Carbon 37 (1999) 13791389

Modification of the surface chemistry of activated carbons

J.L. Figueiredo*, M.F.R. Pereira, M.M.A. Freitas, J.J.M. Orfao

, Faculdade de Engenharia da Universidade do Porto,

Laboratorio
de Catalise
e Materiais, Departamento de Engenharia Qumica
4099 Porto, Portugal
Received 8 September 1998; accepted 18 December 1998

Abstract
A NORIT activated carbon was modified by different chemical and thermal treatments (including oxidation in the gas and
liquid phases) in order to obtain materials with different surface properties. Several techniques were used to characterize
these materials including nitrogen adsorption, chemical and thermal analyses, XPS, TPD and DRIFTS. The results obtained
by TPD agree quantitatively with the elemental and proximate analyses of the oxidized materials, and qualitatively with the
observations by DRIFTS. A simple deconvolution method is proposed to analyse the TPD spectra, allowing for the
quantitative determination of the amount of each functional group on the surface. A multiple gaussian function has been
shown to fit the data adequately, the parameters obtained for each fit matching very well the features observed in the
experimentally determined TPD spectra.
It is shown that gas phase oxidation of the carbon increases mainly the concentration of hydroxyl and carbonyl surface
groups, while oxidations in the liquid phase increase especially the concentration of carboxylic acids. 1999 Elsevier
Science Ltd. All rights reserved.
Keywords: A. Activated carbon; B. Oxidation; C. TPD; D. Surface oxygen complexes

1. Introduction
Carbon materials are finding an increasing number of
applications in catalysis, either as supports for the active
phases, or as catalysts on their own. Their performance is
determined both by their texture and surface chemistry [1].
In the case of activated carbons, the texture may be
adapted to suit the situation by adequate choice of the
activation procedure. In particular, it is possible to prepare
carbons with different proportions of micro, meso and
macropores [2].
On the other hand, the nature and concentration of
surface functional groups may be modified by suitable
thermal or chemical post-treatments. Oxidation in the gas
or liquid phase can be used to increase the concentration of
surface oxygen groups, while heating under inert atmosphere may be used to selectively remove some of these
functions. Carboxyl, carbonyl, phenol, quinone, and lactone groups, have been identified on carbon surfaces [3,4].
A variety of experimental techniques has been used to

*Corresponding author.

characterize these functional groups, such as chemical

titration methods [3,57], temperature-programmed desorption (TPD) [810], X-ray photoelectron spectroscopy
(XPS) [11,12] and infra-red spectroscopy methods (FTIR,
DRIFTS) [8,1315].
The chemical titration methods, such as proposed by
Boehm [3,5], are especially useful when used in combination with other techniques [16]; however, they are not
practical when dealing with small samples. In addition,
these methods fail to account for a large proportion (as
high as 50%) of the total oxygen content of carbon
materials [17].
XPS is a surface technique which will provide an
estimate of the chemical composition of the few uppermost
layers of the material. In order to get further insight into
the nature of the functional groups on the surface, a
reconstruction of the O1s peak can be performed [18]. The
reconstruction of the C1s region is generally more difficult,
not only as a result of the broadness of the peak, but also
because it is necessary to make an assumption concerning
its nature after oxidation, namely if the surface graphitelike structure remains unchanged or not. This will affect
the curve fitting procedure, an asymmetric peak shape

0008-6223 / 99 / $ see front matter 1999 Elsevier Science Ltd. All rights reserved.
PII: S0008-6223( 98 )00333-9

1380

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

being needed for the graphite-like structure, while a

gaussian peak is required for an aliphatic structure [19].
Infra-red spectroscopic methods can be applied only to
highly oxidized carbons, otherwise the intensity of the
absorption bands is not sufficient. Diffuse Reflectance
FTIR (DRIFTS) is preferable, to avoid the problems
caused by sample dilution, and a recent study has shown
the usefulness of the technique to monitor the surface
functionalities that develop under oxidizing conditions
[14]. The interpretation of the spectra is complicated by the
fact that each group originates several bands at different
wavenumbers, therefore each band may include contributions from various groups (cf. Table 1).
Temperature-programmed methods have become rather
popular. Surface oxygen groups on carbon materials
decompose upon heating by releasing CO and CO 2 at
different temperatures. There is some confusion in the
literature with respect to the assignment of the TPD peaks
to specific surface groups, as the peak temperatures may be
affected by the texture of the material, the heating rate and
the geometry of the experimental system used [2022].
However, some general trends can be established as
summarized in Fig. 1. Thus, a CO 2 peak results from
carboxylic acids at low temperatures, or lactones at higher
temperatures; carboxylic anhydrides originate both a CO
and a CO 2 peak; phenols, ethers, and carbonyls (and
quinones) originate a CO peak. In general, the TPD spectra
obtained with carbon materials show composite CO and
CO 2 peaks which must be deconvoluted before the surface
composition can be estimated.
Thus, in general the quantitative analysis of the surface
functional groups is not straightforward. In the present
work, activated carbons were modified by thermal or
chemical post-treatments in order to obtain materials with
different surface properties which were then characterized
by different techniques, in an attempt to determine quantitatively the amount of each chemical function on the
surface. A simple deconvolution method is proposed to

analyse the TPD profiles, and the results obtained seem to

validate this procedure.

