- Email: [email protected]
Calibration of a novel instrument for the investigation of small permeation fluxes of gases through membranes
Measurement 59 (2015) 241–247
Contents lists available at ScienceDirect
Measurement journal homepage: www.elsevier.com/locate/measurement
Calibration of a novel instrument for the investigation of small permeation fluxes of gases through membranes } k ⇑, Ferenc Réti, Gábor Kiss Béla Sebo Budapest University of Technology and Economics, Department of Atomic Physics, Budafoki út 8, 1111 Budapest, Hungary
a r t i c l e
i n f o
Article history: Received 19 January 2014 Received in revised form 16 April 2014 Accepted 19 September 2014 Available online 28 September 2014 Keywords: Vacuum Gas permeation Calibration Reproducibility Membrane transport
a b s t r a c t In this study the calibration and the determination of the sensitivity and reproducibility of a permeation measurement equipment are presented. It is showed that the calibration constant describing the relationship between the material current entering a vacuum chamber through a membrane and the pressure increase over the base pressure in this chamber is dependent on the type of gas used but independent of the membrane temperature. The reproducibility of the calibration constant is approx. ±13%, which can be accepted as the reproducibility of the equipment. The results indicate that the minimum material current required for the determination of the transport parameters in a mounted membrane is in the range of 1011 bmol s1 c and the maximum measurable permeating current is 5 orders of magnitude higher than this. Using various mountable membrane thicknesses this means that the transport properties of a wide range of materials can be investigated with the equipment. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction The investigation of gas permeation through different membranes is of great interest. There are a lot of industrial processes based on the permeation properties of inorganic membranes such as gas purification or separation (e.g. [1–7]). As it is possible to deposit thin metallic layers (such as palladium or palladium alloys) on highly permeable and relatively cheap polymeric membranes mechanically supporting the thin layers (e.g. [8,9]), the permeation properties of metallic as well as polymeric materials are of interest. Despite the interest in the permeation through different materials and the processes involved in the permeation, there are numerous questions left. First of all it is important to acquire reproducible and accurate data concerning the transport of gasses in different materials. Another interesting question is the role of the surface or ⇑ Corresponding author. Tel.: +36 1 4634208; fax: +36 1 4634194. } k). E-mail address: [email protected] (B. Sebo http://dx.doi.org/10.1016/j.measurement.2014.09.057 0263-2241/Ó 2014 Elsevier Ltd. All rights reserved.
surface processes e.g. the effect of surface contaminants (such as an oxide layer on a metal surface) and different layers and treatments on the transport properties of a material. In order to study the effect of surface contaminants and transport properties of metals with low hydrogen permeability a new high sensitivity instrument was constructed and described earlier [10]. As a new instrument raises a number of questions concerning its calibration, sensitivity or the reproducibility of the measured data, the present study tries to address these issues. In case of a high vacuum chamber pumped continuously, the achieved base pressure is the result of a dynamic equilibrium between the pumping speed of the high vacuum pump and the desorption speed of the contaminants from the wall of the chamber. After reaching this dynamic equilibrium, additional material (e.g. due to permeation through a membrane) entering the chamber causes the pressure to rise. In case of pressures lower than approx. 103 mbar, turbomolecular pumps generally have a constant pumping speed (e.g. [11]), meaning that in case of a
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
242
material current (J [mol s1]) entering into a vacuum chamber in dynamic equilibrium (e.g. permeation through a membrane), the pressure increase (Dp [mbar]) in the chamber is expected to be linearly proportional to the material flux entering the chamber (temperature is kept constant) [10]:
J ¼ k Dp
ð1Þ 1
where k bmol s1 mbar c is the proportionality constant. As this constant is not expected to vary with time if other parameters are kept constant, a similar relationship can be assumed to the amount of the permeated material ðQ ½molÞ with the integration of Eq. (1):
Q¼
Z
J dt ¼ k
Z
Dp dt
ð2Þ
The equations presented are the underlying fundamentals of the permeation measurement equipment described earlier [10]. The proportionality constant k is vital as it determines the sensitivity and shows the reproducibility of the measurements. It can be assumed, that k is dependent on the temperature and the type of permeating gas. 2. Materials and methods 2.1. Samples and permeant gases The measurements reported here were conducted through a 0.1 ± 0.02 mm thick polytetrafluoroethylene (PTFE) membrane with a density of 2.1–2.2 g cm3 (KOLOFOL Ltd., Hungary). As PTFE is chemically resistant, heat resistant, readily available and relatively cheap and has good mechanical properties and a relatively high reported permeability to gases [12,13], it is a suitable material to carry out the experiments required for the calibration of the permeation measurement equipment. The following high purity gases were supplied by Messer Hungarogáz Ltd. (Hungary) and used as permeants: He (99.996%), H2 (99.999%), N2 (99.995%), Ar (99.9995%) and CO2 (99.995%).
Fig. 1. The scheme of the high sensitivity permeation measurement setup for the investigation of permeation properties of membranes from [10].
the active surface (the part of the surface which plays a role in the permeation) of the membranes. The vacuum chamber is pumped by a turbomolecular pump and a rotary vane pump. The pressure of the chamber is measured with a hot cathode ionization gauge (MKS 909AR, MKS Instruments Inc. USA). The high pressure chamber is connected to a gas system through which any arbitrary type of gas can be introduced into the chamber and its pressure registered with the aid of an industrial pressure transmitter (Swagelok, USA). The investigated membranes can be heated from the outside to 50–200 °C. The turbomolecular pump and the vacuum chamber can be separated with a valve. If the built in liquid nitrogen cold trap is filled, it is possible to make integral (static method) measurements with even higher sensitivity than dynamic method measurements [10]. Although measurements made according to the static method present a way to conduct more sensitive measurements, they have a drawback as the highest measurable permeating material current is very limited. As a PTFE membrane was chosen to calibrate the equipment, the relatively high amount of permeating gas rendered the application of the static method impractical.
