Hydrogen Permeability of Palladium Alloy Membrane at High Temperatures in the Impurity Gases Co-existing Atmospheres

Transcription

1 WEC 6 / 3-6 une 006 Lyn France ydrgen Permeability f Palladium Ally Membrane at igh Temperatures in the Impurity Gases C-existing Atmspheres A. Unemt a,*, A. Kaimai a, T. Otake a, K. Yashir a, T. Kawada a,. Mizusaki a, T. Tsuneki b, I. Yasuda b a Institute f Multidisciplinary Research fr Advanced Materials, Thku Univ. -- Katahira, Aba-ku, Sendai, , apan Tel: Fax: * unemt@mail.tagen.thku.ac.jp b Tky Gas C., Ltd., ydrgen Business Prject, R&D Divisin -6-5, Shibaura, Minat-ku, Tky, , apan ABSTRACT: ydrgen permeatin measurements were carried ut with Ag 3wt% Pd ally membrane f 0µm in thickness s as t clarify the effect f c-existing gases n hydrgen permeability thrugh the steam refrming reactin prcess. C-intrductin f impurity gases such as O, CO, CO, C 4 with hydrgen did nt affect hydrgen permeability f palladium ally membrane at high temperatures (773K, 873K). The rate determining step was the bulk diffusin f in the membrane. wever, effects f c-existing gases were visible at lw temperatures in - O, -CO, -CO, -C 4 atmspheres. Rate equatin f surface reactin was btained frm the result f hydrgen permeatin measurements. The results f perfrmance simulatin by the use f surface reactin rate equatin suggested that the effect f impurity gases n hydrgen permeability was negligibly small at high temperatures. KEYWORDS: Palladium Ally Membrane, Membrane Refrmer, Impurity Gas, Surface Reactin, igh Temperatures. Intrductin Palladium series allys have attracted attentins fr lng decades as hydrgen separatin membrane due t their high hydrgen permeability and selectivity [-3]. In 004, Tky Gas C., Ltd. had develped a prttype system f Membrane Refrmer which utilizes a palladium ally t prduce hydrgen frm natural gas thrugh the steam refrming reactin. It achieved a hydrgen prductin capacity f 40Nm 3 /h and purity f 5N []. Steam refrming reactin, using methane as the main cmpnent f natural gas is as fllws, C + 4 O CO 3 + () At the same time, shift reactin ccurs, CO + O CO + () These reactins shift t the right side when hydrgen is remved frm the system by the palladium allys. Thus, Membrane Refrmer is cnsidered as a highly-efficient system which refrms fuel gas and purifies hydrgen in nly ne step. Step 3: Cmbinatin Step : Adsrptin/Dissciatin Aux Gas Step : Bulk Diffusin Upstream Membrane Dwnstream Fig. Schematic illustratin f hydrgen permeatin prcess thrugh the palladium ally membrane. /9

2 WEC 6 / 3-6 une 006 Lyn France ydrgen permeatin pathway cnsists f three steps as shwn in Fig.. Diffusin flux f,, in the membrane is expressed by Fick s st law as fllws, C = D (3) x where, D dentes the diffusin cefficient f in slids, C is hydrgen cncentratin. The relatin between equilibrium hydrgen cncentratin, C,eq., and hydrgen partial pressure, p, is expressed by Sieverts law as fllws, C, eq. K p = (4) where, K dentes Sieverts cnstant. We define the hydrgen activity in slids, a, as, p a = (5) p where, a is unity if the slid is in equilibrium with 0 5 Pa hydrgen. Frm the lcal equilibrium assumptin, eq. (3) is rewritten as, ( p ) a = D K x (3 ) Integratin f eq. (3 ) thrugh thickness f the membrane, l, leads, = D K ( p ) a,up a l,dwn (6) where, a,up and a,dwn are hydrgen activities in the slid just beneath the surfaces in the upstream and dwnstream chambers, respectively. If the surface f the membrane is in equilibrium with the gas phase bth in the upstream and the dwnstream chambers, eq. (6) is mdified as fllws, p,up p,dwn = D K (6 ) l where, p,up and p,dwn dente hydrgen partial pressures in the upstream chamber and in the dwnstream chamber, respectively. Estimated hydrgen flux is decreased if the surface reactin rate is cmparable t, r lwer than the diffusin flux. Figure schematically shws the hydrgen activity distributins when the driving frce fr surface reactin is (a) small and (b) large. (a) (b) Driving frce fr surface reactin is small Driving frce fr surface reactin is large Upstream a x Membrane a x Dwnstream Upstream a x Membrane a a x a Dwnstream Fig. Schematic illustratin f hydrgen activity distributins. The driving frce fr surface reactin prcess is (a) small and (b) large. In pure hydrgen system, the surface reactin rate n Pd based allys knwn t be fast enugh s that hydrgen permeatin flux is well described by the diffusin cntrlled kinetics. In a prttype mdule f Membrane Refrmer, hwever, the btained hydrgen flux was nearly a half cmpared with the estimated ne frm eq. (5) []. /9