2. Experimental
A NORIT ROX 0.8 activated carbon (pellets of 0.8 mm
diameter and 5 mm length) was used as supplied and after
different oxidation treatments: in the gas phase, with 5%
O 2 (in N 2 ) at 698 K for different times in order to achieve
the desired burn-off (B.O.), or with 50% N 2 O (in N 2 ) for 8
h at 773 K; and in the liquid phase, with 5M nitric acid at
the boiling temperature for 6 h, or with 10 M H 2 O 2 at
room temperature. The materials oxidized in the liquid
phase were subsequently washed with distilled water until
neutral pH, and dried in an oven at 383 K for 24 h. The
following samples were prepared: A1-original carbon; A2
A1 oxidized with 5% O 2 for 3 h at 698 K (B.O.58.7%);
A3A1 oxidized with 5% O 2 for 6 h at 698 K (B.O.5
12.5%); A4A1 oxidized with 5% O 2 for 10 h at 698 K
(B.O.521.5%); A5A1 oxidized with 5% O 2 for 20 h at
698 K (B.O.541.5%); A6A4 treated under nitrogen at
873 K for 1 h; A7A4 treated under nitrogen at 1023 K for
1 h; A8A4 treated under nitrogen at 1373 K for 1 h;
A9A1 oxidized with 50% N 2 O for 8 hours at 773 K
(B.O.521%); A11A1 oxidized with 5M nitric acid for 6
h at boiling temperature; A12A1 oxidized with 10 M
H2O2.
The textural characterization of the materials was based
on the N 2 adsorption isotherms, determined at 77 K with a
Coulter Omnisorp 100 CX apparatus. The micropore
volumes and mesopore surface areas were determined by
the t-method, using the standard isotherm for carbon
materials proposed by Reinoso et al. [27]. The adsorption
data were also analysed with the Dubinin equation. When a
type IV deviation occurred, two microporous structures
were taken into account, and the corresponding volumes,
W01 and W02 , were calculated [28]. The Stoeckli equation

Table 1
IR assignments of functional groups on carbon surfaces (adapted from Ref. [14])
Group or functionality

Assignment regions (cm 21 )

10001500

CO in ethers (stretching)
Alcohols
Phenolic groups:
COH (stretching)
OH
Carbonates; carboxyl-carbonates
C5C aromatic (stretching)
Quinones
Carboxylic acids
Lactones
Carboxylic anhydrides
CH (stretching)

15002050

10001300
10491276
10001220
11601200
11001500

11201200
11601370
9801300

20503700
32003640

25003620
15901600
15851600
15501680
16651760
16751790
17401880

25003300

26003000

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

1381

Fig. 1. Surface groups on carbon and their decomposition by TPD [9,10,18,2326].

[29] was used to estimate the average micropore width of

the smaller pores, using an affinity coefficient for nitrogen
b 50.34.
The DRIFTS spectra were obtained without sample
dilution in a Nicolet 510P spectrophotometer equipped
with a diffuse reflectance attachment, by adding 256 scans
at a resolution of 4 cm 21 . The samples were finely ground
in a mortar for this analysis. Sample A8, where all surface
groups were removed, was used as background; in this
way, it was possible to obtain a better base-line correction
and spectra with increased signal-to-noise ratio [30].
XPS analysis was performed at CEMUP (Centro de
Materiais da Universidade do Porto) with a VG Scientific
ESCALAB 200A spectrometer utilizing a non-monochromatized Mg Ka radiation (1253.6 eV). The vacuum in
the analysis chamber was always ,1310 27 Pa.
The TPD profiles were obtained with a custom built
set-up, consisting of a U-shaped tubular micro-reactor,
placed inside an electrical furnace. The mass flow rate of
the helium carrier gas (69 mg / s) and the heating rate of the
furnace (5 K / min) were controlled with appropriate units.
The amounts of CO and CO 2 desorbed from the carbon
samples (0.1 g) were monitored with a SPECTRAMASS
Dataquad quadrupole mass spectrometer.
The materials were further characterized by elemental

and proximate analyses, with a CARLO ERBA EA 1108

Elemental Analyser and a Mettler TA 4000 thermal
analyser, respectively.

