Ortho-para catalysis in liquid-hydrogen production – Nvlpubs

by DH Weitzel · 1958 · Cited by 37 — use .o! h?terogeneous catalysis to establish ortho-para. eqUIhbnum. They used catalyze the ortho-parahydrogen conversion was undertaken. 2. Catalyst

275 KB – 7 Pages

PAGE – 1 ============
Journal of Research of the National Bureau of Standards Vol. 60, No.3, March, 1958 Research P aper 2840 Ortho-Para Catalysis In Liquid-Hydrogen Production I D. H. Weitzel, W. V. Loebenstein, J. W. Draper, and O. E. Park A series of selected or specially prepared catalysts were s tudied for their ability to accelerate the ortho to para conversion o f hydrogen. The results of this study a re presented and the performance of various catalysts are compared with that of chromic oxide on alu: mina pellets. An outstanding cata.lyst, hydrous Jerric oxide g ranules, was selected for further study a nd used 111 the liquefiers of the NatIOnal Bureau of Standards Cryogenic Engineering Laboratory. One and one-half liters of this catalyst has now been used to convert more t han 100,000 liters of liquid hydrogen to 90 to 95 percent para at an average rate of about 235 liters of liquid per hour. There is to date no evidence of decrease in efficiency with continued u se. 1. Introduction Freshly liquefied h ydrogen that has not been consists of a 3-to-1 ortho-para mixture. There IS a slow but definite c hange in the mixture on stawhich complicates the problem of keeping the hqUId for any great length of time. The e xothermic of conversion of ortho-to parahydrogen at 20° K IS about 254 calories per mole, whereas the thermic heat of vaporization of liquid hydrogen i s 216 calories per mole [1].2 As a result of this slow change, a thermally isolated tank of liquid hydrogen prepared without conversion to the para form will lose about 1 percent of its volume during the flrst day of storage. In the absence of this internal evlution of heat, the heat transfer to a well-insulated Dewar may result in a loss of less than 1 p ercent pel’ day. The obvious solution of the above difficulty is the conversion to the para form either in the ga’s phase before liquefaction or in the liquid phase immeately after liquefaction, butin any case before delivery to the storage Dewar. 1.1. Background Ifarkas and Saehsse [2] observed the catalytic effect of the paramagnetic molecule of oxygen in the gas phase on the conversion of parahydrogen to normal hydrogen. They also showed that the diamagnetic gases nitrogen, nitrous oxide, carbon dioxide nia, hydriodic acid, sulfur dioxide, and carbonyl caused no conversion, whereas the effects of the paramagnetic gases, nitric oxide and nitrogen d.ioxide, were s imilar to that of oxygen. The CIated molecule J204 was ineffective. Bonhoeffer and Harteck [3] were the first to make use .o! h?terogeneous catalysis to establish ortho-para eqUIhbnum. They used charcoal at liquid-air peratures and were able to establish equilibrium quickly when normal hydrogen was passed over the catalyst. However, when parahydrogen was 1 This work w as SUPI)Orted by Lbe U. . Atomic Energy Commission. , Figures in brackets indicate the Jiteratnre references at the e nd of tbis paper. passed over charcoal at room temperature, 110 version took place [4]. Taylor and collaborators [5, 6] studied the calytic activity of tbe metallic o}”rides and found that the paramagnetic s ubstances chromic o}”ride, cerium oxide, and neodymium oxide brought about rapid conversion, whereas zinc oxide, lanthanum oxide, and vanadium pentoxide, having low or negligible paramagnetism, s howed low or negligible conversion ciencies. They conclude that the magnetic character of the surface of the catalyst is a controlling factor and may aecount for their earlier success with metallic nickel, as well a for the results of Emmett and Harkne s [ 7] with Van del’ ‘\IV aal’s adsorption on iron thetic ammonia catalysts. 1.2 Chromic Oxide Catalyst Grilly [8] made usc of commercially available chromic oxide on alumina for both gas and liquid phase conversion, his object being improvement in the keeping quality of the liquid produced. He successfully modified a liquefier so as to operate at 25 liters pel’ hour for regular production of 85-percent para liquid. Essentially p1ll’e (95 to 97 percent) parahydrogen was first produced in quantity at the National Bureau of Standards Cryogenic Engineering Laboratory in 1953 [9]. The catalyst was placed in the bottom of the liq,nefiel’ reservoir. Liquid hydrogen from the expanSlOn valve filtered through the catalyst before reaching the transfer siphon for delivery to the storage Dewar. Due to the urgency of the proj ect, tho chromic oxide eatftlyst, which had proved effective in Grilly’s work, was used. It consisted of Ys-in. by pellets of approximately 20-percent chromic oxide on alumina. About 65 liters of this cataly t resulted in sion to 95 to 97 percent parahydrogen when the production was about 240 liters of liquid hydrogen per hour. Many thousands of liters of liquid parahydrogen were produced with the aid of this catalyst bed. So long as the catalyst was reactivated by heating and pumping (104° ·c, 25 mm Hg, for 24 hI’ 01′ more) whenever exposure to atmospheric air had 221