3 WEC 6 / 3-6 une 006 Lyn France A pssible cause f the reductin f the btained hydrgen flux is an interference f the impurity gases which cme frm the prcess f stream refrming reactin such as O, CO, CO and C 4. S far, several wrks have been perfrmed t investigate the effect f impurity gases with hydrgen n hydrgen permeability as summarized in table. wever, their details have nt been clarified yet. Thus, this study aims t clarify hw the impurity gases in a steam refrming reactr affect hydrgen permeability f Ag 3wt% Pd membrane by a precise analysis f hydrgen permeatin flux.. Experimental Cmmercially available Ag 3wt% Pd ally membrane (Ishifuku Metal Industry C., Ltd.) f 0µm in thickness was emplyed fr hydrgen permeatin measurements. It was cut int a disk which has 0.5cm effective areas and set int a tw-rm type apparatus as shwn in fig.3, supprted by stainless gaskets. Ttal pressure f bth the upstream chamber and the dwnstream chamber was atm. ydrgen including impurity gas was intrduced int the upstream chamber by 00cc/min. Argn was intrduced int the dwnstream chamber by 0~00cc/min. ydrgen partial pressures in the upstream chamber, p,up, and the dwnstream chamber, p,dwn, were determined as fllws, Thermal Cuple Upstream Chamber Specimen: Ag 3wt% Pd Dwnstream Chamber Q,up,up +Impurity Gas: 00cc/min Q up,ttal Q dwn,ttal Q inlet,dwn : 0~00ccm Gas utlet (T G. C.) Furnace Gas utlet (T F. F. M.) Fig. 3 Schematic illustratin f experimental apparatus: Flw rates f intrduced mixed gas int the upstream chamber and the dwnstream chamber were 00cc/min and 0~00cc/min, respectively. Effective area fr hydrgen permeatin was 0.5cm. Q,up p, up/pa = 0300 (7) Q ttal,up p /Pa = 0300 (8) Q, dwn ttal,dwn Q ttal,up and Q ttal,dwn dente ttal fluxes frm utlet in the upstream chamber and the dwnstream chamber, respectively. Q,up is the intrduced hydrgen flux int the upstream chamber. Permeated hydrgen flux,, was estimated by, = Q Q (9) ttal,dwn inlet,dwn where, Q inlet,dwn dentes intrduced Argn int the dwnstream chamber. All these fluxes were determined by using a Film Flw Meter (ORIBA STEC / (VP-3, VP-3U)). Gaseus cmpnents f utlet gas frm upstream chamber were analyzed by a gas chrmatgraph (Agilent / 3000A Micr G. C.). 3/9

4 WEC 6 / 3-6 une 006 Lyn France Table Summary f reprted results t investigate the effect f impurity gases with hydrgen n hydrgen permeability f palladium allys. Specimen Thickness / µm Kinds f Impurity Examined Gas with ydrgen Temperature / K r Nt Reference Ag 3wt% Pd 0 C 4 +CO +CO+ O 873 [] Pure Pd (Unknwn) C 4, C 4, CO 573<T<673 Nt <573, >673 [3] (Ag-Au-Ru) 5wt% Pd (Unknwn) C 4, CO, CO 643~647 Nt [4] Pure Pd 700 CO <473 >473 Nt CO <573 >573 Nt [5] Pure Pd Pd 76 Ag 4 Pd-(Y, Gd)-Ag 700 C Pd-Y-(In, Sn, Pb) 3, C 4 53<T<73 [6]* Pd 83.3 Ag 5 In.5 Y 0. Pd Ag 0 Au.53 Ru.03 Y.0 Pt 0.5 Al 0.6 Pure Pd Pd 76 Ag 4 Pd-(Y, Gd)-Ag 700 CO 53<T<73 Pd-Y-(In, Sn, Pb) [7]* Pd 83.3 Ag 5 In.5 Y 0. Pd Ag 0 Au.53 Ru.03 Y.0 Pt 0.5 Al 0.6 Pd/Stainless membrane Wall thickness:.6mm Pre size: 0.µm O, CO 653 (± 5) [8] Pure Pd 50 C 4, C 3 8 C 3 6 (+ O) O C 4 + O C O 63<T< <T<873 63<T<873 63<T<873 63<T<873 Nt Pre-xidized Pd Arund 00 CO 43<T<473 [0] *Suffix in referenced data [6] and [7] stand fr atmic %. [9] 4/9

