A Simple Technique for the Measurement of H
2
Sorption Capacities
John M. Zielinski,* Peter McKeon, and Michael F. Kimak
Air Products and Chemicals, Inc., 7201 Hamilton Boule
V
ard, Allentown, Pennsyl
V
ania 18195
An accurate (and lowcost) experimental technique has been developed to screen the effectiveness of anadsorbent in improving gas storage capacity within a pressurized vessel. Specifically, the capsule techniqueis shown to be effective in directly measuring the total H
2
contained within a pressurized vessel and can beused to evaluate the amount of gas in the free space and adsorbed on the solid, that is, a sorption isotherm.The capsule technique was benchmarked by measuring isotherm data for CH
4
on an activated carbon sampleand was then subsequently evaluated for use with H
2
. The capsule data are in excellent agreement with thetotal storage capacities expected from calculations using equation of state information. In addition, H
2
isothermdata from the sorption capsule are found to be within 1% of values obtained from a more sophisticateddifferential pressure adsorption unit (DPAU). Conditions for when the adsorbent aids or hinders storage arealso discussed in terms of the 2010 DOE H
2
storage targets.
Introduction
Many researchers are currently committed to developingenabling technologies for the successful introduction of hydrogen as an alternative fuel for both stationary and transportationapplications. One of the key technical hurdles to widespreaduse of hydrogen fuel cells is its ability to be stored at highdensities in a practical manner. To support the development of advanced hydrogen storage materials and processes, our laboratory has designed and built two instruments: (1) a differentialpressure adsorption unit (DPAU), capable of accurately measuring H
2
sorption isotherms up to
∼
2000 psia with as little as100 mg of sample,
1,2
and (2) a sorption capsule, which is ideallysuited for rapidly screening candidate adsorbents and whichdirectly provides the total hydrogen loading in a vesselcontaining adsorbent. The latter is the subject of this Article.If solid adsorbents are placed within a gas cylinder, theyoccupy a portion of the volumetric space. Despite this loss of gasphase volume, if the gas

solid interactions are sufficientlyfavorable, there is the potential to reversibly store more totalmolecules of adsorbate within this type of a system than withina conventional pressurized gas cylinder. Alternatively, one maybe able to store the same amount of H
2
in a container containingadsorbent at lower pressures than in a pressurized emptycontainer, thereby yielding a storage system that is inherentlysafer (i.e., is at lower pressure) and that has less of a wallthickness requirement for the container. In turn, the reductionof wall thickness would lead to lower cost containers. Thesuccessful implementation of such an adsorbentbased storagesystem is centered on the development of adsorbent materialsthat have sufficient reversible H
2
sorption characteristics.Many experimental techniques have been developed tomeasure gas

solid equilibrium data based on knowledge of thetotal moles of adsorbate contained within a system and anexperimental assessment of the moles of the adsorbate residingin the gas phase by techniques such as IR spectroscopy,
3
NMRspectroscopy,
4
GC headspace analysis,
5,6
and through simpleuse of a pressure transducer.
7
The moles of gas adsorbed onthe solid phase, therefore, can be inferred by difference of thesetwo quantities.In this work, we present our experimental methodology fora sorption capsule technique and provide benchmarking datato examine its effectiveness in measuring the total loading of agas within a pressurized vessel as well as the more difficultexperiment of evaluating a sorption isotherm, using a pressuretransducer to evaluate the amount of adsorbate in the gas phase.Experimental limitations will be discussed along with conditionsunder which the presence of the adsorbent is found to hinderthe total storage capacity.
Experimental SectionMaterials.
GX31 Supercarbon was obtained from Amoco.All of the gases used were obtained from Airgas. The hydrogenused was Research Grade (99.9995%), the helium was ultrapureHe BIP PLUS (
<
20 ppb water,
<
10 ppb O
2
), and the methanewas ultrahigh purity (99.99%). Hydrogen and methane werefurther purified by passing the gases through an active metalpointofuse purifier (Matheson TriGas, model MN12).
Sample Preparation.
Samples for adsorption testing wereactivated by degassing the materials during a series of temperature ramps and isothermal soaks while under a dynamicvacuum. Typically,
∼
1.5

