Fusion Engineering and Design 83 (2008) 655–660

Contents lists available at ScienceDirect

Fusion Engineering and Design

journa l homepage: www.e lsev ier .com/ locate / fusengdes

Adsorption capacity of hydrogen isotopes on mordenite

Yoshinori Kawamura ∗, Yoshihiro Onishi1, Kenji Okuno1, Toshihiko Yamanishi

, a morationtopen colue facte dev

acityigatedt ofthermwereucedwere

Tritium Technology Group, Directorate of Fusion Research and Development,Japan Atomic Energy Agency, 2-4 Shirane Shirakata, Tokai, Ibaraki 319-1195, Japan

a r t i c l e i n f o

Article history:Received 18 May 2007Received in revised form 26 December 2007Accepted 8 February 2008Available online 22 April 2008

Keywords:Fusion reactorHydrogen isotopeTritiumSynthesis zeoliteAdsorptionLangmuir model

a b s t r a c t

In a fusion reactor systemof system control and opemethods for hydrogen isomaterial of the separatioof cation and so on. If thclarified, it may lead to ththis work, adsorption capzeolite (NaY) were investadsorption per unit weighof CaA. The adsorption isocoefficients of H2 and D2

were estimated by the redparameters of the zeolite

1. Introduction

In a nuclear fusion reactor system, measurement and analy-sis of hydrogen isotopes including tritium are necessary from theviewpoint of safety of system control and operation. A gas chro-matograph (GC) with a cryogenic separation column is one of themethods for hydrogen isotope analysis. Alumina is a typical packedmaterial of the cryogenic separation column, and it is used at liq-uid nitrogen temperature (77 K). However, use of liquid nitrogen isa cause of long analysis time (tens of minutes). The long analysistime becomes a weak point when a GC with a cryogenic separationcolumn is used as process monitor. The present author has pro-posed to use a micro gas chromatograph (micro GC) for hydrogenisotope analysis and has developed the cryogenic separation col-umn for micro GC, and has reduced the analysis time to less thana few minutes successfully [1–3]. However, liquid nitrogen is stillused to cool the separation column, and it is an obstruction to goodsensitivity and easy installation of micro GC. The development ofthe column material having separation capability at the temper-ature, which can be reached by an electric device, is one of the

∗ Corresponding author. Tel.: +81 29 282 6393; fax: +81 29 282 5917.E-mail address: kawamura.yoshinori@jaea.go.jp (Y. Kawamura).

1 Radiochemistry Research Laboratory, Faculty of Science, Shizuoka Unversity, 836Ohya, Suruga-Ward, Shizuoka 457-8256, Japan.

0920-3796/$ – see front matter © 2008 Elsevier B.V. All rights reserved.doi:10.1016/j.fusengdes.2008.02.004

nitoring of hydrogen isotopes including tritium is necessary for the safety. A gas chromatography using a cryogenic separation column is one of the

analysis. Synthesis zeolite such as molecular sieve 5A (CaA) is a candidatemn, and its property varies by the ratio of silica to alumina, the kinds

or affected the hydrogen adsorption property of the synthesis zeolite iselopment of the new zeolite optimized to the separation column. So, in

of hydrogen (H2) and deuterium (D2) for mordenite (MOR) and NaY typeat various temperatures, and were compared with CaA. The amount of

MOR was larger than that of CaA, and that of NaY was smaller than thats were expressed by sum of two Langmuir equations, and the Langmuir

proposed. Furthermore, the Langmuir coefficients of HD, HT, DT and T2

mass. The correlation between the adsorption properties and the physicalnot confirmed.