3. Results and discussion

The results obtained for all samples by the various
characterization techniques are shown in Tables 26, and
Table 2
Textural properties of all materials
Sample

Vmicro
(cm 3 / g)

Sme
(m 2 / g)

W01
(cm 3 / g)

W02
(cm 3 / g)

L1
(nm)

A1
A2
A3
A4
A5
A6
A7
A8
A9
A11
A12

0.372
0.442
0.497
0.537
0.575
0.564
0.557
0.546
0.592
0.393
0.388

126
150
149
167
173
186
178
164
217
146
157

0.347
0.403
0.450
0.459
0.445
0.543
0.536
0.514
0.522
0.364
0.355

0.029
0.036
0.045
0.083
0.146
0.028
0.028
0.034
0.077
0.030
0.034

0.91
1.1
12
1.4
1.6
1.2
1.3
1.1
1.3
0.91
0.94

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

1382

Table 3
Proximate and elemental analyses of all materials
Sample

Proximate analysis

A1
A2
A3
A4
A5
A6
A7
A8
A9
A11
A12

Elemental analysis

Moisture
(wt%)

Volatiles
(wt%)

C fixed
(wt%)

Ash
(wt%)

C
(wt%)

H
(wt%)

N
(wt%)

O
(wt%)

3.91
1.06
1.89
1.77
5.19
2.17
0.59
1.83
1.78
3.26
2.64

6.76
10.13
13.95
17.75
19.50
12.81
5.82
3.19
14.80
7.30
7.93

86.53
85.59
81.21
77.72
71.51
81.52
90.10
90.98
79.44
87.52
87.47

2.80
3.26
2.95
2.75
3.81
3.51
3.49
3.99
3.98
1.92
1.96

95.23
92.77
90.96
89.72
86.34
92.72
96.07
98.10
87.88
85.79
94.21

0.40
0.25
0.28
0.29
0.31
0.35
0.42
0.20
0.60
0.87
0.40

0.48
0.41
0.45
0.64
0.54
0.65
0.54
0.73
2.61
1.06
0.54

3.89
6.56
8.31
9.34
12.82
6.27
2.96
0.97
8.90
12.28
4.86

Table 4
XPS results for the O1s region, values given in % of the total amount
Sample

Binding energy (eV)

A1
A2
A3
A4
A5
A6
A7
A8
A9
A11
A12
a

531.1 a

532.3 a

533.3 a

534.2 a

1.90
2.57
3.41
4.605
4.86
2.74
1.69
0.86
4.26
3.87
2.22

2.08
1.38
1.66
2.18
3.66
2.31
1.46
1.34
1.36
5.04
3.86

1.66
3.38
4.49
6.06
6.99
3.69
1.25
1.11
5.93
3.41
4.30

0.81
0
0
0
0
0
0
0
0
1.05
2.00

O total

C total

Ot / Ct

6.46
7.33
9.55
12.83
15.51
8.74
4.40
3.31
11.55
13.37
12.38

93.51
92.67
90.45
87.17
84.51
91.26
95.60
96.68
85.08
86.63
87.62

n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
3.37
n.d
n.d.

0.069
0.079
0.106
0.147
0.183
0.096
0.046
0.034
0.136
0.154
0.141

Oxygen atoms of the relevant surface functional groups as shown in Fig. 1.

will be analysed and discussed in three stages: (a) General

characterization, with emphasis on the different oxidation
methods (samples A1, A4, A9, A11, A12); (b) Heattreatments at different temperatures (samples A4, A6, A7,
A8); and (c) Different degrees of oxidation by the same
treatment (samples A1, A2, A3, A4, A5).
Table 5
Amounts of CO and CO 2 released, obtained by integration of the
areas under the TPD peaks of the ROX activated carbon before
and after different oxidative treatments
Sample

CO 2 (mmol / g)

CO (mmol / g)

CO / CO 2

A1
A2
A3
A4
A5
A9
All
A12

156
214
265
351
534
200
1104
420

743
3254
3919
4806
6252
3536
2702
1321

4.75
15.2
14.8
13.7
11.7
17.7
2.45
3.15

3.1. Different oxidation treatments

3.1.1. Textural characterization
The textural properties of all materials are collected in
Table 2, while some of the nitrogen adsorption isotherms
are shown in Fig. 2. It may be concluded that the liquid
phase treatments have no significant impact on the texture
of the carbon materials (A11 and A12 vs. A1), while gas
phase oxidation increases the micropore volume, the
Table 6
Amounts of CO and CO 2 released, obtained by integration of the
areas under the TPD peaks of the carbons before (sample A4) and
after heat treatments at different temperatures (samples A6A8)
Sample

CO 2 (mmol / g)

CO (mmol / g)

CO / CO 2

A4
A6
A7
A8

351
88.9
67.6
35.4

4806
3017
581
133

13.7
33.9
8.59
3.77

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

1383

treated samples (A6, A7, A8), a gradual decrease in

volatile matter and oxygen content is observed as the
temperature increases.

