Carbon 42 (2004) 1351–1355

www.elsevier.com/locate/carbon

The Wheeler–Jonas equation: a versatile tool for the prediction

of carbon bed breakthrough times
a,*
P. Lodewyckx , G.O. Wood b, S.K. Ryu c

a
Department of Chemistry, Royal Military Academy, Renaissancelaan 30, B-1000 Brussels, Belgium
b
Industrial Hygiene & Safety Group, Los Alamos National Laboratory, MS K486, Los Alamos, NM 87545, USA
c
Department of Chemical Engineering, Chungnam National University, 305-764 Taejon, South Korea
Received 10 July 2003; accepted 28 December 2003

Available online 10 February 2004

Abstract
This paper gives an overview of the recent developments in the use of the Wheeler–Jonas equation. Extensive experimental work
has been done by measuring breakthrough times of different types of activated carbon beds, under different experimental conditions,
for a large variety of gases and vapours. This includes the use of activated carbon fibre beds, the presence of moisture on the carbon
and in the air stream, non-constant flow patterns and adsorption of chemisorbed species. In all cases the applicability of the
Wheeler–Jonas has been demonstrated, i.e. one can use this equation to extrapolate single laboratory breakthrough results by simply
varying the independent variables of the equation (amount of adsorbent, flow rate, inlet and breakthrough concentrations). In most
cases it is even possible to perform ab initio breakthrough calculations for a well-defined carbon bed. To achieve this new supporting
equations had to be derived to allow the estimation of the dependent variables, We (the equilibrium adsorption capacity) and kv (the
overall mass transfer coefficient), under different circumstances. In conclusion, the scope of the Wheeler–Jonas (or Reaction Kinetic)
equation extends largely beyond its commonly accepted boundaries. This is primarily due to its apparent simplicity: the combination
of a single capacity term and an overall kinetic effect strongly enhances its applicability to different adsorption circumstances. In this
way it is far more potent than many of the more modern equations that require the exact knowledge of several, not readily available,
input parameters.
Ó 2004 Elsevier Ltd. All rights reserved.

Keywords: A. Activated carbon; Carbon fibers; D. Adsorption properties; Transport properties

1. Introduction 2. Theoretical

Over the last few decades several authors groups have 2.1. General
been using the Wheeler–Jonas (a.k.a. the Reaction Ki-
netic) equation to estimate breakthrough times of filter The Wheeler–Jonas equation [1,2] (see Eq. (1)) permits
canisters filled with granular activated carbon (GAC) the estimation of the breakthrough time of a granular
against organic vapours. Recently several groups, activated carbon (GAC) bed, based solely on measurable
including the authors of this paper, have successfully and readily available macroscopic parameters:
extended the use of this equation to other types of

adsorbents. They have also broadened the scope of M  We We  qb cin  cout
tb ¼  ln ; ð1Þ
applicability of the equation to other vapours and other Q  cin kv :cin cout
environmental circumstances. This paper gives an
where tb is the breakthrough time to reach cout (min), M
overview of these recent developments in the use of an
the weight of the carbon bed (gcarbon ), We the equilibrium
old and trusted predictive equation.
adsorption capacity (g/gcarbon ), Q the volumetric flow
rate (cm3 /min), cin the contaminant concentration in air
(g/cm3 ), cout the chosen breakthrough concentration (g/
*
Corresponding author. Tel.: +32-2-701-4020; fax: +32-2-701-4504. cm3 ), qb the bulk density of the carbon bed (gcarbon /cm3 )
E-mail address: [email protected] (P. Lodewyckx). and kv the overall adsorption rate coefficient (min1 ).
0008-6223/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2004.01.016
1352 P. Lodewyckx et al. / Carbon 42 (2004) 1351–1355

In the case of pure physisorption, especially for or- 2.3. Presence of water vapour
ganic vapours, this equation needs two parameters of
the carbon-adsorbate system to be calculated: the static It is well known that water vapour, be it in the
adsorption capacity We and the overall adsorption rate ambient air or pre-adsorbed on the carbon, will enter in
coefficient kv . The first one is usually calculated from the competition with the organic vapour. This will usually
Dubinin–Radushkevich equation (Eq. (2)) [3] and does result in a loss of adsorption capacity as well as a
not pose any major problems diminishing of the adsorption rate.