5 WEC 6 / 3-6 une 006 Lyn France 3. Results & Discussin 3. The Effect f Impurity Gases n ydrgen Permeability at igh Temperatures The examined cnditins fr hydrgen permeatin measurements were shwn in table. Frm the results in - O, -CO - O, -CO- O, -C 4 - O systems, we fund that hydrgen flux was nt reduced cmpared t in pure r -Ar systems even the impurity gases were c-intrduced. was prprtinal t the difference f square rts f hydrgen partial pressures between in the upstream and the dwnstream chambers (p,up / -p,dwn / ) by fllwing eq. (6 ). Therefre, the rate determining step is the bulk diffusin f in membrane in abvementined cnditins cntrary t the result by Tky Gas C., Ltd []. The cause f the reductin by half in hydrgen flux in Membrane Refrmer is nt attributed t the effect f impurity gases. Further details f the results will be discussed elsewhere Table Examined cnditins fr hydrgen permeatin measurements at high temperatures (873K, 773K). Adequate amunt f water vapr was c-intrduced when carbn cntaining gases are c-intrduced with hydrgen t avid carbn depsitin. Atmsphere - O -CO- O -CO - O -C 4 - O Partial Pressure f Impurity Gas O up t 9.6kPa CO up t 4.kPa with 9.6kPa O CO up t 8.3kPa with 9.6kPa O (at 873K) CO up t8.5kpa with 9.6kPa O (at 773K) C 4 up t 6.kPa with 9.6kPa O 3. The Effect f Impurity Gases n ydrgen Permeability at Temperatures The results f hydrgen permeatin measurements in - O system at 63K and 445K are shwn in Fig. 4. ydrgen permeatin flux was decreased at 445K and the degree f reductin depended n the O cncentratin. T btain the driving frce fr the surface reactin, we calculated the activity f hydrgen beneath the surface f membrane, a,up, by the mdificatin f eq. (6). Since the surface reactin rate is fast enugh in pure, the dwnstream surface f the membrane is assumed t be equilibrium with the gas phase. Then, a,dwn in eq. (6) can be replaced by the hydrgen partial pressure in the gas phase as, = D K ( p ) a,up p p l,dwn (6 ) a,up can be btained as, l a,up = + p,dwn (6 ) p D K Figure 5 shws permeated hydrgen flux vs. ptential drp at the upstream surface. As a first apprximatin, the surface reactin rate is described as a linear functin f a,up as, p,up = k a,up (0) p 5/9

6 WEC 6 / 3-6 une 006 Lyn France Fig. 4 The results f hydrgen permeatin measurements in - O system at 63K and 445K. Fig. 5 Relatins between driving frce fr surface reactin prcess f hydrgen permeatin and hydrgen permeatin flux at varius temperatures in - O system. k is cnstant. Figure 6 shws the dependence f k n partial pressure f water vapr. Althugh the physical meaning is nt clear, the relatin between k and partial pressure f water vapr, p O, can be rughly written as, O k p () Therefre, eq. (0) is mdified as, = k p a,up O,up p p (0 ) Arrhenius type temperature dependence f k in eq. (0 ) was bserved as shwn in Fig. 7. Then, we btained relatins f k with abslute temperature, T, as, k = k Ea R T exp () E a dentes activatin energy. R is gas cnstant. Therefre, eq. (0 ) is mdified as, = k p O a,up p p,up Ea exp R T The same calculatins were applied t -CO, -CO, -C 4 atmspheres. Kinds f impurity gas dependence f k, k and E a are summarized in Table 3. In -CO atmspheres, the values f k f Membrane D is abut 3.6 times higher than that f Membrane C even the membranes f the same cmpsitin is used fr hydrgen permeatin measurements. The authrs culd nt specify the reasn. Pssible reasns are differences f surface mrphlgy, segregatin f silver and s n. (0 ) 6/9

7 WEC 6 / 3-6 une 006 Lyn France Fig. 6 Dependence f cnstant k n partial pressure f water vapr in eq. (0). Fig. 7 Temperature dependence f cnstant k in eq. (0 ). In -C 4 atmspheres, k did nt shw clear dependence n methane partial pressure. It depended n the partial pressure f methane t - t 0 pwer. Therefre, the values f k and E a were nt specified. Gaseus cmpnents were nt changed, carbn depsitin n the specimen was nt bserved in applied atmspheres. Table 3 Summary f used membrane ID, kinds f impurity gas dependences f k, k and E a. Membrane ID Kinds f Impurity Gas with ydrgen Relatin f k with Partial Pressure f Impurity Gas O A O p B CO 0 p CO k / sec - m - E a / (ml ) C CO D p CO B ~0 C E 4 p C Perfrmance Simulatin by Using Obtained Surface Reactin Rate Equatins The authrs tried t simulate the perfrmance f Membrane Refrmer, regarding hw hydrgen flux is reduced, when O, CO r CO is c-intrduced as an impurity gas with hydrgen int the upstream chamber. ere, reductin rati f btained hydrgen flux, m, is defined as, m = (3) where, dentes the btained hydrgen flux in pure system. The value f a,up is btained by slving the eqs. (6) and (0). Then, by substitutin f a,up int eq. (3), the value f m is given as, 7/9