Fig. 2. Nitrogen adsorption isotherms at 77 K on the ROX

activated carbon before and after different oxidative treatments.

mesopore surface area and the average width of the

micropores (A4 and A9 vs. A1). These changes increase as
the degree of oxidation increases (A1 vs. A2, A3, A4 and
A5). In particular, the volume of the larger micropores
(W02 ) increases remarkably for higher degrees of B.O. (A4,
A5). The samples which were submitted to thermal
treatments under inert atmosphere (A6, A7, A8) do not
show significant textural differences with respect to the
parent material, A4.
It is generally reported that the liquid phase oxidations
do not change significantly the texture of the activated
carbons [7], although under more drastic conditions (concentrated acid, heating until complete evaporation) a
decrease in surface area and pore volume has been
observed while the average micropore width increases, due
to the collapse of the pore walls [31]. In the gas phase
oxidations the pore volume increases with burn-off, but it
may be expected that, for higher degrees of oxidation,
there will be a decrease first in the micropore volume and
subsequently in the mesopore volume, due to the collapse
of the pores [32].

3.1.2. Proximate and elemental analyses

The proximate and elemental analyses of the carbon
samples before and after different oxidative treatments are
shown in Table 3. It is apparent that the gas phase
treatments lead to a substantial increase in the volatile
matter content (samples A4, A9), while the liquid phase
treatments originate only a modest increase in this parameter (A11, A12). Since the volatiles include the products
released by the surface groups that decompose up to 1223
K, it is obvious that the gas phase treatments are the most
effective. This is accompanied by a slight increase in ash
content (due to carbon burn-off), whereas some ash is
removed by the liquid phase treatments (which, by the
way, do not cause any significant carbon burn-off). The
increase in oxygen content is also observed by elemental
analysis, although the result for A11 seems suspicious. The
increase in nitrogen content is noticeable for samples A11
and especially A9. When comparing A4 with the heat-

3.1.3. Surface analysis by XPS

Table 4 shows the results obtained by XPS. The
amounts determined for C, O, N were calculated from the
corresponding peak areas divided by the appropriate
sensitivity factors (1.000 for C, 2.850 for O and 1.770 for
N [33]). The ratio of total oxygen to total carbon, O t / C t ,
which indicates the degree of surface oxidation, is also
included. The reconstruction of the O1s peak gives additional information on the nature of the surface oxygen
groups. The functionalities shown in Fig. 1 will contribute
with peaks located at 531.1 eV (1), 532.3 eV (2), 533.3 eV
(3) and 534.2 eV (4), in addition to chemisorbed water, at
535.9 eV [18]. The peaks were deconvoluted by using a
sum of Lorentzian-Gaussian functions, fixing the width at
half-height equal to 2.0 eV and allowing the position of the
peak center to vary within 60.2 eV of the reported value
[34].
All treated samples show an increase in the oxygen
content. It is observed that the gas phase treatments (A4
and A9) increase mainly the groups (1) and (3), carbonyl /
quinone and cetoenol, respectively, while the liquid phase
treatments (A11 and A12) increase groups (2) and (4),
hydroxyl1anhydrides and carboxylic acids, respectively.
By comparing A4 with the heat-treated samples (A6, A7,
A8), one can observe the gradual decrease in all types of
surface oxygen groups. However, the total oxygen content
of A8 is too high, probably reflecting the effect of surface
re-oxidation by exposure to the atmosphere during sample
handling.
3.1.4. TPD
Fig. 3 shows the TPD spectra of the ROX activated
carbon before and after the different oxidative treatments.
In all cases, an increase in the amount of surface oxygen
groups is evidenced by the increase of the CO and CO 2
peaks. The CO 2 spectra are quite distinct: Thus, the liquid
phase treatments increase the CO 2 evolution at low
temperatures (from 373 to 673 K), while oxidation in the
gas phase induces a CO 2 peak at higher temperatures. In
the later case, the small peak at low temperature shown by
the original carbon is no longer observed, as a result of the
higher temperature used in this treatment (698 K). The
TPD peaks can be tentatively assigned to the different
functional groups by comparison with the data summarized
in Fig. 1, taken from the available literature. The first CO 2
peak may be undoubtedly attributed to carboxylic acid
functions, while the higher temperature CO 2 peak shows a
shoulder at about 820 K and a maximum at about 930 K,
which may result from carboxylic anhydrides and lactones.
The gas phase treatment gives much more CO than the
nitric acid treatment, but in both cases the TPD composite
peaks show two maxima at around 900 K and 1070 K,

1384

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

Fig. 4. DRIFTS spectra recorded for one of the carbons before

(sample A4) and after heat treatments at different temperatures
(samples A6 and A7).