© 2008 Elsevier B.V. All rights reserved.

solutions for this weak point. Synthesis zeolite is a probable can-didate of the column material. Its structure varies by the ratio ofsilica (SiO2) to alumina (Al2O3), the kinds of cation and so on, andit gives the unique function to each zeolite. If the factor effected tothe hydrogen adsorption property of the synthesis zeolite is clari-fied, it may lead to the development of the new zeolite optimized to

the hydrogen isotope separation column. Therefore, investigation ofthe hydrogen adsorption properties about various synthesis zeoliteis necessary. Molecular sieve 5A (CaA) is a typical packed mate-rial of separation column, and its hydrogen adsorption capacity atvarious temperatures has been already investigated [4]. Mordenite(MOR) is also a kind of a synthesis zeolite, and it has been reportedthat the separation column with it can separate hydrogen isotopemixture at fairly high temperature [5,6]. However, its specific datahave been not clearly indicated. The investigation of sorption rateand its isotope effect is necessary to evaluate a material that issuitable for GC separation column or not. And, to evaluate the sorp-tion rate and its isotope effect, the investigation of the adsorptioncapacity (adsorption isotherm) of hydrogen isotopes on the mate-rial is needed. This is also necessary to investigate the factor thataffects the isotope effect of hydrogen adsorption on the material.So, in this work, as a first step of the evaluation as a GC separationcolumn and of the investigation of the factor effected to the iso-tope effect on hydrogen adsorption, hydrogen (H2) and deuterium(D2) adsorption capacities of MOR are investigated. And then, thatof Y type zeolite (NaY) is also investigated to compare with CaAand MOR.

eering

n-Pentane (mp) 143Dry ice in ethanol 195

System volume [cm3]Reference volume 104.3Sample container 186.7Piping 35.6

The experimental temperature was changed from 77 K to 195 Kby using various refrigerants. When the sample container wasimmersed into the refrigerant, temperature gradient was on the

656 Y. Kawamura et al. / Fusion Engin

2. Experimental

MOR and NaY used in this work were the high silica zeolite (HSZ-642NAA and HSZ-320NAA) purchased from TOSOH Co. The cationin common with these zeolites is sodium ion (CaA includes not onlysodium ion but also calcium ion). The ratio of silica/alumina of MORand NaY are about 20 and 5, respectively, and are larger than CaA(Si/Al = 2). It is necessary to notice that MOR used in this work doesnot include calcium ion and is not same one in Refs. [5,6]. Samplespecifications are listed in Table 1.

Fig. 1 shows the schematic diagram of the experimental appa-ratus. A volumetric method was applied to the measurement ofadsorption isotherm. The measurements of the system volume (thesample container, piping and the sample) were carried out usinghelium gas, and were based on the reference volume whose volumewas known. The experimental procedures were as follows:

(1) The sample was charged into the sample container, and waskept at 473 K under the vacuum condition for 12 h or more to

eliminate the residual water.

(2) The sample container was immersed into the refrigerant to cooldown the sample to the experimental temperature.

(3) The experimental apparatus was isolated from the vacuum sys-tem by closing the isolation valve, and the system pressure wasrecorded, and then, the sample container was isolated by clos-ing the sample isolation valve.

(4) The sample gas (H2 or D2) was introduced to the referencevolume including the piping, and the system pressure wasrecorded.

(5) The sample isolation valve was opened to start adsorption. Thesystem pressure change was recorded until the equilibriumstate was attained.

(6) The amount of the sample gas residual in gas phase of the exper-imental apparatus was calculated from the system pressure,the system volume and temperature, and then, the amount ofadsorption was obtained by subtracting it from the amount ofthe sample gas introduced at procedure (4).

(7) The sample container was isolated again, and the procedurewas returned to (4). Then, the procedures (4–7) were repeateduntil the system pressure was attained to the certain pressure.

Table 1Sample specification

Brand name HSZ-642NAA HSZ-320NAA MS-5A

Type MOR NaY CaACation Na+ Na+ Na+, Ca2+

Pore size (A) 7 8 5BET surface area (m2/g) 360 700 650SiO2/Al2O3 17.8 5.5 2Na (Ca)/Al2O3 1.05 1 0.38 (0.42)

Fig. 1. A schematic diagram of the experimental apparatus.

and Design 83 (2008) 655–660

Table 2An experimental condition

Sample

MOR NaY

Charged weight [g] 17.02 15.03Adsorbate H2, D2

Temperature [K]N2 (boling point) 77Ar (bp) 87Isopentane (melting point) 113

sample container. So, the net volume of the sample containerwas not able to use for the calculation of adsorbed amount.Therefore, the apparent volume of the sample container wasmeasured at each experimental temperature, and was usedfor the calculation. The experimental conditions were listed inTable 2.