PAGE – 2 ============
‘occurred, there appeared to be no reduction in its effectiveness for bringing about the conversion to near equilibrium concentrations, The results obtained with chromic oxide were fying, but a search for a more effective catalyst was continued in the hope of finding one that would cupy less space in the liquid reservoir. The large volume of chromic oxide catalyst required resulted in the formation of considerable quantities of dust or fines that could be carried through the system. To prevent this, filters and baffle plates had to be installed to prevent any s hifting or motion I of the catalyst pellets. This was probably the most serious difficulty overcome in the adaptation of chromic oxide on alumina to the NBS Oryogenic Engineering Laboratory liquefiers. Another disadvantage was the delay in initial cool-down resulting from the heat capacity of the large catalyst mass. Since it was evident that t he design of more pact parahydrogen liquefiers of various capacities would be simplified if a more efficient catalyst were available, a systematic study of materials that might catalyze the ortho-parahydrogen conversion was undertaken. 2. Catalyst Survey The choice of materials to be tested was limited by such practical considerations as resistance to sion, availability, and economic feasibility. Beyond ‘these considerations the selection of samples was based on the reports of previous investigators and on experience witb the chromic oxide on alumina. lor and Diamond [10] found definite cOl’l’elation between catalytic effectiveness and both magnetic susceptibility and adsorptive capacity. Of interest also was the theory advanced by Wigner [11], which states that the velocity constant of the reaction is prop9,rtional to the square of the effective magnetic moment in Bohr magnetons. The effective moment f-Lell is given by the relation f-Leu=2·/S(S+ 1), where S represents total spin of the ion. If n is the number of unpaired electrons in the ion, then S=n/2. Therefore, f-Le!l=..jn(n+2). For chromic ion (Cr+3) having three unpaired electrons (n=3), f-Lett=3.87 magnetons, and for ferric ion (Fe+3) having five unpaired electrons (n=5), f-Le!f=5.92 magnetons. Therefore, the ratio of the velocity constants cording to Wigner’s theory, should be (5.92)2 (3.87)2=2.34. Accordingly, the ferric ion should be two and third times as effective for ortho-parahydrogen conversion as chromic ion, other things being equal. 2.1. Catalyst Preparations With the above considerations in mind, 15 catalysts were obtained either commercially or prepared in the laboratory. A brief description of each material follows: Paramagnetic ferric oxide (15%) on Florex: Fuller’s earth granules obtained from the Floridin Co. were impregnated with a concentrated solution of ferrous ammonium nitrate. This was followed by heating the granules in a vacuum at 2000 C for 48 hours. This technique for the preparation of ported ferric oxide free from ferromagnetism has been described by Selwood [12]. Paramagnetic ferric oxide (2 %) on porous glass: Porous glass granules, obtained from the Oorning Glass Works, were impregnated and oxidized as described above. Magnetite (Fe304) unsupported: Granules of a synthetic magnetite were obtained from the U. S. Bureau of Mines. The granules were reduced in hydrogen at 4500 C, then oxidized in air at 4900 O. Paramagnetic ferric oxide (15%) and chromia (9.3%) on alumina: This mixed catalyst was prepared by impregnating pellets of Cr203 on Davison alumina with netic Fe203′ Ferric ammonium sulfate, unsupported: This material was chosen because of its unusually high magnetic susceptibility at low temperatures. Hydrous ferric oxide, unsupported: The preparation of this material consisted of cipitation of the gel by addition of sodium hydroxide to a solution of ferric chloride, followed by washing the gel with distilled water. The gel was then