carbons before (sample A4) and after heat treatments at

different temperatures (samples A6, A7 and A8), while the
corresponding TPD spectra are shown in Fig. 5.
Table 1 shows the infra-red bands which can be
associated with different oxygen functional groups on
carbon materials [14]. Inspection of Fig. 4 shows that there

Fig. 3. TPD spectra before and after different oxidizing treatments: (a) CO evolution; (b) CO 2 evolution.

which may originate from phenols, ethers and carbonyls /

quinones. Table 5 shows the amounts of CO and CO 2
released, obtained by integration of the areas under the
TPD peaks, together with the ratio CO / CO 2 . This ratio
increases significantly after the gas-phase treatments (A4,
A9), and decreases slightly after the liquid-phase treatments (A11, A12).
These results are consistent with an increase in carboxylic acid groups after nitric acid treatment, whereas gas
phase oxidation increases anhydride, lactone, phenol and
carbonyl / quinone surface groups. Both the TPD results
and the proximate analyses show an increase in surface
groups according to the sequence A1,A12,A11,A9,
A4. The same is observed by elemental analysis, with the
exception of A11, as noted above.
The XPS results are not directly comparable with the
other techniques, since only the uppermost layers (1015
nm) are analysed. The results suggest that the liquid phase
treated materials may be more oxidized on the surface than
in the bulk.

3.2. Heat-treated samples

Fig. 4 shows the DRIFTS spectra recorded for one of the

Fig. 5. TPD spectra for one of the carbons before (sample A4)
and after heat treatments at different temperatures (samples A6
A8): (a) CO evolution; (b) CO 2 evolution.

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

1385

are three bands of interest: one at 1750 cm 21 , associated

with C5O stretching in lactones and carboxylic anhydrides, one at 1600 cm 21 associated with quinone and
ceto-enol groups [35], and a broad band centered around
1250 cm 21 associated with CO stretching in ethers,
lactones, phenols and carboxylic anhydrides.
The band at 1750 cm 21 , associated with carboxylic
anhydrides and lactones, disappears when heating to 873 K
(A6); the two other bands decrease in sample A7.
Inspection of Fig. 5 shows that the shoulder in the CO
peak disappears in sample A6, while in the CO 2 spectrum
only a small peak is left at high temperatures. Sample A7
does not yield any CO 2 , and only a small peak is left in the
CO spectrum. All groups have disappeared from sample
A8. Table 6 gives the CO and CO 2 contents of these
samples determined from the TPD spectra.
By coupling together these observations (Figs. 4 and 5),
it may be concluded that heating up to 873 K removes the
carboxylic anhydrides and most of the lactones; some
phenol groups are also removed, as shown by the decrease
in intensity of the characteristic DRIFTS bands and by the
decrease in the CO TPD peak. At 1023 K the lactones have
been completely removed (no CO 2 peak), together with
phenols and some carbonyls which reflect in the decrease
of the bands at 1250 and 1600 cm 21 . Some carbonyls and
ethers which were left are completely removed at 1373 K,
as shown both by the absence of a CO peak in the TPD,
and any bands in DRIFTS.
Fig. 6. TPD spectra for the samples oxidized in the gas phase (5%
O 2 in N 2 ) to different extents: (a) CO evolution; (b) CO 2
evolution.

3.3. Samples oxidized in the gas phase to different

extents
The TPD spectra obtained for samples A1, A2, A3, A4
and A5 are shown in Fig. 6, showing a gradual increase
both in CO and CO 2 released as the extent of oxidation
increases. However, the CO peak increases much more
than the CO 2 , leading to a very large increase in the
CO / CO 2 ratio (cf. Table 5): a three-fold increase is
observed in going from the original sample (A1) to the
sample with the lower degree of oxidation (A2), decreasing
slowly thereafter (A2 to A3, A4, A5). Similar behaviour
was reported by Otake and Jenkins [9] with an activated
carbon obtained from a phenol-formaldehyde resin.
It is also interesting to consider the changes observed in
the shape of the CO peak as function of the degree of
oxidation of the carbon (Fig. 6a). In effect, the high
temperature shoulder is more prominent at low burn-off
(sample A2) but the low temperature shoulder increases
with the extent of oxidation and becomes the most intense
in sample A5. This can be interpreted in terms of the
mechanism for carbon gasification proposed by Zhuang et
al. [36]
C f 1 O 2 carbonyls, ethers( 1 CO)

(1)

C f 1 O 2 lactones, anhydrides

(2)

C f 1 carbonyls, ethers 1 O 2 CO 2 1
CO or carbonyls, ethers

(3)

C f 1 carbonyls, ethers 1 O 2 lactones, anhydrides

(4)

lactones, anhydrides CO 2 ( 1 CO) 1 2C f

(5)

carbonyls, ethers CO 1 C f

(6)

where C f is a free active site on the carbon surface. The

phenol groups, not considered by Zhuang et al. [36],
should be included in these reactions. Reactions (1) and
(3) were assumed predominant. Reactions (5) and (6),
corresponding to the decomposition of the surface groups,
should be negligible at the oxidation temperature used
(698 K).
Reactions (1) and (3) are much more important than (2)
and (4), since the amount of CO released by TPD (reaction
6) is much larger than the CO 2 (reaction 5) indicating that
there are much more carbonyl, ether and phenol groups
(reaction 1 and 3) than lactones and anhydrides (reactions
2 and 4). In effect, Table 5 shows that the ratio CO / CO 2 is
always higher than 10.