2.0 g of sample was loaded in anactivation cell within an argon glove box and attached to anASAP 2010 (Micromeritics). The samples were then heated ata rate of 10
°
C/min to 100
°
C and held at that temperature for30 min. The temperature was subsequently increased to 300
°
C at the same ramp rate and held there until a vacuum readingof less than 10

4
Torr was achieved.After the activation, samples were transferred back into theargon glove box and weighed into a highpressure sorptioncapsule cell for gas sorption studies. The activated sample wasthen removed from the glove box, connected to the sorptioncapsule apparatus, and outgassed at ambient temperature toremove the argon sorbed while loading the sample into thecapsule cell. Final weighings to determine sample weight wereperformed using a fiveplace analytical balance. Both theanalytical balance and the sorption capsule reside in a nitrogenpurged Lexan box to avoid complications associated with weightchanges based on condensation of humidity on the externalsurface of the sorption capsule.
Apparatus.
The basis of the sorption capsule technique liesin the ability to accurately quantify the amount of gas containedwithin a vessel by comparing the weight of the evacuated vessel
* To whom correspondence should be addressed. Tel.: (610) 4817975. Fax: (610) 4816578. Email: zielinjm@airproducts.com.
329
Ind. Eng. Chem. Res.
2007,
46,
329

335
10.1021/ie060700y CCC: $37.00 © 2007 American Chemical SocietyPublished on Web 12/06/2006
to the weight of the vessel when it contains pressurized gas.The difference in these two weights directly gives the amountof gas stored within the container at a known pressure,
P
, anda temperature,
T
.The effect that a solid adsorbent has in either enhancing orhindering the amount of gas able to be stored can also be easilydiscerned by a similar procedure. First, a solid adsorbent isloaded into the sorption capsule and the system is evacuated.A comparison of the weight of the evacuated capsule, whichcontains solid, to the evacuated weight when empty yields thesample weight. When gas is introduced to the system containingadsorbent at the same pressure as in the empty cell experiment,
P
, and allowed to achieve equilibrium, gas resides both in thefree space and on the adsorbent. A comparison of the weightof the pressurized cell containing adsorbent with that of apressurized cell containing no solid reveals the enhancement(or hindrance) that the solid plays in terms of its gas storagecapability. As described, this experiment provides the totalamount of gas contained within a pressurized vessel containingsolid. It does not directly yield information regarding thepartitioning of the adsorbate between the solid phase and thegas phase.It is interesting to note that neither the empty cell volume,the free space volume, nor the sample weight are needed toassess whether the presence of the solid improves or impedesgas sorption within the vessel by the method outlined. Thesequantities, however, are essential if a sorption isotherm isdesired. The explicit relationship between the extent of sorption,the volume occupied by the adsorbent, and the system free spaceis developed in the next section of this Article.A schematic of the capsule apparatus is provided in Figure1. Two key components in the capsule experiment employedin our studies are the two sorption cells, which are hooked to agas handling system, and the analytical balance used forweighing them. The analytical balance employed was a SartoriusResearch Series MC1 balance, which has a weighing limit of 210 g and a readability of 0.05 mg. The sorption cells wereconstructed out of T316 stainless steel and were orbitallywelded to a 1/4” VCO male tube weld gland. The cells were
∼
7 cm
3
in volume possessing an O.D. of 12.8 mm and an I.D.of 9.5 mm and were hydrostatic tested to 3200 psig. The shutoff valves on the cells, from Swagelock (SSIRVCO4SC11),are constructed out of stainless steel and have a maximumpressure rating of 6000 psia. The sorption cells were sized sothat the total weight of a cell, with a valve attached, was
∼
180g. This enables one to perform weighing measurements withthe fiveplace Sartorius balance even with an appreciable samplesize.Performing weight measurements accurately to five decimalplaces can be challenging when one considers changes inlaboratory temperature and humidity levels. To circumvent theseproblems, the entire capsule apparatus is housed in a temperaturecontrolled Lexan box, which is nitrogen purged so thatthe relative humidity (RH) was maintained at a steady 10%.The box is also equipped with two rubber gloves so that thecells can be transferred from the gas manifold system to theanalytical balance without exposure to laboratory atmosphereor direct contact with human hands, which can contaminate theoutside of the sorption cells and alter their weight.
Theory
The number of moles of gas,
n
, which can be contained inan empty vessel of volume,
V
E
, at pressure,
P
, and temperature,
T
, can be expressed as:where
z
is a compressibility factor that accounts for gasphasenonidealities, and
R
is the universal gas constant.If the vessel is filled with an adsorbent material and is againpressurized with an adsorbate gas, the total number of molesof gas within the container has contributions from both the gasphase as well as the solid phase such that:where
V
FS
is the free space (gasphase) volume,
M
S
is the massof the solid adsorbent, and
K
is a partition coefficient, indicatingthe distribution of adsorbate between the gas phase and the solidphase, such thatHere,
n
ADS
is the Gibbsian excess moles adsorbed on the solid.In the case of a Langmuirian isotherm, the partition coefficientis given as:The parameters
m
and
b
are termed the saturation capacity andadsorption coefficient, respectively. As indicated in eq 4, thepartition coefficient can be a function of pressure; however, inthe lowpressure limit,
K
ceases changing with pressure andequals the Henry’s constant,
K
H
, which is also equal to theproduct
mb
.Because
V
FS
represents the difference between the empty cellvolume and the volume occupied by the solid adsorbent, onecan write:where
F
S
is the skeletal density of the solid adsorbent.Substitution of eq 5 into eq 2 yields a second, and somewhatmore insightful, expression for the total adsorbate containedwithin a pressurized vessel:
Figure 1.
Schematic of the sorption capsule unit. The sample cells, gasmanifold, pressure transducer, and analytical balance are contained in atemperatureregulated (25
°
C), nitrogenpurged Lexan box. The nitrogenflow rate was regulated so that the relative humidity was maintained at
∼
10%.
n
)
PV
E
zRT
(1)
n
)
PV
FS
zRT
+
KPM
S
(2)
n
ADS
)
KPM
S
(3)
K
)
mb
1
+
bP
(4)
V
FS
)
V
E