filtered, extruded, dried, and activated in air at 1400 C. Hydrous manganese oxide, unsupported: The preparation here was the same as that for the hydrous ferric oxide, except that manganese chloride was substituted for the ferric chloride. Manganese dioxide (18 %) on silica gel: This material was prepared by addition of a tion of Mn(NOa)2 to a deactivated sample of Davison “Intermediate Density” silica gel. The granules were then dried and heated in a stream of air at 4000 C. This caused the nitrate to decompose, ing Mn02′ Ferrous chloride on silica gel: Deactivated silica gel was impregnated with a water solution of ferrous chloride. Nickel (0.5%) on alumina: This catalyst was obtained from Baker & 00., Inc. Nickel (5.3%) and thoria (0.24%) on Davison alumina: Since thoria is Imown to be a good promoter of nickel catalysts at high temperatures, this material was made by using a solution of 0.10 mole nickelous nitrate and 0.001 mole thorium nitrate in water. Alumina pellets were wetted with this solution, 222

PAGE – 3 ============
then dried and the nitra,tes decomposed to oxides in a.str.eam of all’ at 4000 C. Finally, the nickel oxide throughout the pellets was reduced to metallic mckel by treatment with hydrogen at400° C. Gadolinium oxide. Neodymium oxide. Crude cerium oxide: These rare-earth oxides were tested because their successful use as ortho-parahydrogen conversion had been reported by Taylor and others. The oXIdes ,yere tested in the form of unsupported powder. Chromia (20%) on alumina: This. was the Harshaw catalyst that had been III the Cryogenic Engineering Laboratory hquefiers. It was used a ,s a comparison in the evaluation of the other catalysts. 2.2. Activities of Catalysts a. Apparatus The appara.tus. constructed for the hydrogen studIes IS shown schematically in fiO’ure 1 and the analyzer i s represented in figure 2. b , . The analyzer was built around a thermal-conductIVIty and is described in a previous paper [13J. A rev:ersmg v.alve was, however, installed in the :flow.lmes leadmg to the cell. This valve makes it pO,ssIble to and analyzed gas from one SIde of th.e bndge to the other, thus obtammg an average readmg from which a number of errors due to imperfect zero balance and filament matching have been eliminated. A four-way selector valve was also added to the analyzer as shown. . The c?ntral part of the apparatus shown in figure 1 IS a stamless-stecl Dewar enclosed in a sealed and vented m etal container, which in turn is s urrounded by a large liquld-nitrogen bath. The c hamber that co:r::tained3 the ca.talyst being was a copper cylu:der, %-111. (hameter by 4 m. long, suspended vertIcally from the top plate of the apparatus by FIGURE 1. Catalyst evaluation apparat1ts. REVERSING VALVE LIQUEFIER SUPPLY DEWAR REFERENCE CONVERSION APPARATUS FIGURE 2. Flow sheet of ortho-para conductivity cell. means of a stainless-steel tube which also served as the tap during catalys’t cleanup. Hydrogen gas entenng the chamber was first led through a heat exchanger, counter-current to the exhaust :flow then through a coil of copper tubing in.) in bath liquid, and finally through the catalyst. The !low the chamber was downward. The top mch 0′[ chamber was packed with copper sponge to a v?lume of good heat exchange for of vapor formed by heat of conversion durmg lIqmd-phase studies. After passing through the catalyst, the hydrogen was warmed to room partly tlu’ough heat exchange with the gas paI:tly by through tlOnal copper tubmg after emergmg from the Dewars. About 50 ml pel’ minute of the exhaust :flow was de:flected through the analyzer system, and the balance tlu’ough a rotameter and wet-gas meter before be1l10′ vented to the atmosphere. The catalyst temperature was assumed to be the same as that of the bath, and this could be varied by choice of bath liquid and by adjustment of the pressure under which the bath was maintained. For studying thc catalyzed reconversion of para to mal hydrogen, the Dewars c ould be removed and a wire-wound sleeve heater controlled by a variable placed around the catalyst chamber. ThIs heater, together with the combined suspension tube and pumping tap mentioned above, served to reactIvate the catalyst prior to a space-velocity measurement. The same cylinder of purified normal hydrogen was used to prOVIde reference gas lor the analysis cell and. gas. to be c onverted in the catalyst chamber. CalIbratIOn of the analysis system at 99.8 percent was c onveniently obtained at the beglllmng of a run simply by using the exhaust gas from the catalyst chamber at two different (low) rates, ?oth. of which gave the same analysis at 25 percent l?arahydrogen rfrom passmg the SImultaneously through both SIdes of the analYSIS cell. 223