BT 2 2 cs 2.4. Non-linear flow (half-sine flow)
We ¼ Wo dL exp 2
log ; ð2Þ
b c in
Over the decades there has been much debate on the
where We is the adsorption equilibrium capacity (g/
use of testing activated carbon canisters with a linear
gcarbon ), Wo the micropore volume (cm3 /gcarbon ), dL the
(constant) flow. It is clear that in real use (i.e. attached
liquid density of the organic vapour (g/cm3 ), B the
to a gas mask) these filters will not ’see’ a linear flow but
structural constant of the carbon (K2 ), T the test
instead they will be subjected to a half-sine flow. Once
temperature (K), b the affinity coefficient of the organic
again, this should, theoretically, not influence the
vapour (–), cS the concentration at saturation vapour
capacity of the carbon but could have its repercussions
pressure (g/cm3 ) and cin the contaminant concentration
on the dynamics of adsorption: a lower value of kv
in air (g/cm3 ).
would lead to shorter breakthrough times (see Eq. (1)).
The estimation of the second parameter, the overall
adsorption rate coefficient kv , is less straightforward.
2.5. Chemisorption
Initially some very crude estimations were developed,
based on theoretical diffusion models. An alternative
The last extension of the Wheeler–Jonas equation is
was the experimental determination of kv by means of a
its applicability to describe the adsorption of non-or-
single breakthrough experiment [4]. Since then, several
ganic vapours, i.e. the chemisorption of gases. After all,
authors have proposed semi-empirical equations [5–8].
this equation was originally derived for the catalytic
The most recent one has been put forward by Lode-
oxidation of carbon monoxide [1]. It stands to reason
wyckx and Wood [8] (Eq. (3))
that both interaction parameters (capacity and adsorp-
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

 tion rate) will differ from the ones used in physisorption.
0:33 0:75 1:5 We
kv ¼ 800  b  vL  dp  ð3Þ Hence, Eqs. (2) and (3), especially (2), will probably not
MW be valid in this case. Even though the micropore volume
of the carbon still plays an important role in the
with kv the overall adsorption rate coefficient (min1 ), We
adsorption process, it is primarily based on chemical
the adsorption equilibrium capacity (g/gcarbon ), Mw the
interactions between the gas and the impregnants.
molecular weight of the vapour (g/mole), dp the average
diameter of the carbon particle (cm), b the affinity
coefficient of the organic vapour (–) and vL the linear
3. Experiment and results
velocity through the bed (cm/s).
Both Eqs. (2) and (3) have been applied to many
3.1. General
different granular activated carbons and a broad range
of gases and vapours. Some difficulties arise in the
Basically there are two ways to verify the applicability
application of the model for gases with low molecular
of the Wheeler–Jonas relation.
weights. However, in most cases it shows excellent re-
The first one [3] is based on one single breakthrough
sults for organic vapours, both low and high boiling.
curve, i.e. a plot of tb against ln½ðcin  cout Þ=cout . For
The use of the Wheeler–Jonas equation is widespread
high values of the latter (very low values of the break-
in both theoretical research and industrial applications
through concentration cout ) this plot should yield a
[9]. However, Eq. (1) being directly derived from a mass
straight line. Furthermore, one can calculate We and kv
balance over the carbon bed, its general nature should
from this graph, as We is linked to the intercept with the
permit the extension to other types of adsorption
y-axis and kv to the slope of the line (see Eq. (1)).
(chemisorption) and other environmental circumstances.
This is illustrated by Fig. 1. However, this method
presents several disadvantages: it is not always clear how
2.2. Activated carbon fibres (ACF) many points have to be taken into consideration for the
fitting, there is a high uncertainty on the calculated
The adsorption properties of ACF are essentially the values of We and kv (which follows partly from the first
same as for GAC. In this case, however, it is clear one problem) and, most important, all evidence is based on
would expect much faster kinetics. one single breakthrough experiment.
 P. Lodewyckx et al. / Carbon 42 (2004) 1351–1355 1353

180 do not have to be determined experimentally (even

160 though it stays advisable): they can be calculated in the
140 same way as for a granular activated carbon [3,8] using
120 Eqs. (2) and (3). An interesting corollary of these find-
tb (min)

100 ings is that, apparently and contrary to common belief,

80 it appears that the rapid adsorption by ACF is not re-
60 lated to the lack of mesoporosity (micropores situated at
40 the surface of the fibre). But the applicability of Eq. (3)
20 to calculate kv points at the fibre dimensions and the
0 high micropore volume to be the main reasons for the
-2 -1 0 1 2 3 4 5 6
ln((cin-cout)/cout)
enhanced adsorption speed.

Fig. 1. Validity of the Wheeler–Jonas equation––method 1.

