Professional Documents
Culture Documents
Grav Journa
Grav Journa
Grav Journa
ScienceDirect
article info inties in the calculation of the amount of gas absorbed or desorbed by a sample using
has been made. The gravimetric technique provides a near direct measurement of the
Article history: Received 2 October 2013 ously measuring the weight of the sample plus absorbed gas and is frequently used to
Received in revised form 19 February 2014 e of hydrogen in potential hydrogen storage materials. These calculations provide the
Accepted 25 February 2014 Available ncertainty in the final gas uptake measurement given knowledge of the measurement
online 25 March 2014 ogen gas on activated carbon has been modelled to allow individual constants such as
well as measurement uncertainties to be changed while determining the effect on the
urement uncertainties critical to minimising the overall uptake uncertainty have been
Keywords: Uncertainty on with another popular gas up- take method, the Sieverts technique, has been made.
analysis Hydrogen storage Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
Gravimetric
http://dx.doi.org/10.1016/j.ijhydene.2014.02.154 0360-3199/Crown Copyright © 2014, Hydrogen Energy Publications, LLC. Published by Elsevier
Ltd. All rights reserved.
Introduction he pressure measurements in addition to the amplification of un- certainties by
arger amounts of gas (large volumes or high pressures or both). Another factor
which was confirmed and clarified by this analysis was the effect of poor
Two of the more popular methods of determining the uptake of gas into or on to
knowledge of the sample density which is required to determine the volume of
materials, which is crucial to hydrogen storage material characterisation as well
he sample cell not occupied by the sample (void volume), an effect which has
as other areas of research, are the Sieverts technique and the gravimetric
been observed previously [3].
technique. The Sieverts technique employs measurement of the pressure and
temperature of the gas together with knowledge of the volumes involved to Materials of low density, porous materials or materials for which
determine the number of moles of gas before and after the gas is released into density and volume are difficult to experimentally determine or even define, are
the volume containing the sample. The difference between the calculated particularly vulnerable using the Sieverts technique and it is these materials
amount of gas finally in the system using these measurements and that expected which are often of greatest interest for gas storage applications. The poor
from the amount of gas initially must be the amount absorbed/adsorbed (or
desorbed) by the sample. This method has been described in detail in
* Corresponding author. Tel.: þ6 1 7 3735 5023; fax: þ6 1 7 3735 7656.
Ref. [1]. In a recent article by the authors [2], a full uncertainty analysis was
performed for the Sieverts technique, pointing out the susceptibility of the E-mail address: j.webb@griffith.edu.au ( C.J. Webb).
technique to the step-wise accu- mulation of uncertainties and the uncertainty in
repeatability of hydrogen uptake measurements across different laboratories has been indicated by the international round-robin exercises for
hydrogen adsorption by porous carbons [4,5]. In view of the sensitivity to sample volume/ density exposed by our analysis [2], poor knowledge of
the sample density has likely been a major factor in this.
The gravimetric technique offers some potential advan- tages over the Sieverts technique since the weight of the sample is continuously measured.
A sample is typically con- tained in a bucket which is suspended from one side of a balance arm by a hang-down wire, rod or magnetic coupling.
On the other, tare side of the balance arm (for the case of double-sided instruments), a similar hang-down and bucket system is employed with a
tare weight approximately equal to the sample weight. A simplified schematic of a gravimetric instrument is shown in Fig. 1. A variation on this is
the single- sided balance system which does not have a tare weight. Our analysis here is for the more general case of the double-sided balance.
The gravimetric setup is described in more detail in Ref. [1]. The balance then effectively measures the difference in weight force between the
sample side and tare side. Pres- sure is typically applied in steps, but in principle any mecha- nism that increases the pressure such as a piston
intensifier or a flow controller can be used. The uptake is determined from the increased weight of the sample due to the absorbed or adsorbed gas.