PAGE – 4 ============
h. Results To determine the comparative effectiveness of the various catalysts, the space velocity was measured for each in STP volumes per minute per unit volume of catalyst when converting normal 25 percent hydrogen to 90 percent para with the catalyst at 20oK. The results of thIs initial series of ments are shown in table 1, where the catalysts are arranged according to increasing space velocities. Also shown is the relative effectiveness of each lyst when compared with the Harshaw chromic oxide on alumina. TABLE 1. Conversion of 25-percent para to 90-percent para at 20° K, 15 psig Oatalyst Gadolinium oxide, unsupported ________________ _ Crude eerie oxide, unsupported _______________ _ Neodymium oxide. unsupported _______________ _ FeOI, on silica geL _____________________________ _ 2% paramagnetic Fe,O, on porous glass ___ . _____ _ 15% paramagnetic Fe,O, on FloreL ____________ _ Ferric ammonium sulfate, unsupported ________ _ Magnetite, Fe,O., unsupported ________________ _ 20% 0,,0, on alumina (Hm·shaw) ______________ _ 15% paramagnetic Fe,O, and 9.3% OnO, on alumins ______________________________________ _ 5.3% Ni and 0.24% thoria on alumina ___________ _ 18% MnO, on silica geL ________________________ _ 0.5% Ni on alumina ____________________________ _ Hydrous manganese dioxide, unsupported ______ _ Hydrous [erric oxide, unsupported _____________ _ Space velocity STPjmin 10 20 20 20 20 24 30 40 50 50 60 80 100 190 330 Relative space velocity (0″OF1) 0.2 .4 .4 .4 .4 .5 . 6 .8 1.0 1.0 1.2 1.6 2.0 3.8 6.6 The rare-earth oxides did not show up very well, being less than half as active as cln’omic oxide on alumina. (It is difficult to compare these results with the work of Taylor and Diamond [6J because no actual space velocities are reported in their article.) In the present investigation, the laboratory preparations of iron oxide free from ferromagnetism were hardly as good as the reference chromic oxide on alumina. The high magnetic susceptibility of ferric ammonium sulfate at the temperature of the conversion did not result in an outstanding catalyst, but gave an activity between those of paramagnetic iron oxide on Florex and unsupported synthetic magnetite. The nickel and manganese compounds resulted in good catalysts, with % percent nickel on alumina twice as good and manganese dioxide on silica gel 1.6 times as good as the reference chromic oxide on alumina. Although attempts to reduce ferromagnetism of supported iron oxide catalysts did not help the lytic effectiveness, there was a marked improvement in both manganese and iron when the supported oxide was replaced by the unsupported hydrous form. Thus hydrous manganese oxide was about four times and hydrous iron oxide almost seven times as good as chromic oxide on alumina. 3. Hydrous Iron Oxide Catalyst 3.1. Laboratory Studies Additional measurements with the hydrous Iron oxide disclosed at once that the space velocity for liquid-phase conversion could be further increased by a very significant factor simply by using a smaller particle size. The original measurements had been made with 3,pproximately X-in. pellets. Cutting each of these into about four pieces increa.sed the space velocity from 330 to 750, and grinding the material into a fine powder resulted in a STP space velocity of almost 2,000 pel’ minute. Since there are obvious disadvantages to the use of fine powders in liquefiers, it was desirable to prepare more of the material in order to expenment with various mesh sizes and particle modifications. Difficulties were encountered, some of which are referred to by Weiser [14], i. e., properties of the hydrous oxides may show marked variations with small variations in treatment of the sample. though reasonable care was taken to use the same procedure in the preparation of several samples, the space velocities of the resultant products would vary, sometimes by a factor of 2 or 3. It was necessary to prepare a number of batches before sufficient high-activi ty ma,terial was ccllected to support the research a,nd charge one of the large Cryogenic Engineering Laboratory liquefiers . The reasons for the increa,sed activity of the better preparations were not entirely clear. N 0 was found, for exa.mple, between catalytIc activIty and the concentration of various metallic impurities as determined by spectrographic a ,nalysis. creased bulk density was identified