Adsorption and Desorption of Hydrogen by EvaporatedMolybdenum Films at Low TemperaturesBY D. 0. HAYWARD, N. TAYLOR AND F. C . TOMPKINSDept. of Chemistry, Imperial College of Science and Technology, London, S.W.7Received 3 1 st January, 1 966The adsorption and desorption of hydrogen on evaporated molybdenum films has been studiedin the temperature range below 250°K. Sticking probabilities have been measured using a flowtechnique. The initial value (07) is higher than found in flash filament experiments due to themultiple collisions with the surface. At 78°K the adsorbed layer builds up on the outermost regionsof the film and comes into pseudo-equilibrium with the gas phase whilst the less exposed parts arestill bare. Above a coverage of 8 x 1014 molecules per geometric cm2 of film surface, cessation ofthe flow of hydrogen results in a slow decay of pressure P such that 1/P is linear with time.Thisis shown to be due to a slow redistribution of adsorbed hydrogen, via the gas phase, between theconcentrated outer layers and the interior surface of the film. The desorption of hydrogen fromfilms has been studied on warming from 78 to 300°K. Two, and in some cases three, pressurepeaks have been observed, but the spectrum is complicated by surface diffusion and rearrange-ments in the temperature range 170 and 200°K. It is concluded that there is good evidence for onlyone low-temperature state of adsorption of hydrogen on molybdenum; this is largely desorbedbelow 150°K.Much interest has been shown recently in the existence of weakly held statesof adsorption of elementary gases on clean metal surfaces, as demonstrated by suchtechniques as flash desorption, work function measurements and field emissionmicroscopy.Such states, significantly populated only at low temperatures, maywell play an important role in catalytic reactions, especially those which are knownto proceed readily at low temperatures, such as H2/D2 equilibration.The primary purpose of the present work was to investigate the nature of theselow-temperature states for the adsorption of hydrogen on evaporated molybdenumfilms. Experimental data have been obtained on the kinetics of adsorption andredistribution over the surface of the film, the nature of the isotherm at low tem-perature, and the ‘‘ desorption spectra ” of the hydrogen layer.EXPERIMENTALThe apparatus employed was a conventional u.h.v.system capable of an ultimatevacuum of about 7x 10-11 torr. Hydrogen was obtained from the British Oxygen Com-pany in sealed-off ampoules and was purified by passage through a palladium thimble.The apparatus was designed primarily to measure sticking probabilities, although it provedto be suitable for the other measurements. However, as the absolute magnitude of thesticking probability is not of major concern in the present paper, only a brief descriptionof the technique employed in its measurement will be given here. Further details willbe published elsewhere.The method is basically that of Wagener 1 in which a steady stream of gas flows intothe vessel containing the film.The sticking probability is obtained as the ratio of therate of adsorption to the rate of collision with the surface, the latter being calculated fromthe pressure at the film surface. The main problem arises in obtaining a meaningfulvalue for the ambient pressure in a system which is rapidly gettering the incoming gas.The problem has been solved using the arrangement shown in fig. 1 in which the gas is776 ADSORPTION OF HYDROGEN O N MOLYBDENU-Mintroduced at the centre of a bulb on which the film has been deposited. The method issimilar in principle to that used by Clausing,2 although deveIoped by us independently.The molybdenum film is evaporated on to the walls of the reaction vessel from the filamentF, with the diffuser D retracted into the vertical side-arm, and the nickel disc ND heldM.S.FIG. 1.-Cell used for measuring stickingprobabilities.magnetically over the orifice leading to theionization gauge IG.The latter precaution isessential to prevent deposition of film in theionization gauge sidearm otherwise spuriouslyhigh values of sticking probability (sometimesgreater than unity) may be obtained. Afterevaporation, the diffuser D is lowered intoposition and gas from a constant pressurereservoir is introduced through the inlet I.The tubing P leads to the pumps via a Deckervalve. Normally this valve was closed, al-though little difference in the pressure readingswas obtained if it was open, except when thesticking probability was extremely low (< 10-4).MS is a compact 180 degree mass spectrometer,built in this laboratory and is similar in designto that of Goldstone.3The cathodes in both IG and MS wererhenium filaments electrophoretically coatedwith lanthanum