2.2. Permeation measurement system 2.3. Calibration of the instrument As the newly built permeation measurement equipment was described in detail elsewhere [10], only a brief description is given here. The system consists of a high vacuum chamber (base pressure approx. 7 109 mbar) separated by the membrane under investigation from a gastight high pressure chamber with a known volume of 4:84 0:24 cm3 (Fig. 1). The membrane is supported by a plate covered by a fine stainless steel mesh and holes drilled in it, in order to let through the permeated gas. The role of this plate is to prevent the rupture of the investigated membrane caused by the pressure difference between the vacuum chamber and the high pressure chamber. The mounting system was tested with metal and polymeric membranes of different thicknesses between 10 lm and 125 lm in order to provide a gastight mounting which does not cause ruptures and maintains
In order to determine the calibration constant k (Eqs. (1) and (2)), the material current entering the chamber through the PTFE membrane and the resulting pressure increase had to be measured. The process of a calibration measurement at room temperature with H2 gas is shown in Fig. 2. Initially before each measurement, the membrane was empty of permeant, the temperature was held constant and the pressure in the vacuum chamber reached the dynamic equilibrium corresponding to the base pressure of the chamber at the given temperature. After the permeating gas is first introduced to the high pressure chamber its pressure is kept constant. As the permeating flux starts to build up, the pressure in the vacuum chamber increases. Reaching the steady-state permeation, a new dynamic equilibrium pressure is achieved in the vacuum
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
243
Fig. 2. Calibration measurement through a 0.1 mm thick PTFE membrane at room temperature with H2 gas. (a) Shows the evolution of the monitored pressures in the high pressure chamber and the vacuum chamber during the whole measurement with the fitted straight line for the evaluation of the pressure decrease in the high pressure chamber. (b) Shows the pressure increase in the vacuum chamber at the beginning of the experiment.
chamber (higher than the initial base pressure). From the speed of the pressure decrease in the gastightly closed high pressure chamber and the known volume of the chamber, the permeating flux can be determined. The disadvantage of this process is that after the valve at the high pressure chamber is closed, the permeating flux also decreases with decreasing pressure in the high pressure chamber which in turn leads to decreasing pressure in the vacuum chamber. In practice due to the limited sensitivity of the vacuum gauge this pressure decrease was usually not detectable (Fig. 2(a), upper graph, between approx. 750 s and 2000 s). In order to overcome this difficulty the initial steady-state pressure increase was always recorded as the pressure increase of the vacuum chamber (Fig. 2(a), approx. 750 s). In case of the high pressure chamber a straight line was fitted to the pressure vs. time graph and the slope of this line was assumed to be the speed of the pressure decrease (Fig. 2(a) lower graph between approx. 750 s and 2000 s). As the permeating flux is
directly proportional to the pressure above the PTFE membrane [14] the rate of the pressure decrease in the high pressure chamber is expected to become slower and the pressure vs. time graph to deviate from a straight line. As the overall pressure decrease was fairly small, the recorded graphs could be fitted with a straight line (Fig. 2(a) upper graph between approx. 750 s and 2000 s). For the calculation of the material current from the rate of pressure decrease in the high pressure chamber the ideal gas law was used (Eq. (3)). As the investigated gases are real gases the ideal gas law is not entirely applicable to describe their behaviour. The margin of error is estimated to be lower than 5% even in the worst case scenario appearing in this study (room temperature, 5.5 bar pressure, CO2) based on the comparison of calculations using the ideal gas law and the Van der Waals equation.
J¼
dn V dp ¼ dt RT g dt
ð3Þ
244
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
Here J is the material current entering the vacuum chamber through the membrane, V bm3 c is the volume of the high pressure chamber, dp=dt bbar s1 c is the rate of pressure decrease in the high pressure chamber, 1 R ¼ 8:314bJ mol K1 c and T g ½K is the absolute gas temperature in the high pressure chamber. As the pressure transducer at the high pressure chamber has to be kept below 80 °C it has to be actively cooled, the high pressure gas system is therefore partially at room temperature at all times, even during measurements made at elevated temperatures. As a result the temperature in Eq. (3) can differ from the temperature of the membrane and it is defined as an ‘average gas temperature’ between the room temperature and the membrane temperature. In order to measure this temperature a practically non permeable aluminium plate was placed in the system and the pressure in the high pressure chamber was measured at room temperature and at different plate (membrane) temperatures. With the aid of the pressure ratios the ‘average gas temperatures’ were calculated for the different gases and temperatures and the calculation of the calibration constants was carried out with these temperatures. The calibration process was repeated at different pressures in the high pressure chamber and with different membrane temperatures for each gas in order to map the parameter dependence of k. 3. Results and discussion 3.1. Calibration constant in case of different permeating gases Fig. 3 shows the measured permeating fluxes and the corresponding pressure increases in the vacuum chamber at room temperature (25 °C) using H2 as permeating gas which was admitted into the high pressure chamber at different pressures. The calibration constant can be determined by fitting a straight line on the measured points running through the origin as in case of zero permeating flux
Fig. 3. Calculation of the calibration constant (k) by fitting a straight line through the origin on the pressure increase in the vacuum chamber (Dp) permeating flux (J) data points measured during steady-state permeation. The different permeating flux values correspond to different gas pressures in the high pressure chamber.
the pressure increase above the base pressure is zero. The slope of the fitted line is k the calibration constant (Eq. (1)). The permeating flux–pressure increase relationships in case of other permeating gases can be handled similarly and can also be fitted with a straight line (not shown here). Table 1 and Fig. 4 summarize the calibration constants calculated in case of different permeating gases. The calibration constant (k) is dependent on the type of gas used. This can be explained by taking into account that the detector used in the measurement equipment is a hot cathode ionization gauge, whose output is dependent on the gas present in the vacuum chamber. It is calibrated to nitrogen, but when working with other gases Gas Calibration Factors (GCF) can be used:
preal ¼
pshown GCF
ð4Þ
The meaning of Eq. (4) is that to a given gas flux entering into the chamber, different types of gas will show different pressure increases (even assuming equal pumping speeds), with higher pressure increases and smaller calibration constants corresponding to higher GCF (Eq. (1)). Assuming inverse proportionality between the calibration constant and the GCF, their product should be the same in case of each gas. Although the products are relatively close to each other (Fig. 4), their values differ by around 40%, showing that something else is influencing the value of the calibration constant especially in case of lighter gases. Another possible affecting factor is the pumping speed of the turbomolecular pump. Due to the working principles of these pumps, their pumping speed for light particles is considerably lower than their pumping speed for particles with higher mass (which is almost constant). This can explain the deviations in the product of the calibration constants and the gas calibration factors in case of gases having light particles. The effect of pumping speed could have been investigated with varying the open diameter of the flange through which the chamber is pumped. As it was not possible in this setup another approach was used. The electronic control unit of a turbomolecular pump is usually equipped with an option lowering the rotation speed of the pump (stand-by mode) in order to elongate the lifetime of the bearing or to protect the pump from overheating etc. Measurements made with the different gases in stand-by mode of the turbomolecular pump (66% of nominal rotational speed) showed that the calibration constant (k) is dependent on the pumping speed of the pump. In case of the gases with lower molecular (atomic) weight this effect is more pronounced (Fig. 5). A possible explanation can be given by considering the different thermal velocities of the particles. The lighter particles are travelling faster at the same temperature, meaning that they have less chance colliding with the blades of the rotor in the turbomolecular pump, which is why the these pumps are generally less effective in pumping light gases. In case of a lowered angular velocity the blades can still hit the slower particles (although with slightly lower effectiveness) but the faster particles can slip through them causing less effective pumping and the
}k et al. / Measurement 59 (2015) 241–247 B. Sebo Table 1 Calibration constants for different gases measured at room temperature. Permeating gas
Calibration constant (k) measured at room temperature s mol mbar
He H2 N2 Ar CO2
1.4725 102 ± 9.5 105 5.354 103 ± 5.9 105 1.941 103 ± 5.5 105 1.319 103 ± 3.2 105 1.155 103 ± 1.3 105
Fig. 4. Calibration constants (k) measured at room temperature for different gases along with the Gas Calibration Factor (GCF) of the hot cathode ionization gauge and their product. In case the calibration constant is inversely proportional to the Gas Calibration Factor (GCF) their product should be constant.
