d’évaluer la
fiabilité de l’échantillonnage et de l’analyse du mercure et de l’arsenic dans le biogaz ainsi que
dans les bio-huiles.
139
Références (hors publications)
ASTM D5954-98, 2014. Standard Test Method for Mercury Sampling and Measurement in
Natural Gas by Atomic Absorption Spectroscopy, 2014. West Conshohocken, PA,
United States.
ASTM D6350-14, 2014. ASTM International. Standard Test method for Mercury Sampling
and Analysis in Natural Gas by Atomic Fluorescence Spectroscopy, 2014.West
Conshohocken, PA, United States.
Boffetta, P., Merler, E., Vainio, H., 1993. Carcinogenicity of mercure and mercury
compounds. Scand. J. Work. Environ. Health 1–7.
Bouyssiere, B., Baco, F., Savary, L., Garraud, H., Gallup, D.L., Lobinski, R., 2001.
Investigation of speciation of arsenic in gas condensates by capillary gas
chromatography with ICP-MS detection. J. Anal. At. Spectrom. 16, 1329–1332.
https://doi.org/10.1039/b105765k
Cachia, M., 2017. Caractérisation des transferts d’éléments trace métalliques dans une matrice
gaz/eau/roche représentative d’un stockage subsurface de gaz naturel. Université de
Pau et des Pays de l’Adour.
Delgado-morales, W., Zingaro, R.A., Mohan, M.S., 1994. Arsenic in Natural Gas: Analysis
and Characterization of Pipeline Solids by 1H NMR and Other Methods. Int. J.
Environ. Anal. Chem. 57, 313–328. https://doi.org/10.1080/03067319408027463
Filby, R.H., 1994. Origin and nature of trace element species in crude oils, bitumens and
kerogens: implications for correlation and other geochemical studies. Geol. Soc. Spec.
Publ. 203–2019.
Irgolic, K.J., Llger, D., Zingaro, R.A., Spall, D., Puri, B.K., 1991. Determination of arsenic
and arsenic compounds in natural gas samples. Appl. Organomet. Chem. 5, 117–124.
https://doi.org/10.1002/aoc.590050209
ISO 6978.1-2003, 2003. International Standard Organisation (ISO). ISO 6978.1-2003, Natural
gas - Determination of mercury - Part 1: Sampling of mercury by chemisorption on
iodine. ISO: Geneva, Switzerland, 2003.
ISO 6978.2-2003, 2003. International Standard Organisation (ISO). ISO 6978.2-2003, Natural
gas - Determination of mercury - Part 2: Sampling of mercury by amalgmation on
gold/platinum alloy. ISO: Geneva, Switzerland, 2003.
Krupp, E.M., Johnson, C., Rechsteiner, C., Moir, M., Leong, D., Feldmann, J., 2007.
Investigation into the determination of trimethylarsine in natural gas and its
partitioning into gas and condensate phases using (cryotrapping)/gas chromatography
coupled to inductively coupled plasma mass spectrometry and liquid/solid sorption
techniques. Spectrochim. Acta Part B At. Spectrosc. 62, 970–977.
https://doi.org/10.1016/j.sab.2007.07.009
Lambertsson, L., Lord, C.J., Frech, W., Björn, E., 2018. Rapid Dissolution of Cinnabar in
Crude Oils at Reservoir Temperatures Facilitated by Reduced Sulfur Ligands. ACS
Earth Space Chem. https://doi.org/10.1021/acsearthspacechem.8b00096
Leeper, J., 1980. Mercury-LNG’s problem. Hydrocarb. Process 59, 237–240.
Liu, Z.X., Tang, D.Z., Yan, Q.T., Xu, H., Wang, S.Y., Xu, R., Wu, S., 2014. Characteristic of
the Volatile Arsenic in Natural Gas and its Novel Determination Method. Appl. Mech.
140
Mater. 522–524, 1515–1521. https://doi.org/10.4028/www.scientific.net/AMM.522-
524.1515
Norme NF EN 13211, 2001. Association Francaise de Normalisation (AFNOR). NF EN
13211, Qualité de l’Air - Émissions de Sources Fixes - Methode Manuelle ́de
Determination de la Concentration en Mercure Total (Air Quality - Fixed Source
Emissions - Manual Method for Total Mercury Determination). AFNOR: Paris,
France, 2001.
Rumayor, M., Gallego, J.R., Rodríguez-Valdés, E., Díaz-Somoano, M., 2017. An assessment
of the environmental fate of mercury species in highly polluted brownfields by means
of thermal desorption. J. Hazard. Mater. 325, 1–7.
https://doi.org/10.1016/j.jhazmat.2016.11.068
SilcoTek Corporation, 2019. SilcoTek Corporation. Prevent Mercury Loss during Transport
and Storage with SilcoNert, 2000. SilcoTek Corporation: Bellefonte, PA, 2019.
Wilhelm, S.M., Bloom, N., 2000. Mercury in petroleum. Fuel Process. Technol. 63, 1–27.
https://doi.org/10.1016/S0378-3820(99)00068-5
Yeh, S., How, S.W., Lin, C.S., 1968. Arsenical cancer of skin: Histologic study with special
reference to Bowen’s disease. Cancer 21, 312–339. https://doi.org/10.1002/1097-
0142(196802)21:2<312::AID-CNCR2820210222>3.0.CO;2-K
141
Annexes
1. Annexe 1 : Analyse par SEM-EDX du site 2 à 4 de la plaque