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

1386

Table 7
Comparison between the oxygen concentrations (wt%) obtained
by different techniques
Sample

Elem. anal.

TPD

XPS

A2
A3
A4
A5

6.56
8.31
9.34
12.8

6.63
8.01
9.91
13.2

9.54
12.3
16.4
19.7

The shape of the CO peak can be analysed in terms of

this mechanism as follows:
At low degrees of oxidation, the shoulder at 1100 K
(carbonyl, ether) is higher than the shoulder at 900 K
(phenol, anhydrides), indicating that reaction (1) prevails.
However, as the concentration of carbonyl, ether and
phenol groups increases on the surface, reaction (4)
becomes favoured. This originates anhydrides, one of the
components of the CO peak at 900 K. Thus, for high
degrees of oxidation, there is a larger increase in the 900 K
shoulder with respect to the 1100 K shoulder.
It is particularly instructive to compare the results
obtained by TPD with those of the other techniques for
samples A2, A3, A4 and A5, which have all been
processed at the same temperature (698 K). It may be
observed from Tables 7 and 8 that both the concentration
of oxygen obtained by elemental analysis, and the concentration of volatiles determined by proximate analysis,
correlate quite well with the TPD results. Thus, the volatile
matter of these samples corresponds to the amounts of CO
and CO 2 desorbed. The XPS results for oxygen are higher,
indicating that the surface is more oxidized than the bulk.
In order to determine the amount of each surface group,
the deconvolution of the CO and CO 2 TPD spectra of the
gas phase oxidized carbons was based on the following
assumptions, suggested by the results reported above: (a)
The CO 2 spectra are decomposed into two contributions,
corresponding to carboxylic anhydrides and lactones.
There is not any low temperature CO 2 evolution, since
these samples were processed at a temperature higher than
the decomposition of carboxylic acids. (The small peaks
shown by A2 and A3 may result from exposure to the
atmosphere [23]). Although most authors do not distinguish between lactones and anhydrides, quoting the same
decomposition temperature for them [10,18], in this work
it is assumed that the first component corresponds to
Table 8
Comparison between the volatile matter (proximate analysis) and
the amounts of CO1C0 2 released by TPD (expressed in wt%)
Sample

Volatiles

CO1CO 2

A2
A3
A4
A5

10.1
14.0
17.8
19.5

10.0
12.1
15.0
19.9

anhydrides (CO 2 peak [1) and the second corresponds to

the lactones (CO 2 peak [2).
(b) The carboxylic anhydrides decompose by releasing
one CO and one CO 2 molecule [37]. Thus, the first
component in the CO spectra corresponds to the carboxylic
anhydrides, with a peak of the same shape and equal
magnitude of the CO 2 peak [1 (CO peak [1). This peak
is defined from the deconvolution of the CO 2 spectra.
(c) In addition to the carboxylic anhydrides (CO peak
[1), the CO spectra include contributions from phenols
(CO peak [2) and carbonyl / quinones (CO peak [3) [18].
This sequence of decomposition temperatures, first
phenols, then carbonyl / quinones, is justified by the higher
stability of the later groups.
(d) It is assumed that secondary reactions are negligible.
For instance, the Boudouard reaction CO(g)1C[O]5C f 1
CO 2 (g), where C f is a free active site and C[O] is a
surface oxygen complex, would lead to the release of CO 2
at about 1000 K at the expense of CO [38,39]. The absence
of such reactions under the experimental conditions used in
the present work was confirmed by performing the TPD at
different flow rates of the carrier gas.
(e) A multiple gaussian function was used for fitting
each of the TPD spectra, taking the position of the peak
center as initial estimate. The numerical calculations were
based on a non-linear routine which minimized the square
of the deviations, using the Simplex method to perform the
iterations. The use of gaussian functions is justified by the
shape of the TPD peaks, which are a result of a continuous
random distribution of binding energies of the surface
groups [40].
(f) In the CO 2 spectra, the width at half-height was
taken as the same for both peaks. This restriction is needed
when the spectra do not present definite shoulders, such as
sample A5 (in which case one single peak would describe
the situation well). In the other cases, the widths at
half-height obtained when the restriction is removed are
quite similar, which justifies the procedure. This restriction
does not affect significantly the amounts obtained for the
CO components [2 and [3.
This deconvolution procedure proved to fit the data quite
well, as shown in Fig. 7 for the CO and CO 2 TPD spectra
of samples A2, A3, A4 and A5.
Table 9 gives the results obtained, where T M is the
temperature of the peak maximum, W is the width of the
peak at half-height and A is the integrated peak area. It can
be observed that all the fits, for carbons oxidized to
different extents and containing quite different amounts of
surface groups, give approximately the same values of T M
and W for each component; moreover, the T M values
correspond closely to the peak maxima observed in the
TPD spectra, and are within the range of values reported in
the literature (cf. Fig. 1).
Thus, the decomposition temperatures found were circa
820 K for anhydrides, 940 K for lactones, 905 K for
phenols and 1080 K for carbonyls / quinones.