3. Results and discussions

Fig. 2 shows the adsorption capacity of H2 (top) and D2 (bottom)on MOR at various experimental temperatures. The adsorptioncapacity decreased with increase of temperature. The amount of

Fig. 2. Adsorption capacity of H2 (top) and D2 (bottom) on MOR at various temper-atures.

Y. Kawamura et al. / Fusion Engineering

Fig. 3. Adsorption capacity of H2 (top) and D2 (bottom) on NaY at various temper-atures.

D2 adsorption was larger than that of H2 adsorption, and the dif-ference of them became small with increasing temperature. Fig. 3shows the adsorption capacity of H2 (top) and D2 (bottom) on NaYat various temperatures. The tendency of the adsorption capacitywas similar to that of MOR. However, the adsorption capacity perunit weight was smaller than that of MOR even if the BET surfacearea is large. In case of lower temperature, the adsorption capacityat lower pressure was about one order of magnitude smaller thanthat of MOR. On the other hand, at higher pressure, the adsorptioncapacity of NaY was close to that of MOR. Detailed discussion is

carried out later. In these figures, broken lines mean the adsorptioncapacity of H2 and D2 for active alumina at 77 K. This alumina waspurchased from Sigma–Aldrich Co. 19 wt% of MnCl2 was doped tothis alumina for the prevention of the separation of para and ortho.Adsorption capacity is fairly smaller than MOR or NaY. The adsorp-tion capacity is not correlated directly to the isotope separationcapability.

The hydrogen adsorption capacities on MOR and that on NaYwere also not expressed by Langmuir model simply. Nishikawaet al. have proposed to apply sum of two Langumuir equationsas the adsorption isotherms of hydrogen for CaA, NaA(MS4A) andAC(activated carbon) of 77 K [7]. In this work, this model is expe-diently called Two Site Langmuir (TSL) model. In this model, it isassumed that two kinds of activated adsorption sites are on the sur-face of adsorbent, and that the Langmuir model can be applied toeach adsorption site. So, adsorption capacity is expressed as:

Q = a1P

1 + b1P+ a2P

1 + b2P, (1)

where Q is a adsorption capacity [mol/g], P is a pressure of hydrogen[Pa]. “a” [mol/g/Pa] and “b” [1/Pa] is called the Langmuir coefficient.

and Design 83 (2008) 655–660 657

“b” means the equilibrium constant of the adsorption reaction.“a” means the product of the saturated adsorption amount andthe equilibrium constant “b”. Subscript 1 and 2 mean the adsorp-tion site 1 and site 2, respectively. Nishikawa et al. have provedexperimentally that TSL model can be extended to the multi-component system by applying Markham–Benton correlation [8,9].The following equation is the extended form of TSL model for themulti-component system;

Qi = a1,iPi

1 +∑n

i=1b1,iPi

+ a2,iPi

1 +∑n

i=1b2,iPi

, (2)

where n and i mean the number of components in the system and acertain component in the system, respectively. The present authorhas assumed that temperature dependence of adsorption capacityon site 1 and site 2 are different from each other, and has proposedfollowing equation [4];

Q = a0,1 exp(E1/RT)P1 + b0,1 exp(E1/RT)P

+ a0,2 exp(E2/RT)P1 + b0,2 exp(E2/RT)P

, (3)

where E is a apparent adsorption heat [J/mol], R is a gas constant[J/mol/K] and T is an absolute temperature [K]. Subscript 0 meansthat it is a frequency factor. For CaA, the frequency factors andthe apparent adsorption heat have been proposed from the dataobtained between 77 K and 195 K by the present author [4].