hexaboride to reduce theoperating temperature below that at whichsignificant atomization of hydrogen takesplace.Both gauges were operated with anelectron emission current of 10pA, at whichion pumping is insignificant. The massspectrometer and the ionization gauge could be operated simultaneously or separately.When the system was being pumped, all partial pressures were below the detection limit ofthe mass spectrometer (5 x 10-11 torr), and on isolating the system only helium could bedetected within a period of several hours.Desorption spectra were obtained by monitoring the pressure as a function of timeafter the liquid-nitrogen bath around the cell had been removed and the cell allowed towarm up in the atmosphere, with or without lagging around it. The temperature of thefilm was measured approximately by thermocouples fixed to the external wall of the cell.Molybdenum films, weighing between 30 and 50 mg, were usually deposited on the cellwall maintained at about 50°C.The total surface area, as calculated from the total uptakeof hydrogen, was, on average, about ten times the geometric area.RESULTS AND DISCUSSIONSTICKING PROBABILITY DATAFig. 2 shows the variation of sticking probability with coverage when the filmis maintained at 78°K.The initial values are considerably higher than those foundusing metal filaments 4 (-0-3), presumably because incoming molecules may makemore than one collision with the rough surface before being either adsorbed orreflected back into the body of the cell. In the particular experiment shown in fig. 2,the gas supply was stopped for 10-20 min at the points marked A. Up to a coverageof about 8-0 x 1014 molecules per geometric cm2 of film surface this cessation of gasinflow resulted in a very rapid decrease in pressure to the background value ( N 7 x10-11 torr) as shown diagrammatically in fig. 3a. On resumption of flow, the pressureincreased abruptly to its previous value, This behaviour is to be expected when thD. 0. HAYWARD, N. TAYLOR AND F.C. TOMPKINS 77film itself is pumping rapidly and the number of molecules in the gas phase isnegligible. Above this coverage, however, the pressure decreased only slowly tothe background value on stopping the gas supply, despite the high sticking proba-bility, as shown in fig. 3b; and on resumption of gas flow, the sticking probabilitywas initially much higher than before the stoppage, but fell to near the previousvalue on addition of a further 1013 molecules per cm2 (see fig. 2). At 78°K the1 6 3 0 L 2 10 2 0 ' 30 . ' 40 . ' 5 0 ' 'coverage (molecules per cm2)for 10-20 min at the points marked A.FIG. 2.-The variation of sticking probability with coverage at 78°K. The gas supply was closedtime 3FIG. 3.-Diagrammatic representation of pressure changes on closing and opening gas supply to(4 (b)film.A, gaq supply closed ; B, gas supply opened ; PO, background pressure.adsorbed hydrogen is immobile, and the more accessible (" outer ") regions of thefilm are covered first. The slow decay in pressure arises when some of these moreaccessible regions have become saturated and are in pseudo-equilibrium with thegas phase. On stopping the gas supply, the pressure falls, hydrogen desorbs fromthese saturated regions and is transferred via the gas phase to those regions of theinner surface that remain unsaturated. On restarting the gas flow the outer regionsof the film are able to adsorb further quantities of hydrogen and consequently highsticking probabilities are observed. However, these regions rapidly become satur-ated again and the sticking probability falls.On the basis of this model, stickin78 ADSORPTION OF HYDROGEN O N MOLYBDENUMprobabilities calculated from AP2 (see fig. 3b) should remain constant providedAPl B AP2, because the same, virtually saturated surface is exposed to the gas phaseafter each cessation of gas flow. This is found experimentally, as shown in fig. 2.Also, the onset of the slow process corresponds approximately to a break in thesticking probability curve, as is best seen in the inset to fig. 2. This is predicted bythe model outlined above since at the start of the slow process the most exposedregions of the film are no longer adsorbing gas.ISOTHERM AT 78°KThe molybdenum films could not readily be saturated by adsorbing hydrogenat 78°K due to a persistent slow uptake of gas, as is often found with evaporatedmetal films.However, by cycling the temperature of the film between 78 and300°K in the presence of hydrogen, this slow uptake was reduced to negligibleproportions, and it was possible to obtain an isotherm that was effectively reversible.coverage (molecules per cm2)FIG. 4.Temkin isotherm at 78°K.This is shown in fig. 4, where logP is plotted against coverage. The linear plot,which