245
efficiently. Calculating the transport parameters from the data acquired during these measurements also showed, that the difference of the rotational speed has no effect on the calculated values. 3.2. Reproducibility of the measured data Another important property of a measurement equipment is the reproducibility of the measurements made with the equipment, for which the reproducibility of the calibration constant can be a good indication. While collecting the experimental data for this paper, calibration experiments were carried out with H2 gas on various days at room temperature with a given pressure in the high pressure chamber which added up to a total of 50 measurements over a 135 day period. Next to this, after 78 days another measurement cycle with 9 different pressures was done. During this 135 day period the vacuum chamber was heated, the pumps were stopped and restarted and generally every event occurred which can be considered part of the normal operation for the equipment. The results of these measurements are summarized in Table 2. The results indicate that the calibration constant measured initially and almost 80 days later as well as the average of the measurements made during a 135 day period deviate from each other by less than 5%, which is less than the claimed reproducibility of the hot cathode ionization gauge (±5%). Considering the relatively high standard deviation of the measured values during the 135 day long time period along with a 95% confidence interval (2 times the standard deviation) the reproducibility of the device can be determined as approx. ±13%. 3.3. Temperature dependence of the calibration constant
Fig. 5. Difference in the calibration constant (k) with lowered rotational speed of the turbomolecular pump (stand-by mode, 66% nominal rotational speed) measured at room temperature and 4.5 bar absolute pressure in the high pressure chamber expressed in percentage compared to the values measured under the same conditions but at nominal rotational speed. The dashed line indicates the decrease in the rotational speed of the rotor in the turbomolecular pump (34%).
decrease of the calibration constant (k). During normal operation (such as operating the pump in stand-by mode) this does not present a problem as the main components of air have relatively high masses and are still pumped
The pumping speed of a vacuum pump is usually given in volume of gas pumped away in a given time. This means that the number of gas particles pumped away in a given time should be dependent on the gas temperature which in turn should be dependent on the membrane temperature (and the heating). The calibration constant could therefore depend on the membrane temperature. In order to determine the calibration constant-membrane temperature relationship, the calibration process was repeated for each of the gases at different temperatures in 10 °C steps from 50 °C to 140 °C. The results in case of N2 (gas with molecules of two atoms) and Ar (gas in atomic form) are shown in Table 3 and in Fig. 6. In case of both gases the maximum and minimum values differ by less than the reproducibility of the equipment (Table 3), the other gases show similar behaviour (results not shown here). Considering that the measured calibration constants (k) do not show a trend with increasing membrane temperature and the majority of the measured values differ by less than 5% it can be assumed, that the calibration constant is independent of the membrane temperature. A possible explanation can be given by considering the geometry of the membrane heating and the vacuum chamber. There is a liquid nitrogen cold trap in the chamber which is only filled for static method mea-
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
246
Table 2 Reproducibility of the calibration constant. All values measured at room temperature with H2 gas. Initially (9 different pressures) mol Calibration constant (k) for H2 mbar s Deviation from the initial value (%) *
78 days later (9 different pressures)
Average of 50 measurements during a 135 day period
5.354 103 ± 5.9 105
5.154 103 ± 4.5 105
5.243 103 ± 3.4 104*
–
3.7
2.1
The error given here is the standard deviation of the data.
Table 3 Calibration constant (k) measured at different temperatures in case of N2 and Ar. Temperature (°C)
Calibration constant (k) of N2 s mol mbar
Calibration constant (k) of Ar s mol mbar
Room temperature 50 60 70 80 90 100 110 120 130 140
1.941 103 ± 5.5 105 1.835 103 ± 3.3 105 1.949 103 ± 4.6 105 2.05 103 ± 3.2 105 2.014 103 ± 9.9 105 2.006 103 ± 2.7 105 1.994 103 ± 1.8 105 2.046 103 ± 2.3 105 2.034 103 ± 2.6 105 2.018 103 ± 2.8 105 1.967 103 ± 1.6 105
1.319 103 ± 3.2 105 1.305 103 ± 1.7 105 1.296 103 ± 1.1 105 1.278 103 ± 2 105 1.315 103 ± 4.4 106 1.357 103 ± 8.1 106 1.326 103 ± 5.2 106 1.334 103 ± 3.3 106 1.328 103 ± 5.2 106 1.331 103 ± 1.3 105 1.331 103 ± 6.8 106
3.4. Sensitivity of the new permeation measurement equipment
Fig. 6. Calibration constant (k) measured at different temperatures in case of N2 and Ar. The deviations in the values are less than the reproducibility of the measurement setup.
surements and not heated by the membrane heating. It can thus be assumed to be always at room temperature during the calibration measurements as well as the inlet of the turbomolecular pump and the vacuum chamber section right above. Because of the geometry of the chamber there is no direct straight path from the flange where the particles enter the chamber and the inlet of the pump meaning that the particles entering have to collide with the walls before reaching the turbomolecular pump. As the walls with which the gas molecules collide before entering the pump are at room temperature, the gas temperature in the vacuum chamber at the turbomolecular pump can be assumed to be room temperature which is independent of the membrane temperature leading to calibration constants independent of membrane temperature.