3.3. Presence of water vapour

This is largely overcome by the second method [10]. Much work has been done in order to quantify the
In stead of taking one breakthrough curve, one uses effect of water vapour adsorption on the capacity of
several experiments with different amounts of activated activated carbon adsorbents. In several papers [13–15] a
carbon. If this is simply achieved by changing the bed model has been proposed, based on the Wheeler–Jonas
depth, Eq. (1) shows that a plot of breakthrough time equation, to calculate breakthrough times under humid
against the bed weight should yield a straight line. Once conditions. Even though the model can lead to rather
again, We and kv can be calculated from slope and large errors in some specific cases, the overall result is
intercept. Of course, the final test to prove, unambigu- quite good. Fig. 3 shows the agreement between exper-
ously, the applicability of the Wheeler–Jonas equation is imental and calculated values of breakthrough time for
to use it in a predictive role. Based upon a limited seven different organic vapours, under varying condi-
number of experiments (to calculate or to verify the tions of air stream humidity and pre-wetting of the
values of We and kv ) the equation should be able to carbon, for a total of 150 experiments. According to
predict breakthrough times for different amounts of the model, both We and kv are (negatively) influenced by
carbon, inlet concentrations and flow rates. the adsorbed water vapour. As the water is occupying a
part of the micropore system, it will lower the available
3.2. Activated carbon fibres (ACF) adsorption volume, and hence We . The water adsorbed
throughout the complete pore system (i.e. micro- and
All three tests for the applicability of the Wheeler– mesopores) will hamper the adsorption and slow down
Jonas equation have been carried out on ACFs. Some the kinetics (reducing kv ). Of course, one has to take into
results were recently communicated [11,12]. These show account the competition between the organic vapour
that it is possible to predict the breakthrough time of an and water, i.e. the ability of certain vapours to reclaim
ACF bed against organic vapours, as shown in Fig. 2. adsorption space.
They also show that in this case the values of We and kv

120 250

100
200
Theoretical tb (min)

80
150
t (WJ)

60
b

100
40

20 50

0 0
0 20 40 60 80 100 120 0 50 100 150 200 250
t b (exp) Experimental tb (min)

Fig. 2. Calculated vs. experimental values of tb for DMMP on ACF. Fig. 3. Experimental vs. theoretical tb in the presence of water vapour.
1354 P. Lodewyckx et al. / Carbon 42 (2004) 1351–1355

experimental breakthrough time [min]

100 80
90 k v = F[vL0.75]
80
60
70
(Mean Flow)0.75 =12.89
Q (l/min)

60 Mean (Flow)0.75 =10.53

50 Ratio = 1.22 40
40
70-70%RH; 70g
30 Pulsed flow:
30*PI*sin(t/10) 20 0-0%RH; 70g
20 Mean value y=x
10
0 0
0 200 400 600 800 1000 0 20 40 60 80
t (s) breakthrough time according to the Wheeler-Jonas [min]

Fig. 4. Half-sine flow vs. steady flow: influence on kv . Fig. 5. Experimental vs. calculated tb for NH3 .

3.4. Non-linear flow (half-sine flow)

Values of kv fall in two categories: sometimes they can
A very interesting recent study [16] has given more be calculated by Eq. (3), suggesting surface diffusion to
insight in the way We and kv are influenced by changing be the rate-determining step in the adsorption process.
the flow pattern from a steady to a half-sine flow. The In other cases this is impossible, proposing a different
same group found an explanation for the influence on rate-determining step, e.g. the chemical interaction be-
kv [17], however it is still unclear why and how the tween certain impregnants and the adsorbed gas. More
capacity We should diminish due to the changed flow work has to be done in order to quantify relationships
pattern. A very simple explanation for the lower value between We and kv on one hand, and measurable
kv can be given by examining equation (3). Here one parameters such as flow rate, volatility, micropore vol-
can see the dependence of kv on v0:75 L [8]. When using a ume and quantity of impregnants on the other hand.
steady flow (see Fig. 4) this means taking the power These are necessary to perform ab initio calculations of
0.75 of the mean value of vL (calculated from the mean breakthrough times for chemisorbed substances.
value of the volumetric flow rate Q). With a half-sine
flow one should calculate vL for every value of Q (see
Fig. 4), take the power 0.75, and only then calculate
the average value. A few simulations have demon- 4. Conclusions
strated that: v0:75
L < ½vL 0:75 , the difference being roughly
the same as the one between the two experimentally The scope of the Wheeler–Jonas (or Reaction Ki-
determined values of kv . netic) equation extends largely beyond its commonly
accepted boundaries. This is primarily due to its
3.5. Chemisorption apparent simplicity: the combination of a single capacity
term and an overall kinetic effect strongly enhances its
The general nature of the Wheeler–Jonas equation, applicability to different adsorption circumstances. In
i.e. capacity minus the influence of adsorption kinetics, this way it is far more potent than many of the more
should be able to describe the chemisorption of gases modern equations that require the exact knowledge of
and vapours [18]. Experiments were conducted on a several, not readily available, input parameters (such as
wide range of chemisorbed substances such as CNCl, tortuosity, gas–carbon interactions, etc.). The Wheeler–
H2 S, SO2 , Cl2 , COCl2 and NH3 . Some of these data Jonas equation can be used to describe any type of
have already been published [18–20], but all of them adsorption of a single vapour by a suitable adsorbent.
seem to indicate the applicability of the Wheeler–Jonas Theoretically its use may even be extended to co-
equation. Fig. 5 shows some examples of breakthrough adsorption, as already demonstrated by its ability to
times estimated by Eq. (1), based on experimentally predict organic vapour breakthrough times under humid
determined values of We and kv in the case of ammonia. conditions.
The results are very good, both under dry circumstances In some cases it is possible to perform ab initio pre-
and for pre-humidified (70% RH) filters. As expected, dictions of breakthrough times. For other applications,
the experimentally determined values of We cannot be such as chemisorption, its use is, at this time, still limited
calculated by Eq. (2). However, there seems to be one to extrapolations of one or more breakthrough experi-
unique value for each specific gas–carbon couple (as it ments. However, even in these cases it should be possible
should be in order to apply Eq. (1)). For some gases, at to find theoretical expressions of the capacity We and the
very low concentrations, this value of We changes, indi- overall mass transfer coefficient kv , based on readily
cating a different adsorption mechanism. available macroscopic parameters.
 P. Lodewyckx et al. / Carbon 42 (2004) 1351–1355 1355