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

1387

Fig. 7. Results of the deconvolution of TPD spectra using a multiple Gaussian function. (a) Sample A2, (b) Sample A3, (c) Sample A4, (d)
Sample A5. (j, TPD experimental data; - - -, individual peaks; , sum of individual peaks).

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

1388

Table 9
Results of the deconvolution of TPD spectra using a multiple Gaussian function
Sample

CO 2

CO

Peak [l

A2
A3
A4
A5

Peak [2

Peak [1

Peak [2

Peak [3

TM
(K)

W
(K)

A
(mmol/g)

TM
(K)

W
(K)

A
(mmol/g)

TM
(K)

W
(K)

A
(mmol/g)

TM
(K)

W
(K)

A
(mmol/g)

TM
(K)

W
(K)

A
(mmol/g)

810630
812653
82369
836617

172620
143635
11868
128611

83.2637
90.5685
127621
248683

949624
929632
93666
941615

172617
143624
11866
128611

104637
149685
192621
287683

810
812
823
836

172
143
118
128

83.2
90.5
127
248

89264
90463
90763
91267

14764
14063
14564
14567

952656
1321669
16436128
26946352

107863
110763
107364
1064610

18664
18964
19865
186611

2174657
2694670
32636130
34746355

In a recent publication, Calo et al. [37] studied the

influence of hydrogen on the TPD of surface complexes on
activated carbon. They considered that the CO 2 peak
results only from anhydrides, and therefore subtracted it
totally from the CO peak. In spite of this difference, these
authors found for CO peak [2 T M 5955 K and W5144 K,
and for CO peak [3 T M 51100 K and W5134 K.
Considering that a higher heating rate was used by these
authors (10 K / min), the agreement with the present results
is quite reasonable.

4. Conclusions
It has been shown that gas phase oxidation of activated
carbons increases mainly the concentration of hydroxyl
and carbonyl surface groups, while oxidation in the liquid
phase increases especially the concentration of carboxylic
acids.
In terms of total oxygen content, the results obtained by
TPD agree quite well with the elemental analyses of the
oxidized materials. Higher amounts of oxygen are determined by XPS, indicating that the concentration of the
functional groups is higher at the surface of the particles.
The results obtained by TPD agree well with the qualitative observations by DRIFTS.
The amounts of the different surface groups can be
estimated by deconvolution of the TPD spectra. It has been
shown that a multiple gaussian function adequately fits the
data. Moreover, the parameters obtained for each fit
correspond to the features observed in the experimentally
determined TPD spectra.

Acknowledgements
This work was carried out with support from the
PRAXIS XXI Programme (contract numbers 2 / 2.1 / BIO /
34 / 94 and PCEX / C / QUI / 98 / 96). M.F.R. Pereira and
M.M.A. Freitas acknowledge the grants received under the
same programme. The authors are indebted to Dr. Carlos

M. Sa (CEMUP) for assistance with XPS analysis, and to

NORIT N.V., Amersfoort, The Netherlands, for providing
the activated carbon.