with increased catalytic activity, but the procedure for obtaining a high-density product was not apparent. Of course, some increased activity of the unsupported oxide should be anticipated because of the higher age of iron in the sample. This is certainly an simplification and would not explain the range III catalytic activities of the other iron preparations. An excellent correlation was found between lytic activity and surface area as determined from low-temperature nitrogen adsorption by application of the Brunauer, Emmett, and Teller theory [15J. The surface areas of three laboratory preparations of 30 to 100 mesh hydrous iron oxide are shown in figure 3 plotted against their corresponding catalytic ac–E ” w a: ” w u ” 150 ,——r———r 140 J..—-__+_ a: L :J if) 120 L-‘–__ –‘ __ –‘-_-‘–____ -‘-____ -‘-_-‘-‘ 10 20 30 40 50 CATALYTIC ACTIVITY, RELATIVE UNITS FIGURE 3. Correlation of catalytic activity with surface area. 224

PAGE – 5 ============
tivities relative to the reference chromia eatalyst. The crosses passing through eaeh poi nt correspond to the confidence limits of the measurements. It is apparent that the degree of correla,tion is within the range studied. The factors that affect the surface area of the ultimate catalyst have not been thoroughly investigated, but it is felt that variations in the degree and extent of heating during formation and/or activation are of prime importance in this regard. Selwood [12] described in considerable detail. the preparation of iron oxide free from ferromagnetlsI?’ When Selwood’s directIOns were not followed, as In the preparation of unsupported h ydrous ferric oxide, appreciable ferromagnetism was Yet, these were the very samples that exhIblted the hio-hest catalytic activity. Admittedly, very little is Imown regarding the e ffect of f erromagnetism on catalytic activity. In discussing t he applicability of ferromagnetic studies to structU1:e lems Selwood [12] sLates, FerromagnetIsm, Il1 to paramagnetism, is a cooperative phnomenon. A substance does not become femagnetic until the grain size exceeds a certain critical size, sometimes referred to as tbe femagnetic domain. This!s a .s.ituation which som?what weakens the applicabIlity of ferromagnetIc Rtudies to catalyst stl’ucture problems. Those particles which are t.oo finely divicl?d to ferromagnetism are preClsely those whlCh are likely to exhibit the most catalytic activity. Thi is true by virtue of their large urface area, if for no other reason.” From the results of the present work, the influence of ferromagnetism might, indeed, have enhanced the activity of the catalyst. As tated previously, only increase in activity of iron over chromium was to be expected, ba ed on mao-netic considerations. The actual improvement realized was many fold greater than this. The results of the particle-size investigation are interesting. For liquid-phase con;,ersion granules of 20 to 30 mesh proved almost tWlCe as effectIve as 10 to 20 mesh but 40 to 50 mesh was very httle better than 20 to 30 mesh. Tests of particles smaller than 50 mesh showed very little further improvement. For vapor-phase conversion at 76° K was s omewhat different. Here the gam m gomg from 10 to 20 mesh to 40 to 50 mesh was o nly about 15 percent, a nd again beyond 50 mesh there was little or no further improvement. The s pace velocities for ortho-para conversion in the presence of these hydrous iron oxide granules are so high that very small of catalyst, usually 7 ml, were used for measurments .. This was necessary to .avOId complete converSlOn at all but excessIvely hIgh 110w lates. The s maller samples also m ade possible a considerable simplification of the apparatus, as figure 4. The catalyst was merely placed m a slul?-capsule with inlet and outlet hnes t.md dropped through the neck of a standard 50-lIter r ‘It was possible to pick up particles of this catalyst preparation by means o f a small permanent magnet. CONDENSING COIL DEWAR V ENT r LIQUID LEVEL __ FIGURE 4. Drtlw-para catalyst t esting capsule. –1 liquid-hydrogen Dewar. Space velocities fo.l’ version of 25 to 90 percent para, measured m thIS way, had STP values between 2,090 and 2,500 :per minute for liquid-pha e converSlOn when . o-ranules of 20 mesh or malleI’. In terms of hydrogen production, this means 1 liter of catalyst placed in the liquid receiver \ ill convert 150 to 190 liters of liquid per hom from 25 to 90 percent para. 3.2. Hydrous Iron Oxide in eEL Liquefiers Th e hydrous iron oxide catalyst has been used for liquicl-pha e c onver ion in two n:nd 275 KB – 7 Pages