References [12] Lodewyckx P, Ryu SK. Modelisation of the adsorption of organic

vapours on activated carbon fibres by means of the reaction
[1] Wheeler A. Performance of fixed-bed catalytic reactors with kinetic equation. In: Proceedings of the Third International
poison in the feed. J Catal 1969;13:299. Carbon Conference, Beijin, PRC; 2002.
[2] Jonas LA, Rehrmann JA. Predictive equations in gas adsorption [13] Lodewyckx P, The influence of water on the adsorption of organic
kinetics. Carbon 1973;11:59. vapours by activated carbon. University of Antwerp, Antwerp,
[3] Wood GO. Activated carbon adsorption capacities for vapors. Belgium, PhD thesis; 1998.
Carbon 1992;30:593. [14] Lodewyckx P, Vansant EF. The influence of humidity on the
[4] Busmundrud O. Vapour breakthrough in activated carbon beds. adsorption capacity from the Wheeler–Jonas model for the
Carbon 1993;31:279. prediction of breakthrough times of activated carbon beds. Am
[5] Jonas LA, Rehrmann JA. The rate of gas adsorption by activated Ind Hyg Assoc J 1999;60:612–7.
carbon. Carbon 1974;12:95. [15] Lodewyckx P, Vansant EF. The influence of humidity on the
[6] Wood GO, Stampfer JF. Adsorption rate coefficients for overall mass transfer coefficient of the Wheeler–Jonas equation.
gases and vapors on activated carbons. Carbon 1993;31: Am Ind Hyg Assoc J 2000;61:461–8.
195. [16] Suzin Y, Nir I, Kaplan D. The effect of flow pattern on the
[7] Lodewyckx P, Vansant EF. Estimating the overall mass transfer adsorption of DMMP in activated carbon beds and canisters.
coefficient kv of the Wheeler–Jonas equation: a new and simple Carbon 2000;38:1129–33.
model. Am Ind Hyg Assoc J 2000;61:501–5. [17] Nir I, Kaplan D, Suzin Y. Extension of the Wheeler–Jonas model
[8] Wood GO, Lodewyckx P. An extended equation for rate (WJM) for dynamic adsorption to pulsating flow. In: Proceedings
coefficients for adsorption of organic vapors and gases on of the 7th International Conference on Fundamentals of Adsorp-
activated carbons in air-purifying respirator cartridges. Am Ind tion––FOA 7, Nagasaki, Japan; 2001.
Hyg Assoc J 2003;64(5):646–50. [18] Verhoeven L, Lodewyckx P. Using the Wheeler–Jonas equation to
[9] Moyer ES, Peterson JA. Organic vapor (OV) respirator cartridge describe adsorption of inorganic molecules: ammonia. In: Pro-
testing against methylene chloride. Appl Occup Environ Hyg ceedings of the Second International Carbon Conference, Lex-
1993;8:553–63. ington, KY, USA; 2001.
[10] Wood GO, Moyer ES. A review of the Wheeler equation [19] Verhoeven L, Van Rompaey D. Lodewyckx P. The temperature
and comparison of its applications to organic vapor cartridge effects of the adsorption of phosgene on military filters. In:
breakthrough data. Am Ind Hyg Assoc J 1989;50:400– Proceedings of the Second International Carbon Conference,
7. Lexington, KY, USA; 2001.
[11] Ryu SK. Adsorption of DMMP (dimethyl methylphosphonate) [20] Lodewyckx P, Verhoeven L. Using the Wheeler–Jonas equation to
on activated carbon fibers. In: Proceedings of the Second describe the adsorption of inorganic molecules: chlorine. Carbon
International Carbon Conference, Lexington, KY, USA; 2001. 2003;41:1215–9.