Silconert®
142
143
144
2. Annexe 2 : Analyse par SEM-EDX du site 2 à 5 de la plaque
Dursan®
145
146
147
148

Reliability of Mercury and Arsenic Sampling and Analysis

Transféré par

Droits d'auteur :

Formats disponibles

Reliability of Mercury and Arsenic Sampling and Analysis

Transféré par

Informations du document

Description originale:

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Reliability of Mercury and Arsenic Sampling and Analysis

Transféré par

Droits d'auteur :

Formats disponibles

Reliability of mercury and arsenic sampling and analysis

in the petrochemical and oil industry

To cite this version:

HAL Id: tel-04085642

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

L’Université de Pau et des Pays de l’Adour

La fiabilité de l’échantillonnage et de l’analyse du mercure et de

Date de soutenance : 04 juillet 2022

Florence Pannier Présidente du jury Professeur, Université de Pau et des Pays

Table des figures .................................................................................................................................... 9

Figure 1. Schematic presenting the different methodologies for Hg quantification in natural

Figure 4. Diagram of Arsenic Analysis in Natural Gas ........................................................... 63

Figure 6. Initial Hg concentration measured after sampler loading with Hg-contaminated

Figure 7. Gaseous Hg concentration variations over time during stability experiments

Figure 8. Evolution of gaseous Hg concentration in the old silicon-coated cylinder A after

Figure 18. Schéma du système de désorption thermique pour la détermination de l’arsenic

Figure 20. Chromatogramme de l’arsenic du condensat de gaz de (A.) l’échantillon A – 2 et

Table 1. Commercial Instruments Dedicated to Hg Analysis in Gases ................................... 28

Table 4. Summary of Sample Preparation Methods Used fo Arsenic Concentration Analysis

Tableau 6. Concentrations en Hg obtenus après analyses de blancs des pièges en changeant

Tableau 11. Concentration en As désorbé thermiquement des cylindres .............................. 118

AAS Spectromètre d’absorption atomique

Le chapitre 1 de cette thèse est un état de l’art de méthodes d’échantillonnage, de

II - M.E. and A.M. contributed equally to this work.

Mercury is a naturally occurring and problematic component of petroleum products.

2. General Constraints Related to the Sampling and Quantitative

2.1. Natural Gas

The objective of Hg analysis in petroleum products is to assess any risk to downstream

2.1.1.1. Current Knowledge

𝐻2 𝑆 + 𝐹𝑒2 𝑂3 → 2𝐹𝑒𝑂 + 𝑆 0 + 𝐻2 𝑂 (1a)

𝐻𝑔0 + 𝑆 0 → 𝐻𝑔𝑆 (1b)

2.1.1.2. Thermodynamic Considerations

4 𝐻2 𝑆 + 𝐹𝑒2 𝑂3 → 2 𝐹𝑒𝑆2 + 𝐻2 + 3 𝐻2 𝑂 (2)

2.1.2. Implication of Adsorption Problems for New Pipes

2.1.3. Implication of Adsorption Problems for the Sampling System

Sampling procedures must be appropriate to preserve the stability of Hg in liquid

2.2.1. Volatile Hg Loss

2.2.2. Ionic Hg Adsorption on Container Walls

Hg is known to bind to the walls of storage and/or sampling containers/containers.

2.2.3. Precipitation of Hg-Containing Minerals

Particulate Hg is potentially naturally present in condensates. However, no studies have

𝐻𝑔0 + 𝐻𝑔𝐶𝑙2 → 𝐻𝑔2 𝐶𝑙2 (3)

2.2.4. Sample Homogeneity

Any analysis performed on a sample aliquot assumes that the Hg is homogeneously

3.1. Sampling method

3.1.1. Bubbling System

Nippon Instrument WA-5A AAS Au traps, Tedlar bags, gas 1 pg 13 0.1 – 1

WA-5F AFS Au traps, Tedlar bags, gas <0.1 pg 13 0.1 – 1

EMP (3 different instruments) AAS in situ 100 – 10 0.3 – 1

AM-4 AAS Air Monitoring 0.01 2 0.1 – 0.5

LUMEX RA-915+ AAS Direct gas analysis 2 7.5

RA-915M AAS Direct gas analysis 2 7

Light 915 AAS Direct gas analysis 100 3.3

RA-915M + RP-91NG AAS in situ (special NG) 10

AMNG AAS Monitoring for NG 1 - 1000

Jerome 431-X gold film in situ 3000 5 3.2 0.75

Jerome J505 AFS in situ 50 10 - 15 3 1

PSA PSA 10.54X Sampling System + AFS in situ, monitoring (special