Knowing the values of the calibration constants, the detection limit for the permeation of different gases can be estimated based on the background pressure of the vacuum chamber and the minimum reliably detectable pressure increase. Another limit is applicable for the determination of the transport parameters of the gases in a given membrane material as the calculation of diffusivity, permeability and solubility requires the measurement of the build-up of the permeation flux with adequate time and pressure resolution (see the time-lag method (e.g. [15,16])). Given that the background pressure of the vacuum chamber is approx. 7 109 mbar a pressure increase of 1 109 mbar can already be dependably detected. For the determination of the transport parameters the build-up of the permeation flux should be adequately measurable as the integration of the measured pressure increase vs. time (Fig. 2/b) curve is required. In order to achieve a reasonably accurate integrated curve, the pressure increase vs. time curve should have a reasonable number of points and a reasonable time and pressure resolution. To achieve this as a conservative assumption a higher, approx. 1 108 mbar pressure increase is required. The detection and measurement limits for the different gases can be calculated by multiplying the corresponding k value with the pressure increases mentioned above. The resulting limits are summarized in Table 4. The upper limit of the measureable permeating fluxes can also be estimated as the pumping speed of the turbomolecular pumps above 103 mbar pressure is not linear anymore, meaning that the calibration and measurements become difficult. Furthermore, the turbomolecular pump should not be operated above a given pressure. It can thus be concluded that in case of
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
247
Table 4 Estimated minimum detectable permeation fluxes, minimum permeation fluxes for the determination of transport parameters and the maximum measurable permeating fluxes for different gases. The above values can be considered as a conservative assumption and they were calculated by multiplying the k values corresponding to the different gases with the pressures differences given in the text. Permeating gas
Min. detectable mol
Min. flux for transport
1.5 1011 5.4 1012 1.9 1012 1.3 1012 1.2 1012
1.5 1010 5.4 1011 1.9 1011 1.3 1011 1.2 1011
perm. flux He H2 N2 Ar CO2
s
steady-state permeation the equipment should not be used above 1 103 mbar pressure in the vacuum chamber, which gives the maximum measurable permeating flux for the different gases by multiplication with the respective k value (Table 4). It is worth noting, that the minimum permeating flux required for the determination of the transport parameters and the upper limit of the practically measurable permeating flux differ by 5 orders of magnitude. Considering the wide range of usable membrane thicknesses [10] in the measurement setup this means that the transport properties of a wide range of materials with very different transport properties can be investigated with the aid of the equipment. 4. Conclusion In this study the calibration and the determination of the sensitivity and reproducibility of a new permeation equipment described earlier [10] are presented. It is shown that the calibration constant describing the relationship between the material current entering the vacuum chamber through a membrane and the pressure increase over the base pressure in the vacuum chamber is dependent on the type of gas used but independent of the membrane temperature. The latter can be explained by the ‘thermalisation’ of the gas molecules through collision with the walls of the vacuum chamber which remain at room temperature even when the membrane is heated. The type of permeating gas influences the calibration constant through the different ionisation probabilities, thus through the different gas correction factors of the hot cathode pressure gauge used as the detector and through the different pumping speeds of the turbomolecular pump. The calibration constants measured over a longer time period are indicating the reproducibility of the data obtained with the permeation measurement equipment. These measurements showed that the deviation in the values of the determined calibration constants is less than 5%. Considering the standard deviation of the measured values during a 135 day long time period and with a 95% confidence interval (2 times the standard deviation) the reproducibility of the device can be determined as approx. ±13%. With the aid of the calibration constant the minimum detectable (sensitivity) and the maximum measurable permeating material current can also be calculated for
param. determination
mol s
Upper limit of the permeating flux mol s 1.5 105 5.4 106 1.9 106 1.3 106 1.2 106
the different gases used in the calibration measurements. The results showed that the minimum detectable permeating material current and the minimum material current required for the determination of the transport parameters is in the range of 1011 bmol s1 c and the maximum measurable permeating current is 5 orders of magnitude higher than this. Along with the large variety of mountable membrane thicknesses and materials this means that the transport properties of a wide range of materials can be investigated. References [1] J.P. Wu, I.W.M. Brown, M.E. Bowden, T. Kemmitt, Palladium coated porous anodic alumina membranes for gas reforming processes, Solid State Sci. 12 (2010) 1912–1916. [2] N. Itoh, N. Tomura, T. Tsuji, M. Hongo, Deposition of palladium inside straight mesopores of anodic alumina tube and its hydrogen permeability, Micropor. Mesopor. Mater. 39 (2000) 103–111. [3] T. Ozaki, Y. Zhang, M. Komaki, C. Nishimura, Hydrogen permeation characteristics of V–Ni–Al alloys, Int. J. Hydrogen Energy 28 (2003) 1229–1235. [4] N.M. Peachey, R.C. Snow, R.C. Dye, Composite Pd/Ta metal membranes for hydrogen separation, J. Membr. Sci. 111 (1996) 123–133. [5] L.C. Witjens, J.H. Bitter, A.J. van Dillen, W.M. Arnoldbik, F.H.P.M. Habraken, K.P. de Jong, Improving the control of the electroless plating synthesis of Pd/Ag membranes for hydrogen separation using Rutherford backscattering, J. Membr. Sci. 254 (2005) 241–248. [6] T.S. Moss, N.M. Peachey, R.C. Snow, R.C. Dye, Multilayer metal membranes for hydrogen separation, Int. J. Hydrogen Energy 23 (1998) 99–106. [7] Y. Zhang, T. Ozaki, M. Komaki, C. Nishimura, Hydrogen permeation characteristics of vanadium–aluminium alloys, Scripta Mater. 47 (2002) 601–606. [8] A.L. Athayde, R.W. Baker, P. Nguyen, Metal composite membranes for hydrogen separation, J. Membr. Sci. 94 (1994) 299–311. [9] P.V. Mercea, D. Silipasß, V. Mecea, Separation of a gas mixture through a polymer membrane metallized with palladium, Gas Sep. Purif. 4 (1990) 137–140. }k, G. Kiss, G. Dobos, F. Réti, T. Majoros, O.H. Krafcsik, Novel [10] B. Sebo instrument for the investigation of small permeation fluxes of gases through different membranes, Measurement 46 (2013) 3516–3524. [11] K. Jousten, Handbook of Vacuum Technology, John Wiley & Sons, Berlin, 2008. [12] M. Matsuyama, H. Miyake, K. Asiida, K. Wantanabe, Permeation, diffusion and dissolution of hydrogen isotopes, methane and inert gases through/in a tetrafluoroethylene film, J. Nucl. Mater. 110 (1982) 296–300. [13] R.A. Pasternak, M.V. Christensen, J. Heller, Diffusion and permeation of oxygen, nitrogen, carbon dioxide and nitrogen dioxide through polytetrafluoroethylene, Macromolecules 3 (1970) 366–371. [14] J. Crank, G.S. Park, Diffusion in Polymer, Academic Press Inc., London, 1968. [15] H.L. Frisch, The time lag in diffusion, J. Phys. Chem. 61 (1957) 93–95. [16] S.W. Rutherford, D.D. Do, Review of time lag permeation technique as a method for characterisation of porous media and membranes, Adsorption 3 (1997) 283–312.