However, this weight must be corrected for the total buoyancy force of the combined sample plus absorbed gas and to calculate this, the pressure
and temperature must be known for the gas displaced by the volume of the sample as well as the volume of the sample itself. This is equivalent to
the volume correction to determine the correct void volume in the Sieverts technique and therefore requires knowledge of
international journal of hydrogen energy 39 (2014) 7158 e7164 7159
Fig. 1 e Simplified schematic of a gravimetric apparatus employing a symmetric balance with both the sample (S) and the tare weight (T)
suspended in the gas.
the sample volume as a function of gas uptake. The volumes of the fixed attachments (balance arm, hang-down wires, buckets or similar) on both
sides of the balance must also be determined in order to calculate positive and negative buoy- ancy effects at each pressure and temperature used.
In this paper we present a detailed analysis of the uptake uncertainty for the gravimetric technique which addresses the relative effects of
measurement uncertainties, buoyancy, vol- umes and sample density on the uncertainty in the determined gas uptake. Since this analysis is
independent of whether the gas is absorbed or adsorbed and, with appropriate sign change equally valid for desorption, the term “absorption” will
hence- forth be used to mean absorption, adsorption or desorption.
Methods
The concepts of uncertainty and error have already been detailed in Ref. [2] as well as in Ref. [6]. Briefly, “uncertainty” means lack of knowledge
leading to lower precision whereas “error” implies a mistake that is in principle correctable. It is assumed here that all sources of error that can be
eliminated have been removed. The focus here is on the uncertainties introduced and their propagation culminating in the uncer- tainty of the final
uptake calculation. Following on from the work in Ref. [2], and using the same terminology, which re- flects that of the Joint Committee for
Guides in Metrology (JCGM) [7], we will determine first the uptake and then the uncertainty in the uptake for the gravimetric method.
In general, provided the quantities x are uncorrelated and the estimates of uncertainties, u, in the individual quantities are equivalent to statistical
uncertainties, then the uncertainty in a function of N variables can be calculated from the positive square root of the variance, determined by (Eqn
(10) of JCGM)
∑
u2ðyÞ 1⁄4 Ni1⁄41
( vf
where u2(y) is the variance of the function value y 1⁄4 f (x1,.,xN) and
u2(xi) is the variance of the individual xi values.
If n is the number of moles of gas in a system of volume V at pressure P and temperature T, then the uncertainty dn can be calculated from the
non-ideal gas equation to be (Eqn 7 of Ref. [2]).
(dnn
)2
1⁄4
(dPP
)2
þ
(dVV
)2
þ
(dTT
)2
þ
(dZZ
)2
(2)
where dP is the absolute standard uncertainty in the pressure measurement (and similarly for the other quantities) and Z is the compressibility for a
non-ideal gas. jdP/Pj represents the relative uncertainty in the pressure.
In the gravimetric technique, the weight of the sample is measured and, after compensating for the effect of buoyancy, the weight of the uptake of
gas can be determined. During an absorption step at a particular pressure, the balance measures the difference between the weight and the
buoyancy (but typically provides values in units of mass),
1⁄4 mk Bk (3)
MkB À g
is the measurement on the balance at step k, mk
where MkB is the mass of the sample (plus gas absorbed) at the end of step k and Bk is the
buoyancy at step k (assuming the balance is tared with no sample mass at P 1⁄4 0).