References
[1] Rodriguez-Reinoso F. Carbon 1998;36:159.
[2] Wigmans T. In: Figueiredo JL, Moulijn JA, editors, Carbon
and Coal Gasification, Dordrecht, The Netherlands: Martinus
Nijhoff Publishers, 1986, pp. 559599.
[3] Boehm HP. In: Eley DD, Pines H, Weisz PB, editors,
Advances in Catalysis, Vol. 16, New York: Academic Press,
1966, pp. 179274.
[4] Donnet JB. Carbon 1968;6:161.
[5] Boehm HP, Diehl E, Heck W. In: Proc. 2nd London Carbon
and Graphite Conf., London, UK, 1965, p. 369.
[6] Bandoz TJ, Jagiello J, Schwarz JA. An Chem 1992;64:891.
[7] Noh JS, Schwarz JA. Carbon 1990;28:675.
[8] Vinke P, Van der Eijk M, Verbree M, Voskamp AF. Carbon
1994;32:675.
[9] Otake Y, Jenkins RG. Carbon 1993;31:109.
[10] Zhuang Q-L, Kyotany T, Tomita A. Energy and Fuels
1994;8:714.
[11] Park SHP, McClain S, Tian ZR, Suib SL, Karwacki C. Chem
Mater 1997;9:176.
[12] Biniak S, Szymanski G, Siedlewski J, Swiatkowski A.
Carbon 1997;35:1799.
[13] Ishizaki C, Marti I. Carbon 1981;19:409.
[14] Fanning PE, Vannice MA. Carbon 1993;31:721.
[15] Zawadzki J. In: Thrower PA, editor, Chemistry and Physics
of Carbon, Vol. 21, New York: Marcel Dekker, Inc, 1988, pp.
147386.
[16] Papirer E, Dentzer J, Li S, Donnet J. Carbon 1991;29:69.
[17] Boehm HP. High Temperatures-High Pressures 1990;22:275.
[18] Zielke U, Huttinger KJ, Hoffman WP. Carbon 1996;34:983.
[19] Wiertz V, Bertrand P. In: Proceedings of the International
Conference on Polymer-Solid Interfaces, ICPSI-2, Namur,
Belgium, 1996.
[20] Falconer JL, Schwarz JA. Catal Rev-Sci Eng 1983;25:141.
[21] Boehm HP, Bewer G. In: Proc. 4th London Carbon and
Graphite Conf, London, UK, 1974, p. 344.
[22] Boehm HP. Carbon 1994;32:759.

J.L. Figueiredo et al. / Carbon 37 (1999) 1379 1389

[23] Zhuang Q-L, Kyotany T, Tomita A. Carbon 1994;32:539.
[24] Zhuang Q-L, Kyotany T, Tomita A. Ext. Abstracts Carbon
94, Granada, Spain, 1994, p. 466.
[25] Marchon B, Carrazza J, Heinemann H, Somorjai GA. Carbon
1988;26:507.
[26] Driel J. In: Capelle A, de Vooys F, editors, Activated Carbon
A Fascinating Material, The Netherlands: Norit, Amersfoort, 1983, pp. 4057.
[27] Rodriguez-Reinoso F, Martin-Martinez JM, Prado-Burguete
C, McEnaney B. J Phys Chem 1987;91:515.
[28] Linares-Solano A. In: Figueiredo JL, Moulijn JA, editors,
Carbon and Coal Gasification, Dordrecht, The Netherlands:
Martinus Nijhoff Publishers, 1986, pp. 13778.
[29] Stoeckli HF, Ballerini L, De Bernardini S. Carbon
1989;27:501.
[30] Venter JJ, Vannice MA. Carbon 1988;26:889.
F, Maldonado-Hodar

[31] Moreno-Castilla C, Carrasco-Marn

FJ,
Rivera-Utrilla J. Carbon 1998;36:145.

[32] Rodrguez-Reinoso
F. In: Lahaye J, Ehrburger P, editors,
Fundamental Issues in Control of Carbon Gasification Reactivity, Dordrecht, The Netherlands: Kluwer Academic Publishers, 1991, pp. 533571.
[33] Ikeo N, Iijma Y, Niimura N, Sigematsu M, Tazawa T,

[34]
[35]

[36]
[37]

[38]

[39]
[40]

1389

Matsumoto S, Kojima K, Nagsawa Y. Handbook of X-ray

Photoelectron Spectroscopy, JEOL, 1991.
Desimoni E, Casella GI, Morone A, Salvi AM. Surface and
Interface Analysis 1990;15:627.
Lien-Vien D, Colthup NB, Fateley WG, Grasselli JG. In: The
Handbook of Infrared and Raman Characteristic Frequencies
of Organic Molecules, S. Diego, Toronto: Academic Press,
Inc, 1991, pp. 11754, Chapter 9.
Zhuang Q-L, Kyotany T, Tomita A. Energy and Fuels
1995;9:630.
D, Linares-Solano A, Roman
Calo JM, Cazorla-Amoros

Martinez MC, Salinas-Martnez

De Lecea C. Carbon
1997;35:543.
Hall PJ, Calo JM, Lilly WD. In: McEnaney B, Mays TJ,
editors, Proceedings of the International Conference on
Carbon, Carbon 88, Bristol, UK: IOP Publishing Co, 1988,
p. 77.
Hall PJ, Calo JM. Energy and Fuels 1989;3:370.
Calo JM, Hall PJ. In: Lahaye J, Ehrburger P, editors,
Fundamental Issues in the Control of Carbon Gasification
Reactivity, Dordrecht, The Netherlands: Kluwer Academic
Publishers, 1991, pp. 329368.