In TSL model, it is assumed that the adsorption caused by site1 is dominant at lower pressure and lower temperature, and thatthe adsorption caused by site 2 become dominant with increase ofpressure and/or temperature. The physical validity of TSL model isnot confirmed yet. However, this model can express actually thehydrogen adsorption capacity of CaA, NaA and AC well within thepressure range below an atmospheric pressure. The present authorhas applied the vacancy solution theory to the hydrogen adsorp-tion capacity of CaA. It was also able to express well the adsorptioncapacity. However, the temperature dependence of Henry constantcould not be expressed well [10,11]. Therefore, in this work, TSLmodel was applied to the expression of the hydrogen adsorptioncapacity of MOR and NaY.

When the equilibrium pressure is low, the denominator of theLangmuir equation is close to 1.0. So, Eq. (3) is rewritten in the formof Henry equation as;

Q =(

a0,1 exp(

E1

RT

)+ a0,2 exp

(E2

RT

))P. (4)

From the data at lower equilibrium pressure, Henry constant

was obtained. Fig. 4 shows the temperature dependence of obtainedHenry constant of MOR. As shown in this figure, temperaturedependence of Henry constant seems to include different two tem-perature dependences. The Langmuir constants (a0,1 and a0,2) andthe apparent adsorption heats (E1 and E2) were estimated from thisfigure. On the other hand, the saturated adsorption amount at site2 was assumed much larger than that at site 1. So, it was consid-ered that the maximum value of the adsorption capacity at 77 Kwas close to the saturated amount of adsorption caused by site 2.As mentioned before, the Langmuir coefficient “a” is the product ofthe saturated adsorption amount and the equilibrium constant “b”.So, the Langmuir coefficients, b0,1 and b0,2, have been estimated bytrial and error using the maximum adsorption capacity (at 77 K)as the initial value of the saturated adsorption amount of site 2.Obtained Langmuir coefficients and the apparent adsorption heatswere listed in Table 3, and that of CaA were also listed in this tableas a reference. The adsorption isotherms estimated were shown inFigs. 2 and 3 as the solid lines.

To confirm the validity of the obtained Langmuir coefficients,the adsorption capacities of H2 and D2 on MOR and NaY were mea-sured at 175 K. Methanol was used as a refrigerant. Fig. 5 shows

658 Y. Kawamura et al. / Fusion Engineering and Design 83 (2008) 655–660

Fig. 4. A temperature dependence of Henry constant of H2 adsorption on MOR.

a comparison with the estimation and observation of adsorptionisotherms at 175 K. The top is the case of MOR and the bottom is thecase of NaY. In case of MOR, the observation was slightly smallerthan the estimation. On the other hand, the observation agreedwith the estimation in the case of NaY. From these results, adsorp-tion isotherms obtained in this work are considered to be fairlyvalid. Therefore, after this, the discussions are done with obtained

adsorption isotherms.

Fig. 6 shows a ratio of H2 adsorption capacity of MOR or NaYto that of CaA. The H2 adsorption amount on MOR was larger thanthat on CaA under the experimental condition in this work. Espe-cially, at low temperature, the ratio becomes large with decreaseof the pressure. Oppositely, the ratio becomes closer to 1 at higherpressure. The H2 adsorption amount on NaY was smaller than thaton CaA under this experimental condition. At low temperature, theratio becomes smaller than 1.0 with decrease of the pressure. Thesimilar tendency has been observed for D2 also. The pressure andtemperature regions of which the change of the adsorption capacityratio is large agree with the region of which the adsorption to thesite 1 is dominant. As shown in Table 1, the order of the adsorptioncapacity was not corresponding to that of the BET surface area, ofthe pore size and of the silica/alumina ratio. It also seemed to notbe corresponding to the kind of cation. The BET surface area is theapproximation and is not the true surface area. However, the dis-agreement of this order suggests that hydrogen molecule adsorbson the specific place of the surface and does not cover the surfaceuniformly. That is, an active adsorption site is on the surface of zeo-lite, and it is not distributed uniformly. Therefore, it is considered