corresponds to a Temkin isotherm, indicates that the heat of adsorptionvaries linearly with surface coverage within the range of the data.KINETICS OF THE REDISTRIBUTION PROCESSFor the slow process, following cessation of the gas supply, the reciprocal ofthe pressure varies linearly with time as shown in fig.5. This relationship is wellobeyed both near the onset of the slow process when the pressure decay is stillfairly rapid, and also when the film is nearing saturation and the process takes 10-20min. Such a relationship can be derived theoretically if two assumptions are made :that the film consists of a system of pores of uniform cross-section, and that thesaturated areas of the film are at all times in equilibrium with the gas phase. Thesituation is shown diagrammatically in fig. 6 D. 0. HAYWARD, N. TAYLOR AND I;. C. TOMPKINS 79The saturated layer, each element of which is in equilibrium with the gas phaseimmediately above it, covers the outer surface and extends down the pore to thepoint E. The equilibrium has been found experimentally to obey the Temkin isotherm,In P = aN,+k,where NA is the number of molecules adsorbed per cm2, and c( and k are constants.4- 7time (sec)FIG.5 - 4 4 and @).-Plots of reciprocal pressure against time after closure of gas supply to film.l iPO D‘ EFIG. 6.-Diagrammatic representation of the redistribution process down a pore.The arrows indicate the direction of net gas flow.Differentiating with respect to time,This equation applies provided that 8N~Iat is small compared with the rates ofadsorption and desorption, i.e., a pseudo-equilibrium is maintained.By application of Fick’s first law of diffusion to the net flow of gas across theline AB,8 In Plat = adN,/dt. (280 ADSORPTION OF HYDROGEN ON MOLYBDENUMwhere ( a N ~ / d t ) , is the total number of gas molecules flowing across AB per sec,A is the cross-sectional area of the pore and D is the diffusion coefficient, assumedconstant.Similarly, for the flux across the line CD,The total rate of desorption from the saturated walls of-(dNA/dt)Ldx, where L is the circumference of the porethenrate of desorption =the element ABCD iswall.Using eqn. (2),(5)The equivalent of Fick's second law may now be obtained by considering massbalance in the element ABCD. The rate of change in the number of gas phasemolecules due to changes in pressure may be neglected, since it is very small com-pared with the rate of flow and the rate of desorption from the walls. Thereforeor, substituting eqn.(3) and (4), and rearranging, (Z)=&(y).A solution of this equation, applicable to the present problem, isL ( a - b x - x 2 )2aAD ( t + c ) 'p = -(7)where a, b and c are constants, the values of which are determined by the particularboundary conditions imposed. The pressure that is measured is Po, the value of Pat x = 0. Thus,This equation gives a linear relationship between reciprocal pressure and time, asfound experimentally, and gives support to the model of the redistribution processproposed.Po = aL/2aAD(t+c). (9)DESORPTION SPECTRAThe normal procedure in obtaining " desorption spectra " was to admit a knownquantity of hydrogen to a molybdenum film maintained at 78"K, record the pressurechanges on warming to room temperature, re-cool and repeat the process withfurther doses of hydrogen until the film was saturated. The hydrogen added at78°K is concentrated on the outer regions of the film, but, during the warm-up,redistribution takes place, some hydrogen being desorbed from this concentratedlayer, and re-adsorbed on the internal surface, except when the whole film is nearingsaturation.Redistribution of hydrogen also occurs by surface diffusion atsufficiently high temperatures.Fig. 7 and 8 show how the hydrogen pressure varies as the temperature of a filmis increased for a series of warm-ups at increasingly higher hydrogen coverages.In table 1 is recorded the quantity An of hydrogen adsorbed at 78°K prior to eacD. 0. HAYWARD, N. TAYLOR AND F.C. TOMPKINS 81warm-up, the total quantity TZT of hydrogen on the film, and an approximate valuefor the quantity n of hydrogen in the concentrated outer layer, each expressed inmolecules per geometric cm2 of film surface. n is evaluated as (An+ni/R), wheretemp. (“K)FIG. 7.-Desorption spectra between 78 and 300°K for a single molybdenum film,The numbers refer to desorptions at successively higher coverages.3temp. (“K)FIG. 8.