Contents lists available at ScienceDirect
Measurement journal homepage: www.elsevier.com/locate/measurement
Calibration of a novel instrument for the investigation of small permeation fluxes of gases through membranes } k ⇑, Ferenc Réti, Gábor Kiss Béla Sebo Budapest University of Technology and Economics, Department of Atomic Physics, Budafoki út 8, 1111 Budapest, Hungary
a r t i c l e
i n f o
Article history: Received 19 January 2014 Received in revised form 16 April 2014 Accepted 19 September 2014 Available online 28 September 2014 Keywords: Vacuum Gas permeation Calibration Reproducibility Membrane transport
a b s t r a c t In this study the calibration and the determination of the sensitivity and reproducibility of a permeation measurement equipment are presented. It is showed that the calibration constant describing the relationship between the material current entering a vacuum chamber through a membrane and the pressure increase over the base pressure in this chamber is dependent on the type of gas used but independent of the membrane temperature. The reproducibility of the calibration constant is approx. ±13%, which can be accepted as the reproducibility of the equipment. The results indicate that the minimum material current required for the determination of the transport parameters in a mounted membrane is in the range of 1011 bmol s1 c and the maximum measurable permeating current is 5 orders of magnitude higher than this. Using various mountable membrane thicknesses this means that the transport properties of a wide range of materials can be investigated with the equipment. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction The investigation of gas permeation through different membranes is of great interest. There are a lot of industrial processes based on the permeation properties of inorganic membranes such as gas purification or separation (e.g. [1–7]). As it is possible to deposit thin metallic layers (such as palladium or palladium alloys) on highly permeable and relatively cheap polymeric membranes mechanically supporting the thin layers (e.g. [8,9]), the permeation properties of metallic as well as polymeric materials are of interest. Despite the interest in the permeation through different materials and the processes involved in the permeation, there are numerous questions left. First of all it is important to acquire reproducible and accurate data concerning the transport of gasses in different materials. Another interesting question is the role of the surface or ⇑ Corresponding author. Tel.: +36 1 4634208; fax: +36 1 4634194. } k). E-mail address: [email protected] (B. Sebo http://dx.doi.org/10.1016/j.measurement.2014.09.057 0263-2241/Ó 2014 Elsevier Ltd. All rights reserved.
surface processes e.g. the effect of surface contaminants (such as an oxide layer on a metal surface) and different layers and treatments on the transport properties of a material. In order to study the effect of surface contaminants and transport properties of metals with low hydrogen permeability a new high sensitivity instrument was constructed and described earlier [10]. As a new instrument raises a number of questions concerning its calibration, sensitivity or the reproducibility of the measured data, the present study tries to address these issues. In case of a high vacuum chamber pumped continuously, the achieved base pressure is the result of a dynamic equilibrium between the pumping speed of the high vacuum pump and the desorption speed of the contaminants from the wall of the chamber. After reaching this dynamic equilibrium, additional material (e.g. due to permeation through a membrane) entering the chamber causes the pressure to rise. In case of pressures lower than approx. 103 mbar, turbomolecular pumps generally have a constant pumping speed (e.g. [11]), meaning that in case of a
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
242
material current (J [mol s1]) entering into a vacuum chamber in dynamic equilibrium (e.g. permeation through a membrane), the pressure increase (Dp [mbar]) in the chamber is expected to be linearly proportional to the material flux entering the chamber (temperature is kept constant) [10]:
J ¼ k Dp
ð1Þ 1
where k bmol s1 mbar c is the proportionality constant. As this constant is not expected to vary with time if other parameters are kept constant, a similar relationship can be assumed to the amount of the permeated material ðQ ½molÞ with the integration of Eq. (1):
Q¼
Z
J dt ¼ k
Z
Dp dt
ð2Þ
The equations presented are the underlying fundamentals of the permeation measurement equipment described earlier [10]. The proportionality constant k is vital as it determines the sensitivity and shows the reproducibility of the measurements. It can be assumed, that k is dependent on the temperature and the type of permeating gas. 2. Materials and methods 2.1. Samples and permeant gases The measurements reported here were conducted through a 0.1 ± 0.02 mm thick polytetrafluoroethylene (PTFE) membrane with a density of 2.1–2.2 g cm3 (KOLOFOL Ltd., Hungary). As PTFE is chemically resistant, heat resistant, readily available and relatively cheap and has good mechanical properties and a relatively high reported permeability to gases [12,13], it is a suitable material to carry out the experiments required for the calibration of the permeation measurement equipment. The following high purity gases were supplied by Messer Hungarogáz Ltd. (Hungary) and used as permeants: He (99.996%), H2 (99.999%), N2 (99.995%), Ar (99.9995%) and CO2 (99.995%).
Fig. 1. The scheme of the high sensitivity permeation measurement setup for the investigation of permeation properties of membranes from [10].
the active surface (the part of the surface which plays a role in the permeation) of the membranes. The vacuum chamber is pumped by a turbomolecular pump and a rotary vane pump. The pressure of the chamber is measured with a hot cathode ionization gauge (MKS 909AR, MKS Instruments Inc. USA). The high pressure chamber is connected to a gas system through which any arbitrary type of gas can be introduced into the chamber and its pressure registered with the aid of an industrial pressure transmitter (Swagelok, USA). The investigated membranes can be heated from the outside to 50–200 °C. The turbomolecular pump and the vacuum chamber can be separated with a valve. If the built in liquid nitrogen cold trap is filled, it is possible to make integral (static method) measurements with even higher sensitivity than dynamic method measurements [10]. Although measurements made according to the static method present a way to conduct more sensitive measurements, they have a drawback as the highest measurable permeating material current is very limited. As a PTFE membrane was chosen to calibrate the equipment, the relatively high amount of permeating gas rendered the application of the static method impractical.