M
S
F
S
(5)
n
)
PV
E
zRT
+
M
S
P
[
K

1
F
S
zRT
]
(6)
330
Ind. Eng. Chem. Res., Vol. 46, No. 1, 2007
The first term on the righthand side of eq 6 represents thenumber of moles of gas that would be contained in an emptypressurized cell. The second term, therefore, represents the effectthe solid has on the total loading within the cell. Because thebracketed term is a difference, it can be either positive ornegative. To increase the amount of gas within the containercontaining solid above that amount contained in the emptycontainer, the sorption capacity (as indicated by the partitioncoefficient) must be sufficiently large to overcome the free spacevolume loss due to the physical presence of the solid. Thus, if
K
>
1/
F
S
zRT
, the solid increases the total gas capacity; otherwisethe space that the solid occupies contains more gas moleculeswhen the container is empty.To illustrate this point further, in Figure 2 we presentcalculated results for the weight of hydrogen contained withina 42.2 L volume, that is, the size of a commercial BX H
2
cylinder, at 21
°
C and as a function of pressure, for the caseswhen the cylinder is empty or contains 66 lb of AmocoSupercarbon GX31. The weight of GX31 is assumed to bethe maximum loading based on a skeletal density of
∼
2 g/cm
3
and a 0.71 g/cm
3
packing density.
8
The H
2
weight contributiondue to adsorption onto GX31 was calculated from a dualsiteLangmuir (DSL) isotherm correlation of variabletemperaturesorption data measured for this system.
1,2
The DSL model isgiven as:The model parameters for the H
2