Rearranging Eqn (3) gives an expression for the mass of the sample plus absorbed gas
þ B (4)
mk 1⁄4 Mk B g k
This can be expressed as
mk 1⁄4 m0 þ Dmk (5)
where m0 is the mass of the sample initially at P 1⁄4 0 and Dmk is the additional mass due to the absorbed gas at pressure step k giving an equation
for the additional mass due to absorption
þ B þ B À m0
mk 1⁄4 m0 þ Dmk 1⁄4 Mk B g k Dmk 1⁄4 Mk B g k
(6)
This is valid for both a traditional double-sided balance and a single-sided balance (no tare side). The buoyancy force is the weight of displaced
gas,
1⁄4 M T
Bkg RZgPksk Vks s
À M k
gP Vt
Tk þ M T
RZkt t RZgPksk Vks x
k(7)
where Mg is the formula weight of the gas used, g is the ac- celeration
due to gravity, Vs and
Vt are
the total volume of components (excluding
is the sample volume at step k, k 1⁄4 ZðPk k Þ
sample) on the sample and tare sides of the balance respectively, VX k P is the system pressure, Zks , T s
and Zk 1⁄4 ZðPk k Þ are and Tk are the temperatures of the
t , T t the compressibilities of the gas in the sample and tare sides of the balance and Tks t
gas in
the sample and tare sides of the balance respectively.
From Eqns (6) and (7), the total uptake (absorbed quantity in moles) of gas at step k can be determined,
Dm
nabsðkÞ 1⁄4 Mg k
Pkk T
1⁄4 RZ sV s
k À Pk
s Vt
Tk þ Pkk T
RZkt t RZ sV x
A full derivation of the gravimetric uptake is given in Ref. [1]. Eqn (8) above can be used for a single-sided balance uptake by setting Vt to zero.
From this and using Eqns (1) and (2), the absolute uncer- tainty in the uptake can be calculated:
ðdnabsðkÞÞ2 1⁄4
((dPPk k
)2
þ
\2
þ
\2)( PkVs
\
2
Tk (
RZks sþ (
dPP
k
(dVs
)Vs 2
þ
k
(dTk s T
s
\2
(
þ (dPPk k
)2
þ
(dVt
)Vt 2
þ
(dTk t Tkt
k
t )2
(dZk t Z
\2
\2 þ
þ
þ
(dZk s
\
dZk s 2)( Pk Vx
\
k 2
Tk (
RZks sþ (
dMM
B
k
kB
)2)(Mk BMg\ 2
( k
þ (dmM
B0
(dMg M
g
\2
þ
)2)(m0Mg(dMg
)2
Mg
(9)
international journal of hydrogen energy 39 (2014) 7158 e7164 7160
Tk , Pk T and k k /RZk Tk are the number of moles of gas displaced by the
It should be noted that PkVs/RZts s Vt/RZkt kt P V x s s sum of the fixed volumes on
the sample side, the sum of the fixed volumes on the tare side and the volume of the sample at step k, respectively.
This provides a relatively simple expression which can be used to calculate the uncertainty in the uptake of gas in the gravimetric method and can
be incorporated into any spreadsheet which calculates uptake.
In Eqn (9) above, the uptake has been calculated in moles of gas. Typically experimental results are reported using either a gas to material ratio
(such as H/M) where the moles of gas absorbed is divided by the moles of sample present, or by a wt % which is the ratio of the mass of gas
absorbed to the mass of the sample plus gas. As the uptake measurements reported in this paper are expressed in wt%, this is shown below.
m
wt% 1⁄4 g
n
100 1⁄4
m0 þ mg  abs  Mg
(
nabs Â
m0 þ Mg) Â 100 (10)
where mg is the mass of absorbed gas and m0 the initial mass of the sample.
Modelling the uncertainty
Isotherm data (values of excess uptake as a function of applied pressure of gas) from a hydrogen adsorption experiment on a well-known activated
carbon (Takeda 4A) were used as a basis for exploring the consequences to the total uptake uncertainty by varying the size of the volumes used as
well as the magnitude of the uncertainties in the measurements. The data were fitted to a To ́th equation [8] and a number of steps were chosen
along the isotherm to specify uptake vs pressure points. These were used in a spreadsheet to derive the required starting conditions of pressure and
weight, which were then used to calculate the uptake (obviously the same as the original points) together with the corresponding uncer- tainty in
the uptake (using Eqn (9)). As the process always generates the original data, this allows the volumes and un- certainties in measurements to be
varied to investigate the effect on the uptake uncertainty.