2.2. Permeation measurement system 2.3. Calibration of the instrument As the newly built permeation measurement equipment was described in detail elsewhere [10], only a brief description is given here. The system consists of a high vacuum chamber (base pressure approx. 7 109 mbar) separated by the membrane under investigation from a gastight high pressure chamber with a known volume of 4:84 0:24 cm3 (Fig. 1). The membrane is supported by a plate covered by a fine stainless steel mesh and holes drilled in it, in order to let through the permeated gas. The role of this plate is to prevent the rupture of the investigated membrane caused by the pressure difference between the vacuum chamber and the high pressure chamber. The mounting system was tested with metal and polymeric membranes of different thicknesses between 10 lm and 125 lm in order to provide a gastight mounting which does not cause ruptures and maintains
In order to determine the calibration constant k (Eqs. (1) and (2)), the material current entering the chamber through the PTFE membrane and the resulting pressure increase had to be measured. The process of a calibration measurement at room temperature with H2 gas is shown in Fig. 2. Initially before each measurement, the membrane was empty of permeant, the temperature was held constant and the pressure in the vacuum chamber reached the dynamic equilibrium corresponding to the base pressure of the chamber at the given temperature. After the permeating gas is first introduced to the high pressure chamber its pressure is kept constant. As the permeating flux starts to build up, the pressure in the vacuum chamber increases. Reaching the steady-state permeation, a new dynamic equilibrium pressure is achieved in the vacuum
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
243
Fig. 2. Calibration measurement through a 0.1 mm thick PTFE membrane at room temperature with H2 gas. (a) Shows the evolution of the monitored pressures in the high pressure chamber and the vacuum chamber during the whole measurement with the fitted straight line for the evaluation of the pressure decrease in the high pressure chamber. (b) Shows the pressure increase in the vacuum chamber at the beginning of the experiment.
chamber (higher than the initial base pressure). From the speed of the pressure decrease in the gastightly closed high pressure chamber and the known volume of the chamber, the permeating flux can be determined. The disadvantage of this process is that after the valve at the high pressure chamber is closed, the permeating flux also decreases with decreasing pressure in the high pressure chamber which in turn leads to decreasing pressure in the vacuum chamber. In practice due to the limited sensitivity of the vacuum gauge this pressure decrease was usually not detectable (Fig. 2(a), upper graph, between approx. 750 s and 2000 s). In order to overcome this difficulty the initial steady-state pressure increase was always recorded as the pressure increase of the vacuum chamber (Fig. 2(a), approx. 750 s). In case of the high pressure chamber a straight line was fitted to the pressure vs. time graph and the slope of this line was assumed to be the speed of the pressure decrease (Fig. 2(a) lower graph between approx. 750 s and 2000 s). As the permeating flux is
directly proportional to the pressure above the PTFE membrane [14] the rate of the pressure decrease in the high pressure chamber is expected to become slower and the pressure vs. time graph to deviate from a straight line. As the overall pressure decrease was fairly small, the recorded graphs could be fitted with a straight line (Fig. 2(a) upper graph between approx. 750 s and 2000 s). For the calculation of the material current from the rate of pressure decrease in the high pressure chamber the ideal gas law was used (Eq. (3)). As the investigated gases are real gases the ideal gas law is not entirely applicable to describe their behaviour. The margin of error is estimated to be lower than 5% even in the worst case scenario appearing in this study (room temperature, 5.5 bar pressure, CO2) based on the comparison of calculations using the ideal gas law and the Van der Waals equation.
J¼
dn V dp ¼ dt RT g dt
ð3Þ
244
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
Here J is the material current entering the vacuum chamber through the membrane, V bm3 c is the volume of the high pressure chamber, dp=dt bbar s1 c is the rate of pressure decrease in the high pressure chamber, 1 R ¼ 8:314bJ mol K1 c and T g ½K is the absolute gas temperature in the high pressure chamber. As the pressure transducer at the high pressure chamber has to be kept below 80 °C it has to be actively cooled, the high pressure gas system is therefore partially at room temperature at all times, even during measurements made at elevated temperatures. As a result the temperature in Eq. (3) can differ from the temperature of the membrane and it is defined as an ‘average gas temperature’ between the room temperature and the membrane temperature. In order to measure this temperature a practically non permeable aluminium plate was placed in the system and the pressure in the high pressure chamber was measured at room temperature and at different plate (membrane) temperatures. With the aid of the pressure ratios the ‘average gas temperatures’ were calculated for the different gases and temperatures and the calculation of the calibration constants was carried out with these temperatures. The calibration process was repeated at different pressures in the high pressure chamber and with different membrane temperatures for each gas in order to map the parameter dependence of k. 3. Results and discussion 3.1. Calibration constant in case of different permeating gases Fig. 3 shows the measured permeating fluxes and the corresponding pressure increases in the vacuum chamber at room temperature (25 °C) using H2 as permeating gas which was admitted into the high pressure chamber at different pressures. The calibration constant can be determined by fitting a straight line on the measured points running through the origin as in case of zero permeating flux
Fig. 3. Calculation of the calibration constant (k) by fitting a straight line through the origin on the pressure increase in the vacuum chamber (Dp) permeating flux (J) data points measured during steady-state permeation. The different permeating flux values correspond to different gas pressures in the high pressure chamber.
the pressure increase above the base pressure is zero. The slope of the fitted line is k the calibration constant (Eq. (1)). The permeating flux–pressure increase relationships in case of other permeating gases can be handled similarly and can also be fitted with a straight line (not shown here). Table 1 and Fig. 4 summarize the calibration constants calculated in case of different permeating gases. The calibration constant (k) is dependent on the type of gas used. This can be explained by taking into account that the detector used in the measurement equipment is a hot cathode ionization gauge, whose output is dependent on the gas present in the vacuum chamber. It is calibrated to nitrogen, but when working with other gases Gas Calibration Factors (GCF) can be used:
preal ¼
pshown GCF
ð4Þ
The meaning of Eq. (4) is that to a given gas flux entering into the chamber, different types of gas will show different pressure increases (even assuming equal pumping speeds), with higher pressure increases and smaller calibration constants corresponding to higher GCF (Eq. (1)). Assuming inverse proportionality between the calibration constant and the GCF, their product should be the same in case of each gas. Although the products are relatively close to each other (Fig. 4), their values differ by around 40%, showing that something else is influencing the value of the calibration constant especially in case of lighter gases. Another possible affecting factor is the pumping speed of the turbomolecular pump. Due to the working principles of these pumps, their pumping speed for light particles is considerably lower than their pumping speed for particles with higher mass (which is almost constant). This can explain the deviations in the product of the calibration constants and the gas calibration factors in case of gases having light particles. The effect of pumping speed could have been investigated with varying the open diameter of the flange through which the chamber is pumped. As it was not possible in this setup another approach was used. The electronic control unit of a turbomolecular pump is usually equipped with an option lowering the rotation speed of the pump (stand-by mode) in order to elongate the lifetime of the bearing or to protect the pump from overheating etc. Measurements made with the different gases in stand-by mode of the turbomolecular pump (66% of nominal rotational speed) showed that the calibration constant (k) is dependent on the pumping speed of the pump. In case of the gases with lower molecular (atomic) weight this effect is more pronounced (Fig. 5). A possible explanation can be given by considering the different thermal velocities of the particles. The lighter particles are travelling faster at the same temperature, meaning that they have less chance colliding with the blades of the rotor in the turbomolecular pump, which is why the these pumps are generally less effective in pumping light gases. In case of a lowered angular velocity the blades can still hit the slower particles (although with slightly lower effectiveness) but the faster particles can slip through them causing less effective pumping and the
}k et al. / Measurement 59 (2015) 241–247 B. Sebo Table 1 Calibration constants for different gases measured at room temperature. Permeating gas
Calibration constant (k) measured at room temperature s mol mbar
He H2 N2 Ar CO2
1.4725 102 ± 9.5 105 5.354 103 ± 5.9 105 1.941 103 ± 5.5 105 1.319 103 ± 3.2 105 1.155 103 ± 1.3 105
Fig. 4. Calibration constants (k) measured at room temperature for different gases along with the Gas Calibration Factor (GCF) of the hot cathode ionization gauge and their product. In case the calibration constant is inversely proportional to the Gas Calibration Factor (GCF) their product should be constant.