Table 3The observed Langmuir coefficients of H2 and D2

Adsorbate H2

Adsorbent MOR NaY

Site 1a0,1 (×10−13 mol/g/Pa) 1.4 1.0b0,1 (×10−10 1/Pa) 1.1 1.3E1 (J/mol) 12,600 9350

Site 2a0,2 (×10−10 mol/g/Pa) 1.6 0.69b0,2 (×10−8 1/Pa) 4.1 1.5E2 (J/mol) 4,900 4,950

Fig. 5. Comparison with the observed adsorption capacity and the estimation at175 K (top, MOR; bottom, NaY).

that adsorption capacity does not depend on only one parametersuch as BET surface area.

Fig. 7 shows isotope effect of adsorption capacity of each zeoliteat various temperatures. A gray solid line means the isotope effectof adsorption capacity on alumina (mentioned at Figs. 2 and 3) at77 K. Isotope effect of adsorption capacity on alumina is so small incomparison with other zeolites. This fact also shows that adsorp-tion capacity is not correlated directly to the separation capabilityof GC column. The isotope effect of adsorption capacity on zeolitesbecomes small with increase of temperature and/or pressure. Thatis, in the case of zeolite, the isotope effect of the adsorption capacitycorresponding to the site 1 is large, and that to the site 2 is small. Nocorrelation between the isotope effect and the parameters listed inTable 1 were observed. On the other hand, the present author hasinvestigated the hydrogen adsorption capacity of various activatedcarbons. Activated carbon (AC) consists of only carbon, and is incontrast to zeolite. Furthermore, AC is added to a function of the

D2

CaA MOR NaY CaA

1.5 2.8 3.8 51.1 1.7 3.5 3.5

11,220 12,450 9200 10,970

1.4 1.4 0.6 1.34.5 4 1.4 4.1

4,990 5,100 5,120 5,150

Y. Kawamura et al. / Fusion Engineering

Fig. 6. A comparison with the H2 adsorption capacity ratio MOR/CaA and NaY/CaA.

(1) The adsorption to the site 1 becomes dominant at lower tem-perature and at lower pressure (the small amount of adsorbate).The apparent adsorption heat depends on the kinds of adsor-bent as shown in Table 3. Therefore, the adsorption to the site 1means the fairly strong interaction between the adsorbate and

Fig. 7. A comparison of the isotope effect of adsorption capacity among MOR, NaYand CaA.

molecular sieving by changing the start material, and it is calledthe molecular sieving carbon. Fig. 8 shows the adsorption capacityof H2 and D2 of various molecular sieving carbons at 77 K. LGK-827,LGK-828 and LGK-829 are the molecular sieving carbon offered byTakeda Pharmaceutical Company Limited. Their pore sizes are 3 A,4 A and 5 A, and are corresponding to KA (MS3A), NaA and CaA,respectively. The adsorption isotherms (solid line and broken line)

Fig. 8. Adsorption capacity of H2 and D2 on various activated carbon at 77 K.

and Design 83 (2008) 655–660 659

in Fig. 8 are corresponding to AC reported by the present author[12], and are also expressed by TSL model. As shown in this figure,isotope effect of adsorption capacity is small even if it is corre-sponding to the adsorption caused by the site 1, and it seems tonot depend on the pore size. The adsorption capacity also does notdepend on the pore size. This cause is probably that the carbon isthe only constituent atom of the activated carbon. That is, an activeadsorption site is uniformly distributed on AC surface regardless ofits pore size. And, it is considered that the isotope effect of interac-tion between hydrogen isotope and active adsorption site is not solarge. Oppositely, it is considered that the adsorption capacity andits conspicuous isotope effect in case of zeolite are affected over-all to the constituent atoms and the crystal structure. That is, theactive adsorption site is partially distributed around micro pore.The constituent atoms and the crystal structure affect the poresize and the interaction of the hydrogen isotope and the activeadsorption site. However, the interaction is not correlated directlyto the pore size. Then, the present author expediently consideredas follows;