Results and discussion
The experimental uptake data for hydrogen on the Takeda 4A activated carbon sample are shown in Fig. 2. Eqn (9) was used to calculate the
uncertainties presuming a gravimetric bal- ance with a measurement uncertainty of Æ2 mg and tare and sample side component volumes of 0.240
cc. These values were obtained or measured using a Sartorius MP8 microbal- ance. Other typical values assumed, together with their un-
certainties were
A 1.37 g sample of activated carbon (Takeda 4A) measured
at room temperature. The uncertainty in the pressure measurements was 0.05 bar e appropriate for a pressure transducer calibrated with a quoted
uncertainty of 0.015% of FSR (340 bar). The uncertainty in the temperature measurements was
0.05 K.
he volumes of the components on each side of the bal-
T
ance were calibrated to Æ0.005 cc. The sample density was 1.85 Æ 0.02 g/cc.
The uncertainty in the uptake measurement is shown as “error bars” in the figure and it can be seen that the uncer- tainty increases with increasing
pressure. From Eqn (9), it is clear that the pressure, temperature, volume and compress- ibility uncertainties are amplified by the quantities of gas for the three
volumes of interest e the volume of components on the tare and sample sides of the balance and the volume of the sample. Consequently the uncertainty
increases as the pres- sure increases and the density of gas increases, i.e. as the buoyancy force increases. The maximum uptake is 0.431 Æ 0.018 wt% H2 (a
This uncertainty of 0.018 wt% H2 is to be compared with the corresponding
uncertainty calculation for the Sieverts tech- nique which is shown in Fig.
3, which has a maximum uptake of 0.429 Æ 0.040 wt% H2 (9.3%). The conditions of the Sieverts virtual
experiment are those specified in Ref. [2], which are
indicative of a well-designed apparatus using the best available measuring equipment. These conditions are specified below:
the reference and cell volumes were both 5 cc, the uncertainties in the pressure, temperature and volume measurements were d
P 1⁄4 0.05 bar, dT 1⁄4 0.05
K, dV 1⁄4 0.005 cc and the density of the sample was also assumed to be known to
approximately 1% (dr 1⁄4 0.02 g/cc).
While at low pressures (<50 bar) the uncertainty in the uptake in the Sieverts method (Fig. 3) is smaller than the un- certainty in the uptake for the
gravimetric method, the accu- mulation of uncertainty in the Sieverts technique compounds the uptake uncertainty such that by ca 70 bar, the un- certainties are
similar. After this pressure the uncertainty in the uptake is always lower for the gravimetric method.
This indicates one of the benefits of a method such as the gravimetric technique in which the uncertainties do not accumulate as they do in the
Sieverts technique. The gravi- metric method has two other advantages over the Sieverts technique. The volumes used to calculate the buoyancy are
Fig. 2 e Hydrogen adsorption isotherm at 300 K using the gravimetric technique. The maximum uptake is 0.431 ± 0.018 wt%.
typically small (as they are often just the volumes of hang- down wires and buckets) compared to the volumes needed to prepare gas and contain a sample in
the Sieverts system and this significantly reduces the uncertainty due to the quantity of gas displaced by the volumes of interest. In the gravimetric technique,
the amount of gas in the system volume is irrele- vant to the uncertainty calculation; only the volume of gas displaced by the component volumes features in the
un- certainties. This also reduces the dependence on high preci- sion pressure and temperature measurements as the system is relatively insensitive to these
uncertainties. In addition, due to the nature of the gravimetric method, the weight of the sample is typically known to greater precision. Other factors which
need to be taken into account in this technique include the systematic error in “g” and the effect of higher mass im- purities in the uptake gas.