245
efficiently. Calculating the transport parameters from the data acquired during these measurements also showed, that the difference of the rotational speed has no effect on the calculated values. 3.2. Reproducibility of the measured data Another important property of a measurement equipment is the reproducibility of the measurements made with the equipment, for which the reproducibility of the calibration constant can be a good indication. While collecting the experimental data for this paper, calibration experiments were carried out with H2 gas on various days at room temperature with a given pressure in the high pressure chamber which added up to a total of 50 measurements over a 135 day period. Next to this, after 78 days another measurement cycle with 9 different pressures was done. During this 135 day period the vacuum chamber was heated, the pumps were stopped and restarted and generally every event occurred which can be considered part of the normal operation for the equipment. The results of these measurements are summarized in Table 2. The results indicate that the calibration constant measured initially and almost 80 days later as well as the average of the measurements made during a 135 day period deviate from each other by less than 5%, which is less than the claimed reproducibility of the hot cathode ionization gauge (±5%). Considering the relatively high standard deviation of the measured values during the 135 day long time period along with a 95% confidence interval (2 times the standard deviation) the reproducibility of the device can be determined as approx. ±13%. 3.3. Temperature dependence of the calibration constant
Fig. 5. Difference in the calibration constant (k) with lowered rotational speed of the turbomolecular pump (stand-by mode, 66% nominal rotational speed) measured at room temperature and 4.5 bar absolute pressure in the high pressure chamber expressed in percentage compared to the values measured under the same conditions but at nominal rotational speed. The dashed line indicates the decrease in the rotational speed of the rotor in the turbomolecular pump (34%).
decrease of the calibration constant (k). During normal operation (such as operating the pump in stand-by mode) this does not present a problem as the main components of air have relatively high masses and are still pumped
The pumping speed of a vacuum pump is usually given in volume of gas pumped away in a given time. This means that the number of gas particles pumped away in a given time should be dependent on the gas temperature which in turn should be dependent on the membrane temperature (and the heating). The calibration constant could therefore depend on the membrane temperature. In order to determine the calibration constant-membrane temperature relationship, the calibration process was repeated for each of the gases at different temperatures in 10 °C steps from 50 °C to 140 °C. The results in case of N2 (gas with molecules of two atoms) and Ar (gas in atomic form) are shown in Table 3 and in Fig. 6. In case of both gases the maximum and minimum values differ by less than the reproducibility of the equipment (Table 3), the other gases show similar behaviour (results not shown here). Considering that the measured calibration constants (k) do not show a trend with increasing membrane temperature and the majority of the measured values differ by less than 5% it can be assumed, that the calibration constant is independent of the membrane temperature. A possible explanation can be given by considering the geometry of the membrane heating and the vacuum chamber. There is a liquid nitrogen cold trap in the chamber which is only filled for static method mea-
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
246
Table 2 Reproducibility of the calibration constant. All values measured at room temperature with H2 gas. Initially (9 different pressures) mol Calibration constant (k) for H2 mbar s Deviation from the initial value (%) *
78 days later (9 different pressures)
Average of 50 measurements during a 135 day period
5.354 103 ± 5.9 105
5.154 103 ± 4.5 105
5.243 103 ± 3.4 104*
–
3.7
2.1
The error given here is the standard deviation of the data.
Table 3 Calibration constant (k) measured at different temperatures in case of N2 and Ar. Temperature (°C)
Calibration constant (k) of N2 s mol mbar
Calibration constant (k) of Ar s mol mbar
Room temperature 50 60 70 80 90 100 110 120 130 140
1.941 103 ± 5.5 105 1.835 103 ± 3.3 105 1.949 103 ± 4.6 105 2.05 103 ± 3.2 105 2.014 103 ± 9.9 105 2.006 103 ± 2.7 105 1.994 103 ± 1.8 105 2.046 103 ± 2.3 105 2.034 103 ± 2.6 105 2.018 103 ± 2.8 105 1.967 103 ± 1.6 105
1.319 103 ± 3.2 105 1.305 103 ± 1.7 105 1.296 103 ± 1.1 105 1.278 103 ± 2 105 1.315 103 ± 4.4 106 1.357 103 ± 8.1 106 1.326 103 ± 5.2 106 1.334 103 ± 3.3 106 1.328 103 ± 5.2 106 1.331 103 ± 1.3 105 1.331 103 ± 6.8 106
3.4. Sensitivity of the new permeation measurement equipment
Fig. 6. Calibration constant (k) measured at different temperatures in case of N2 and Ar. The deviations in the values are less than the reproducibility of the measurement setup.
surements and not heated by the membrane heating. It can thus be assumed to be always at room temperature during the calibration measurements as well as the inlet of the turbomolecular pump and the vacuum chamber section right above. Because of the geometry of the chamber there is no direct straight path from the flange where the particles enter the chamber and the inlet of the pump meaning that the particles entering have to collide with the walls before reaching the turbomolecular pump. As the walls with which the gas molecules collide before entering the pump are at room temperature, the gas temperature in the vacuum chamber at the turbomolecular pump can be assumed to be room temperature which is independent of the membrane temperature leading to calibration constants independent of membrane temperature.