the adsorbent surface (activate adsorption site).(2) The adsorption to the site 2 becomes dominant with increase of

pressure (the amount of adsorbate) or temperature. The appar-ent adsorption heat does not depend on the kinds of adsorbentas shown in Table 3. Therefore, the adsorption to the site 2means the interaction between the adsorbate and the inactivesite and/or the interaction between each adsorbate. Occupiedarea of a H2 molecule has been reported to 0.121 nm2 [13]. Cov-erage estimated from obtained data is much smaller than 1.0.So, the adsorption to the site 2 corresponds to the interactionbetween the adsorbate and the inactive site.

The other reason of the application of TSL model to this work isthe easy estimation of Langmuir coefficient of other species (HD,HT, DT and T2) based on the data of H2 and D2. The one of the pur-poses of this study is the development of the separation column

Fig. 9. A correlation between the Langmuir coefficients and the reduced mass onMOR at 77 K.

660 Y. Kawamura et al. / Fusion Engineering

Table 4

on molecular sieves 5A at low temperature, J. Nucl. Sci. Technol. 37 (2000)

The estimated Langmuir coefficients of HD, HT, DT and T2

Adsorbate

HD HT DT T2

AdsorbentMOR

Site 1a0,1 (×10−13 mol/g Pa) 1.9 2.1 3.4 4.2b0,1 (×10−10 1/Pa) 1.3 1.4 1.9 2.2E1 (J/mol) 12,540 12,510 12,410 12,360

Site 2a0,2 (×10−10 mol/g Pa) 1.5 1.5 1.4 1.3b0,2 (×10−8 1/Pa) 4.1 4.0 4.0 3.9E2 (J/mol) 4,980 5,020 5,150 5,220

NaYSite 1

a0,1 (×10−13 mol/g Pa) 1.7 2.2 5.4 8.3b0,1 (×10−10 1/Pa) 2.0 2.3 4.5 6.2E1 (J/mol) 9,290 9,260 9,160 9,110

Site 2a0,2 (×10−10 mol/g Pa) 0.65 0.64 0.58 0.55b0,2 (×10−8 1/Pa) 1.5 1.4 1.4 1.3E2 (J/mol) 5,020 5,050 5,170 5,220

material which can separate 6 hydrogen isotopes. So, the Langmuircoefficients of HD, HT, DT and T2 are necessary. However, the mea-surement of them by the volumetric method is difficult. Nishikawaet al. have reported that there is a correlation between the Langmuircoefficients and the reduced mass of hydrogen isotope, and haveestimated the coefficients of DT and T2 from the coefficients of H2,HD, HT and D2 [14]. The Langmuir coefficients of T2 on CaA (MS5A),NaA (MS4A) and AC at 77 K have been obtained experimentally bythe present author, and it has been confirmed that the estimatedcoefficients have good accuracy [12]. The Langmuir coefficients of

other species of MOR and NaY were also estimated from the dataof H2 and D2. Fig. 9 shows the relation between the reduced massand the Langmuir coefficients of MOR at 77 K. From this figure, theLangmuir coefficients of HD, HT, DT and T2 on MOR at 77 K are esti-mated. In other temperatures, the Langmuir coefficients were alsoestimated with same procedure. And then, the apparent adsorptionheats were also estimated based on the obtained Langmuir coeffi-cients. The obtained Langmuir coefficients were listed in Table 4.To enhance the accuracy of these coefficients, the observation ofadsorption isotherm of T2 are necessary.

4. Conclusion

To investigate the factor influenced to isotope effect of hydrogenadsorption on synthesis zeolite, the adsorption characteristics ofhydrogen isotopes on mordenite (MOR) and Y type zeolite (NaY)were investigated. This is the first step for evaluation of a materialthat is suitable for GC separation column or not.