-Desorption spectrum with film isolated from the pumps.Desorption from the same film as in fig. 782 ADSORPTION OF HYDROGEN ON MOLYBDENUMn> is the total amount of gas adsorbed up to the previous warm-up, and R is the ratioof real area to geometric area of the film (about 7.5 in this case). Below n = 6.8 x 1014molecules cm-2 no desorption is observed on raising the temperature of the film.This is in good agreement with other runs on 10 films in which approximately6-5 x 1014 molecules of hydrogen cm-2 must be added to a clean film at 78°K beforea desorption spectrum is observed.The slow redistribution of hydrogen via the gas phase at 78°K obeys the kineticspreviously discussed, and is always associated with occurrence of a desorptionspectrum. The discrepancy in the coverages at which the two phenomena are firstobserved arises because of differences in the quantities of weakly held hydrogenthat can be detected by the two methods.TABLE 1"T AntotaI HZ adsorbed(molecules cm-2 x 10-14H2 increment at 78°K(molecules cm-2 x 10-14)HZ concn.i: outer layer,(molecules cm-2 x 10-14)spectrum Of the desorption4.579.2 113.4418.6223.4527.8332.0745.574.574.644.235-1 84.834.384.2413.504.575.2 15-386.867.167.327.72no desorption3,9 912345In desorptions 1 to 4 the film was pumped for most of the time, although periodicisolation from the pumps made no detectable difference to the pressure-time curves,presumably because pumping by the inner areas of the film was much greater thanthat through the exhaust port. In desorption 5 the whole surface area of the filmwas close to saturation, and the warm-up was carried out with the system isolated.Pressure peaks or shoulders can be observed in fig.7 and 8 at 130, 170 and 220°K.The sharp drop in pressure between 170 and 200°K is attributed to the onset omobility within the adsorbed layer.This results in weakly held hydrogen, presentat 78°K in the concentrated outer layer of the film, migrating to strongly adsorbingsites on the inncr bare surface. This hydrogen would otherwise have desorbed attemperatures greater than 170°K. Thus, on this basis, the pressure maximum at170°K is not attributed to the existence of a separate adsorbed phase, because thepressure-temperature profile does not reflect the variation of population densitywith bond energy at this point.That mobility becomes important at 170°K is shown by the following experi-ment. Desorption data were obtained as before except that the liquid-nitroger,cooling bath was replaced around the cell at various stages during the desorpiionprocedure.In addition, sticking probabilities were measured during the adsorp-tion of hydrogen at 78"K, aiid they are displayed in fig. 9. After each teinperaturecycle there is an increase in the sticking probability measured at 78"K, due to theredistribution of the hydrogen that was concentrated in the outer regions of the film.The addition of mere gas at 78°K causes a fall in the sticking probability, and thenumber of molecules required to reduce it to its value prior to the temperature cyclegives a measure of the extent of the redistribution process. However, the redistributionis partly via the gas phase, and partiy via the surface. The contribution of the formerATJG can be calculated approximately from the pressure-time profile during the warm-up, and that of the latter ANs by subtracting the gas phase contribution from the totalD.0. HAYWARD, N. TAYLOR AND F. C. TOMPKINS 83The figures in table 2 are calculated in this way for the run shown in fig. 9. It maybe seen that at 170°K only a small number of adatoms have migrated, whereas at 200°Kthe majority have done so.coverage (molecules per cm2)FIG. 9.-The effect of redistribution on sticking probability.Open circles, sticking probabilities measured at 78°K ; filled circles, sticking probabilities measuredat 300°K; A, cooled to 78°K; B, warmsd to 300°K; C, warmed to 190°K and recooled; D,warmed to 200°K and rzcooled ; E, warmed to 170°K and recooled.However, complete equilibration does not appear to be readily achieved evenat higher temperatures, as evidenced by curve number 3 in fig.7. i n this run thetotal coverage of hydrogen on the film was defiritely below that required for satura-tion, even at 300"K, and yet desorption was still occurring at 230°K and above.TABLE 2total number of H2molecules adsorbed(cm-2x 10-14)18.1823-6431.9237.2339.0342.2550.0453.4154.82AN gas phase(molecules cm-2x 10-14)0.0420.120-290.320-970.650.940.472.3AN surface2.76.04-20.93-06.02.00-52.0(moleculzs cm-2 x 10-14)highest temp. reachedduring warm-up (OK)190300200170190300190170190If equilibration were complete at 200"K, only the high energy sites would be coveredabove this temperature, and desorption would not commence until the film wassaturated under the prevailing conditions of temperature and pressure.In runnumber 4, where the total coverage is higher, saturation appears to have been achievedabove about 240°K as the pressure is rising steeply. In fig. 8 the film is close tosaturation at 78°K and is apparently in pseudo-equilibrium with gas phase between140 and 170"K, above which the layer starts to