GX31 system are providedin Table 1. The bracketed term provides the temperature andpressure dependence of the partition coefficient,
K
.The dotted line in Figure 2 indicates the grams of H
2
contained in the empty vessel as a function of pressure. As onemight expect, the higher is the pressure, the higher is the H
2
loading. For comparison, the specifications for a fresh BX H
2
cylinder are 6000 psia at 21
°
C, which corresponds to
∼
1140g (2.5 lb) H
2
. Because a BX cylinder weighs
∼
300 lb, thestorage of H
2
is 0.8% by weight (2.5 lb/302.5 lb) and 27 g H
2
/Lby volume. For comparison, the DOE 2010 storage targets are6 wt % and 45 g H
2
/L.
9
The solid line, indicated as “a”, represents the total H
2
loadingwithin the 42.2 L vessel when it contains 66 lb of GX31.Notably, at low pressures the presence of the adsorbent aidsthe storage capacity within the vessel, while at pressures above4700 psia the GX31 is expected to hinder hydrogen storage.The maximum improvement in storage capacity at a fixedpressure occurs at 1700 psia where the enhancement is
∼
71 g(0.16 lb). This is a minimal improvement over simply pressurizing the empty container and is clearly not sufficient enhancement for a practical device.An alternate way of viewing the sorption enhancement is toexamine the largest pressure savings one can obtain for aparticular gas loading. At 2000 psia, the total H
2
capacity inthe empty cell is
∼
448 g. With GX31 loaded into the storagecell, the same total H
2
capacity is achieved at 1665 psia, whichis a savings of 335 psia. Being able to store comparable H
2
loadings at lower pressures implies savings in the wall thickness(and, consequently, weight) of the container used for storage.In addition, storage of a lower pressure gas is inherently saferthan storage of a highpressure gas.Also included in Figure 2 is a solid line (“b”), which indicatesthe total expected storage capacity of the system if the adsorbentisotherm were tripled. Clearly, this increase in the H
2
isothermhas a profound impact on the extra amount of H
2
that can bestored at a fixed pressure and the pressure savings at a fixedloading. For this scenario, the energy density at 6000 psia is 45g H
2
/L, which meets the DOE 2010 volumetric storage target.This type of increase in the sorption isotherm is consideredachievable by many and is precisely the reason that materialscientists are striving to develop improved materials for H
2
storage.If the same BX cylinder were used as the storage container,however, the gravimetric capacity would still only be 1.1 wt%, because the 66 lb of adsorbent must be added to the 300 lbweight of the cylinder to fairly consider the system storagecapability. Clearly, developing storage devices that are lightweight and highstrength is critical to the H
2
storage effort.
Results and DiscussionTotal Capacity Measurements (Empty Cell).
The sorptioncapsule is ideally suited for performing measurements of thetotal loading of a gas within a pressurized container. Initially,an evacuated cell is attached to the gas handling system (GHS)and evacuated until a pressure of 10

2
Torr is reached. Theshutoff valve on the evacuated cell is then closed, and the cellis removed from the GHS and is weighed on the analyticalbalance. This procedure is used whether the sample cell is emptyor contains an adsorbent material. The cell is then reattachedto the gas handling system, the valve is opened, and the systemis pressurized with the gas of interest. Once equilibrium isestablished, as indicated by the pressure transducer on the GHS,the valve is once again shut off, and the cell is detached fromthe GHS and weighed on the analytical balance. This proceduredirectly measures the weight of the gas introduced into thesample chamber.To test our experimental protocol for weighing gas, thevolume of the manifold,
V
m
(indicated in Figure 1), wasevaluated by performing pressure expansions using a calibrated
Figure 2.
Variable pressure hydrogen loading calculations for a BX H
2
cylinder volume (42.2 L) at 21
°
C. The dotted line indicates the H
2
containedin the system when there is no adsorbent in the cell (eq 1). The solid line(a) illustrates the H
2
capacity when 66 lb of Amoco Supercarbon GX31 isloaded into the cell (eq 2). The solid line (b) represents the total H
2
loadingin the system if the adsorbent capacity were tripled (eq 2). The calculationshave assumed that the skeletal density,
F
S
, is 2.0 g/cm
3
and a packing densityof 0.71 g/cm
3
.
Table 1. DSL Parameters for the System H
2

GX31
m
1
4.486 mmol/g
m
2
4.733 mmol/g
b
o
4.198
×
10

6
psia

1
Q
b
2584.9 cal/mol
d
o
2.459
×
10

5
psia

1
Q
d
1550.9 cal/mol
n
ADS
)
[
m
1
b
0
exp
(
Q
b
RT
)
1
+
b
0
exp
(
Q
b
RT
)
P
+
m
2
d
0
exp
(
Q
d
RT
)
1
+
d
0
exp
(
Q
d
RT
)
P
]
PM
S
(7)
Ind. Eng. Chem. Res., Vol. 46, No. 1, 2007
331
standard volume. Our manifold volume was determined to be9.6 cm
3
. The volumes of our empty sample cells,
V
E
, weresubsequently evaluated by performing pressure expansions fromthe manifold volume. The two sample cells employed in thisinvestigation had volumes of 6.8 and 7.1 cm
3
, respectively.The two empty sample cells were then pressurized withhelium and hydrogen, respectively, at isothermal conditions.Because the system pressure, temperature, and volume withineach sample cell were known, the weight of gas within thesample cell could be calculated from an equation of state(EOS).
10