The adsorption capacity of H2 and D2 on MOR and that on NaYwere quantified at the temperature range between 77 K and 195 K.The adsorption isotherms were expressed by sum of two kinds of

[

[[

[

[

and Design 83 (2008) 655–660

Langmuir equation, and the Langmuir coefficients and the appar-ent adsorption heats for all 6 hydrogen isotopes were proposed.The adsorption capacity of MOR is larger than that of CaA (MS5A),and that of NaY is smaller than that of CaA. The isotope effect ofadsorption capacity on CaA is the largest among these zeolites.The adsorption capacity and the isotope effect clearly depend onthe kinds of zeolite, but the factors which affect them is still notclear.

To understand the factors that influence the hydrogen adsorp-tion performance of the zeolite, the approach from the viewpointof the computational chemistry may also be necessary.

Acknowledgements

The present authors wish to acknowledge Dr. H. Takatsu andDr. T. Nishitani for their strong supports and encouragements tothis study. The present authors are grateful to Dr. H. Watanabe andTakeda pharmaceutical company Ltd. about the free offer of themolecular sieving carbon samples. The present authors appreciateall staff and contractors of TPL for their devoted efforts to operatefacility safely.

References

[1] Y. Kawamura, Y. Iwai, T. Yamanishi, S. Konishi, M. Nishi, Analysis of hydro-gen isotopes with a micro gas chromatograph, Fusion Eng. Des. 49–50 (2000)855–861.

[2] Y. Kawamura, S. Konishi, M. Nishi, Development of a micro gas chromatographfor the analysis of hydrogen isotope mixtures in the fusion fuel cycle, FusionEng. Des. 58–59 (2001) 389–394.

[3] R. Lasser, S. Grunhagen, Y. Kawamura, Use of micro gas chromatography in thefuel cycle of fusion reactor, Fusion Eng. Des. 69 (2003) 813–817.

[4] Y. Kawamura, S. Konishi, M. Nishi, Adsorption isotherms of hydrogen isotopes

536–542.[5] J. Wendel, H. Wertenbach, M. Glugla, R.-D. Penzhorn, B. Spelta, I. Ricapito, et al.,

Quantitative analysis of hydrogen isotopes with a mordenite column at 173 Kby radio gas chromatography, Fusion Technol. 28 (1995) 1090–1094.

[6] U. Engelmann, R. Furrer, R. Van Den Bergh, G. Vassallo, Modified mordenitecolumns for hydrogen isotopes gas chromatography, in: Proc. of 4th Interna-tional Symposium on Fusion Nuclear Technology (ISFNT-4), FT-P11, Apr. 6–11,1997, Meiji Kinenkan, Tokyo, Japan, 1997, p. 240.

[7] M. Nishikawa, K. Tanaka, M. Uetake, Study on a method to recover tritium fromblanket sweep gas, Fusion Technol. 26 (1994) 17–26.

[8] M. Nishikawa, K. Tanaka, M. Uetake, M. Enoeda, Y. Kawamura, K. Okuno, Adsorp-tion isotherm and separation factor for multicomponent hydrogen isotopes incryosorption method for recovery of tritium from blanket sweep gas, FusionTechnol. 28 (1995) 711–716.

[9] E.C. Markham, A.F. Benton, The adsorption of gas mixtures by silica, J. Am. Chem.Soc. 53 (1931) 497–507.

10] S. Suwanayuen, R.P. Danner, A gas adsorption isotherm equation based onvacancy solution theory, AIChE J. 26 (1980) 68–76.

11] Y. Kawamura, private communications.12] Y. Kawamura, M. Enoeda, R.S. Willms, P.M. Zielinski, R.H. Wilhelm, M. Nishi,

Adsorption isotherms for tritium on various adsorbents at liquid nitrogen tem-perature, Fusion Technol. 37 (2000) 54–61.

13] A.L. McClellan, H.F. Harnsberger, Cross-sectional areas of molecules adsorbedon solid surfaces, J. Colloid Interface Sci. 23 (1967) 577–599.

14] M. Nishikawa, K. Tanaka, M. Uetake, T. Shiraishi, Isotope effect in cryosorptionof tritium to molecular sieves and activated carbon at 77 K, Fusion Technol. 34(1998) 234–240.