8 WEC 6 / 3-6 une 006 Lyn France m = D D K K ( p ) ( p ) k l (4) The values f m, calculated by the use f eq. (4) are summarized in table 4. We btained lw value f m even the partial pressure f impurity gas is high. This result shws agreement with the result f hydrgen permeatin measurement at 873K as written in 3.. The reductin by half in btained hydrgen flux in Membrane Refrmer [] culd nt be explained nly by the effect f impurity gases n surface reactin prcess. Table 4 The result f perfrmance simulatin by the use f eq. (4); ydrgen permeatin cefficient 0 (=D k) / (ml ) sec - m - Pa -/ : []. l: 0µm. T: 873K. Atmspheres - O -CO -CO Membrane ID A B C D Partial Pressure f m 00 / % Impurity Gas / % Cnclusins In rder t clarify if the impurity gases with hydrgen affect hydrgen permeability f Ag 3wt% Pd membrane f 0µm in thickness. ydrgen permeatin measurements were carried ut. We fund that hydrgen flux was nt reduced at 873K and 773K in - O, -CO - O, -C 4 - O and -C 4 - O atmspheres. On the ther hand, btained hydrgen flux was bviusly reduced belw temperatures f 63K in - O system cmpared t hydrgen flux in -Ar system. Then, we btained practical rate equatin f the surface reactin by analyzing the relatin between ptential drp arund surface in upstream chamber and hydrgen flux. By using it, we estimated reductin rati f hydrgen flux at perating temperatures f Membrane Refrmer (873K). The results suggest that reductin rati f hydrgen flux are negligibly small even impurity gases such as O, CO and CO are c-existed with hydrgen at high temperatures. 8/9

9 WEC 6 / 3-6 une 006 Lyn France Acknwledgement The authrs sincerely appreciate financial supprt by New Energy and Industrial Technlgy Develpment Organizatin (NEDO) thrugh the Develpment f ighly Efficient ydrgen Prductin Membranes. References: [] I. Yasuda et al., Develpment f Membrane Refrmer fr ighly-efficient ydrgen Prductin frm Natural Gas, 5 th Wrld ydrgen Energy Cnference, 7 une- uly, 004, Ykhama. [] Y. Shirasaki et al., Perfrmance Simulatin f ydrgen Separatin Mdules fr Membrane Refrmer, 004 Fuel Cell Seminar, -5, Nvember, 004, San Antni. [3] R. B. McBride and D. L. McKinley, A new hydrgen recvery rute, Chem. Eng. Prg., 6 (965), 8-85 [4]. Yshida, S. Knishi and Y. Naruse, Effects f Impurities n ydrgen Permeability thrugh Palladium Ally Membranes at Cmparatively igh Pressures and Temperatures,. Less-Cmmn Met., 89 (983) [5] M. Aman, C. Nishimura and K. Kmaki, Effects f igh Cncentratin CO and CO n ydrgen Permeatin thrugh the Palladium Membrane, Mater. Trans., IM, 3 (990), [6] F. L. Chen et al., ydrgen Permeatin thrugh Palladium-based Ally Membranes in Mixtures f 0% Methane and Ethylene in the ydrgen, Int.. ydrgen Energ., (996), [7] Y. Sakamt et al., Effect f Carbn Mnxide n ydrgen Permeatin in Sme Palladium-based Ally Membranes, Int.. ydrgen Energ., (996), [8] A. Li, W. Liang and R. ughes, The Effect f Carbn Mnxide and Steam n the ydrgen Permeability f a Pd/Stainless Steel Membrane,. Membrane Sci., 65 (000), 35-4 [9] S.. ung et al., Effects f C-existing ydrcarbns n ydrgen Permeatin thrugh a Palladium Membrane,. Membrane Sci., 70 (000), [0] D. Wang, T. B. Flanagan and K. L. Shanahan, Permeatin f ydrgen thrugh Pre-xidized Pd Membranes in the Presence and Absence f CO,. Ally Cmpd., 37 (004) /9