12
In Figure 3, we provide a comparison between theweight predicted by the equation of state and our experimentaldata. The data are found to agree with the EOS to within
∼
1%error.
Total Capacity and Isotherm Measurements (AdsorbentPresent).
In traditional volumetric sorption studies, heliumexpansions are performed to assess the free space within asample cell. Because the manifold volume and individual cellvolumes of the sorption capsule unit have been calibrated, thistraditional method can be applied to assess the free space oncean adsorbent is loaded into the system. There are two alternativesby which the free space volume can be determined in thesorption capsule experiment.In the first method, the cell can be pressurized with heliumto a known pressure at a known temperature, and the cell canbe weighed to gauge the mass of helium within the cell. Thishelium is presumed to reside exclusively in the gas phase,because helium is generally considered a nonadsorbing gas.Because pressure and temperature are known, the helium densitycan be determined from an equation of state. Consequently, thecell free space volume can be determined by dividing the massof helium measured by the helium gas density.The second method requires knowledge of the skeletal densityof the adsorbent,
F
S
. If a known mass of solid,
M
S
, is loaded inthe sample cell, the free space volume can be calculated directlyfrom eq 5. All three of these techniques have been employedto estimate the free space volume using quartz as the “adsorbent”and have been found to agree to within
∼
1%.To benchmark the sorption capsule technique using a realadsorbent, CH
4
and H
2
adsorption experiments at 25
°
C wereperformed using Amoco Supercarbon GX31 as the solidadsorbent. For the CH
4
experiments, 1.6235 g of GX31 wasloaded into the sample chamber, the free space volume wasanalyzed by helium expansions, and the system was subsequently pressurized with methane and allowed to equilibrate.The total amount of methane contained in the system wasdetermined by closing the valve on the sample chamber andweighing the entire sorption capsule. The capsule could thenbe reattached to the system and exposed to methane at a newpressure. In this way, the total loading of methane could bedetermined as a function of pressure. The free space volumedoes not need to be known to determine the total amount of methane contained within the system. This result is providedin Figure 4.Although
V
FS
is not needed to determine the total amount of methane in the system, it is needed if a sorption isotherm isdesired. Included in Figure 4 is the calculated amount of methane in the gas phase of the sorption capsule using anequation of state and knowledge of
P
,
T
, and
V
FS
. By dividingthe mass of methane on the solid adsorbent by the mass of solid,one can determine the sorption capacity (mmol/g), that is,surface excess at the corresponding
P
and
T
.The methane isotherm resulting from the measurementsshown in Figure 4 is provided in Figure 5 along with a literatureisotherm from a very sensitive differential pressure adsorptionunit (DPAU).
1
Five repeat points were measured at 250 psia toillustrate the reproducibility of the sorption capsule method. Theagreement between the two experimental data sets is excellentand highlights the capability of the relatively inexpensivesorption capsule unit.One point that needs to be highlighted is the magnitude of the
y
axis in Figure 4. The mass of methane weighed in theseexperiments is on the order of 100 mg, which is easilyaccomplished with a fiveplace analytical balance that has areadability of 0.05 mg. For the methane/GX31 system, onaverage
∼
60% of the total methane mass within the system is
Figure 3.
Comparison of the experimentally measured total mass of helium(
O
) and hydrogen (
b
) contained in pressurized cells at 25
°
C to that expectedfrom an equation of state analysis based on knowledge of the pressure,temperature, and system volume.
Figure 4.
Comparison of the experimentally measured total mass of methane (
O
) at 25
°
C in a system containing 1.6235 g of AmocoSupercarbon GX31. The solid line indicates the contribution of gasphasemethane to the total methane weight in the system.
Figure 5.
Isotherm data at 25
°
C for methane adsorption on AmocoSupercarbon GX31 measured by the sorption capsule technique (
b
) andby a differential pressure adsorption unit (
O
). Five repeat runs wereperformed at 250 psia to illustrate the reproducibility of the sorption capsulemethod.
332
Ind. Eng. Chem. Res., Vol. 46, No. 1, 2007