Obviously, the effect of buoyancy increases with the weight of the gas, so for any gas other than hydrogen the effect of buoyancy will be more significant. For
a calculation on the uptake of argon on the same carbon sample assuming the up- take is the same in wt%, the maximum uptake is 0.43 Æ 0.37 wt% Ar which
has a significantly higher relative uncertainty. However, this is not necessarily a realistic expectation for a different gas as it is a much smaller uptake in terms
of moles of gas e which affects the total uncertainty. If, instead, the uptake is assumed to be the same number of moles of gas, the weight of the uptake will
also be higher. If the same uptake in moles of gas adsorbed is used as for the hydrogen experiment, the maximum uptake is 8.66 Æ 0.32 wt% Ar (a relative
uncer- tainty of 3.7%). This is actually a lower relative uncertainty than the same uptake (in moles of gas) for the case of hydrogen gas. The explanation is
found in Eqn (9), where it can be seen that the molecular weight of the gas is only present as a quantity in the denominator of the last two summation terms in
the absolute uncertainty in the uptake (in moles). Hence for heavier gases, the relative uncertainty decreases slightly (assuming the same quantity of gas is
absorbed).
Fig. 3 e Hydrogen adsorption isotherm at 300 K using the Sieverts method (Vref [ Vcell [ 5 cc, dP [ 0.05, dT [ 0.05, dV [ 0.005, dr [ 0.02). The maximum
uptake value is 0.429 ± 0.040 wt%.
international journal of hydrogen energy 39 (2014) 7158 e7164 7161
The effect of suspended balance component volumes mplifying effect of the gas quantities in these volumes (on the measurement
n- certainties), so where feasible, the amount of component material on each
de of the balance which contributes to a buoyancy force should be kept to a
Aside from the inherent resolution of the balance itself, the corrections due to
minimum.
volumes displaced either side of the bal- ance and the volume displaced by the
sample are factors in the calculation of the uncertainty. For an indication of the Also, from Eqn (9), it can be seen that if the sample side volume is
effect of different size volumes, the uncertainty in the maximum uptake for an gnificantly greater than the tare side volume (or vice versa), the change in the
isotherm has been calculated for a range of volume values while maintaining ptake uncertainty is propor- tional to the sum of the squares of the amounts of
the other values and uncertainties that affect the uptake uncertainty at their as that would occupy or be excluded by these two volumes. Hence, while
nominal values. The maximum uptake typically occurs at the maximumminimising the component volumes is desirable, there are diminishing returns
pressure used for the uptake measurements. In the original adsorption data, rom reducing the volume of com- ponents on one side without also reducing
measurements were taken up to 300 bar, so the uncertainty in the maximum he volume of components on the other side. It is the sum of the squares of the
uptake corre- sponds to the size of the error bars on the last uptake point olumes which needs to be minimised.
(closest to 300 bar) in the isotherms plots such as, for example, Fig. 2. For a single-sided balance, Eqn (9) can be used with Vt 1⁄4 0 , and so
A number of isotherms were calculated each with differenthis apparatus has an advantage in terms of reducing
buoyancy volumes on the balance. For each isotherm, the uncertainty in the
maximum uptake value was plotted as a function of buoyancy volume. Because
it is typical for volumes on each side of a two-sided balance to be very similar,
The above analysis included the buoyancy of the sample, but for a sample with
a well-known density and hence a well- known volume. The uncertainty in the
maximum uptake has been calculated for a range of values of uncertainty in the
sample density, and the results plotted as a function of sample volume
Fig. 4 e Uncertainty in the maximum uptake (ca 3 00 bar) vs balance uncertainty in Fig. 8. Other values such as the component volumes were
component volumes for the gravimetric technique. maintained at their nominal values
the uptake uncertainty by removing one term from the un- certainty expression.
Conclusions
Fig. 7 e Uncertainty in the maximum uptake (ca 3 00 bar) for the gravimetric
international journal of hydrogen energy 39 (2014) 7158 e7164 7163
In summary, the gravimetric method demonstrates lower uncertainties compared to the Sieverts method in the case where relatively low pressures
(<4 00 bar) e but not extremely low pressures (<50 bar) are used, for equivalent
measurement uncertainties and calibration precision.
references