Knowing the values of the calibration constants, the detection limit for the permeation of different gases can be estimated based on the background pressure of the vacuum chamber and the minimum reliably detectable pressure increase. Another limit is applicable for the determination of the transport parameters of the gases in a given membrane material as the calculation of diffusivity, permeability and solubility requires the measurement of the build-up of the permeation flux with adequate time and pressure resolution (see the time-lag method (e.g. [15,16])). Given that the background pressure of the vacuum chamber is approx. 7 109 mbar a pressure increase of 1 109 mbar can already be dependably detected. For the determination of the transport parameters the build-up of the permeation flux should be adequately measurable as the integration of the measured pressure increase vs. time (Fig. 2/b) curve is required. In order to achieve a reasonably accurate integrated curve, the pressure increase vs. time curve should have a reasonable number of points and a reasonable time and pressure resolution. To achieve this as a conservative assumption a higher, approx. 1 108 mbar pressure increase is required. The detection and measurement limits for the different gases can be calculated by multiplying the corresponding k value with the pressure increases mentioned above. The resulting limits are summarized in Table 4. The upper limit of the measureable permeating fluxes can also be estimated as the pumping speed of the turbomolecular pumps above 103 mbar pressure is not linear anymore, meaning that the calibration and measurements become difficult. Furthermore, the turbomolecular pump should not be operated above a given pressure. It can thus be concluded that in case of
}k et al. / Measurement 59 (2015) 241–247 B. Sebo
247
Table 4 Estimated minimum detectable permeation fluxes, minimum permeation fluxes for the determination of transport parameters and the maximum measurable permeating fluxes for different gases. The above values can be considered as a conservative assumption and they were calculated by multiplying the k values corresponding to the different gases with the pressures differences given in the text. Permeating gas
Min. detectable mol
Min. flux for transport
1.5 1011 5.4 1012 1.9 1012 1.3 1012 1.2 1012
1.5 1010 5.4 1011 1.9 1011 1.3 1011 1.2 1011
perm. flux He H2 N2 Ar CO2
s
steady-state permeation the equipment should not be used above 1 103 mbar pressure in the vacuum chamber, which gives the maximum measurable permeating flux for the different gases by multiplication with the respective k value (Table 4). It is worth noting, that the minimum permeating flux required for the determination of the transport parameters and the upper limit of the practically measurable permeating flux differ by 5 orders of magnitude. Considering the wide range of usable membrane thicknesses [10] in the measurement setup this means that the transport properties of a wide range of materials with very different transport properties can be investigated with the aid of the equipment. 4. Conclusion In this study the calibration and the determination of the sensitivity and reproducibility of a new permeation equipment described earlier [10] are presented. It is shown that the calibration constant describing the relationship between the material current entering the vacuum chamber through a membrane and the pressure increase over the base pressure in the vacuum chamber is dependent on the type of gas used but independent of the membrane temperature. The latter can be explained by the ‘thermalisation’ of the gas molecules through collision with the walls of the vacuum chamber which remain at room temperature even when the membrane is heated. The type of permeating gas influences the calibration constant through the different ionisation probabilities, thus through the different gas correction factors of the hot cathode pressure gauge used as the detector and through the different pumping speeds of the turbomolecular pump. The calibration constants measured over a longer time period are indicating the reproducibility of the data obtained with the permeation measurement equipment. These measurements showed that the deviation in the values of the determined calibration constants is less than 5%. Considering the standard deviation of the measured values during a 135 day long time period and with a 95% confidence interval (2 times the standard deviation) the reproducibility of the device can be determined as approx. ±13%. With the aid of the calibration constant the minimum detectable (sensitivity) and the maximum measurable permeating material current can also be calculated for
param. determination
mol s
Upper limit of the permeating flux mol s 1.5 105 5.4 106 1.9 106 1.3 106 1.2 106
the different gases used in the calibration measurements. The results showed that the minimum detectable permeating material current and the minimum material current required for the determination of the transport parameters is in the range of 1011 bmol s1 c and the maximum measurable permeating current is 5 orders of magnitude higher than this. Along with the large variety of mountable membrane thicknesses and materials this means that the transport properties of a wide range of materials can be investigated. References [1] J.P. Wu, I.W.M. Brown, M.E. Bowden, T. Kemmitt, Palladium coated porous anodic alumina membranes for gas reforming processes, Solid State Sci. 12 (2010) 1912–1916. [2] N. Itoh, N. Tomura, T. Tsuji, M. Hongo, Deposition of palladium inside straight mesopores of anodic alumina tube and its hydrogen permeability, Micropor. Mesopor. Mater. 39 (2000) 103–111. [3] T. Ozaki, Y. Zhang, M. Komaki, C. Nishimura, Hydrogen permeation characteristics of V–Ni–Al alloys, Int. J. Hydrogen Energy 28 (2003) 1229–1235. [4] N.M. Peachey, R.C. Snow, R.C. Dye, Composite Pd/Ta metal membranes for hydrogen separation, J. Membr. Sci. 111 (1996) 123–133. [5] L.C. Witjens, J.H. Bitter, A.J. van Dillen, W.M. Arnoldbik, F.H.P.M. Habraken, K.P. de Jong, Improving the control of the electroless plating synthesis of Pd/Ag membranes for hydrogen separation using Rutherford backscattering, J. Membr. Sci. 254 (2005) 241–248. [6] T.S. Moss, N.M. Peachey, R.C. Snow, R.C. Dye, Multilayer metal membranes for hydrogen separation, Int. J. Hydrogen Energy 23 (1998) 99–106. [7] Y. Zhang, T. Ozaki, M. Komaki, C. Nishimura, Hydrogen permeation characteristics of vanadium–aluminium alloys, Scripta Mater. 47 (2002) 601–606. [8] A.L. Athayde, R.W. Baker, P. Nguyen, Metal composite membranes for hydrogen separation, J. Membr. Sci. 94 (1994) 299–311. [9] P.V. Mercea, D. Silipasß, V. Mecea, Separation of a gas mixture through a polymer membrane metallized with palladium, Gas Sep. Purif. 4 (1990) 137–140. }k, G. Kiss, G. Dobos, F. Réti, T. Majoros, O.H. Krafcsik, Novel [10] B. Sebo instrument for the investigation of small permeation fluxes of gases through different membranes, Measurement 46 (2013) 3516–3524. [11] K. Jousten, Handbook of Vacuum Technology, John Wiley & Sons, Berlin, 2008. [12] M. Matsuyama, H. Miyake, K. Asiida, K. Wantanabe, Permeation, diffusion and dissolution of hydrogen isotopes, methane and inert gases through/in a tetrafluoroethylene film, J. Nucl. Mater. 110 (1982) 296–300. [13] R.A. Pasternak, M.V. Christensen, J. Heller, Diffusion and permeation of oxygen, nitrogen, carbon dioxide and nitrogen dioxide through polytetrafluoroethylene, Macromolecules 3 (1970) 366–371. [14] J. Crank, G.S. Park, Diffusion in Polymer, Academic Press Inc., London, 1968. [15] H.L. Frisch, The time lag in diffusion, J. Phys. Chem. 61 (1957) 93–95. [16] S.W. Rutherford, D.D. Do, Review of time lag permeation technique as a method for characterisation of porous media and membranes, Adsorption 3 (1997) 283–312.