rearrange. As the film is isolatedfrom the pumps and the number of molecules in the gas phase is negligible com-pared with those adsorbed on the film, the last part of the pressure-temperatureplot in fig. S should correspond to an isostere, provided equilibrium is achieved.It is found that this is not so below 250°K but, from other desorption spectra84 ADSORPTION OF HYDROGEN ON MOLYBDENUMisosteres were successfully constructed for temperatures between 250 and 270°K givingheats of approximately 17 kcal/mole.Fig.10 shows the desorption spectra for two films, both of which had beeneffectively saturated with hydrogen at 78"K, the rate of adsorption at this tem-perature having decreased to a negligible value. Part of the difference in the twodesorption spectra is due to the fact that in run (a) an attempt was made to equilib-rate the adsorbed layer by cycling the temperature between 78 and 220°K duringthe adsorption, whereas in run (b) the film was saturated with hydrogen at 300"K,and the remainder of the hydrogen was added with the film maintained at 78°K.The sharp decrease in pressure between 170 and 200°K in run (b) is again attributedto a rearrangement within the adsorbed layer, so that more hydrogen is accommodatedon the surface.I 1x id210 -5 -I Itemp.("K)too I 5 0 200 2 5 0FIG. 10.-Desorption spectra for two films.(a) cell lagged, adsorbed layer equilibrated ;(b) cell unlagged, adsorbed layer not equili-brated. Both desorptions were carried out withcontinuous pumping.150 200 2 !temp. ("K)3FIG. 11 .-Number of molecules desorbed perOK against temperature. (a) and (b) refer tothe same desorptions as in fig. 10.However, fig. 10 (and to a lesser extent, fig. 7) is somewhat misleading becausethe rate of temperature rise is continuously decreasing and this has a spurious effecton the shape of the pressure-temperature profile.Further, in run (b) the cell wasunlagged and the warm-up was rapid, whereas in run (a) the cell was lagged. It canbe seen from fig. 11 that if the number of molecules desorbed per cm2 per degreerise in temperature is plotted (instead of the instantaneous pressure), the shape ofthe desorption curve is considerably altered. This is especially true at the highesttemperatures where the rate of rise of temperature tends to zero. Since the systemis being pumped, it follows that the pressure must also tend to zero, irrespective ofthe population distribution of the various binding energies.NUMBER O F LOW TEMPERATURE STATES OF ADSORPTIONIt has been common practice recently to identify each maximum or shoulderon a pressure-time desorption profile with a separate state of adsorption.Theprevious discussion suggests that this may not always be a valid deduction, evenwhen using metal filaments or sheets which expose the whole of their surface directlyto the gas phaseD. 0. HAYWARD, N. TAYLOR AND F. C. TOMPKINS 85The present investigation provides evidence for, at most, two states of adsorptionin the low temperature range : one desorbing below about 150"K, and anotherdesorbing around 220"K, although the latter is doubtful, since it is not observedin all runs. Two studies4Y5 of the adsorption of hydrogen on molybdenum fila-ments or ribbons have been made, but, unfortunately, neither provides informationon the temperature region below 200°K.Pasternak and Wiesendanger 4 foundevidence for two states of adsorption when desorbing from 225"K, but since theirsystem contained no means of chemical identification, the possibility that one ofthe states was due to CO or some other contaminant cannot be ruled out (seeHickmott,6 and Moore and Unterwald 5). However, valid comparisons can prob-ably be made with the related H2/W system ; Hickmott 6 found one state of ad-sorption above 195"K, designated j?, and another desorbing below this temper-ature, designated a. Recently, Ricca, Medana and Saini 7 have obtained desorptionspectra for hydrogen adsorbed on a tungsten sheet that show six pressure peaks,labelled y1, 7 2 , al, UZ, p1 and p2, the y peaks being identified with Hickmott's a phase.However, the differentiation of the desorption above 190°K into a and p peaks maybe artificial due to the fact that the desorption was carried out in two stages, 78 to300°K and 300 to 600°K.No such differentiation is observed by other workers 5 9 6who desorbed in a single flash.In neither of the two investigations discussed above was an attempt made toachieve equilibrium in the adsorbed layer at 78°K by cycling the temperature duringadsorption of hydrogen, although Ricca et aZ.7 saturated the surface at 300°K beforecooling to 78°K in some of their experiments. It is relevant, therefore, to comparetheir desorption spectra with curve (b) in fig. 11, rather than curve (d). This com-parison would suggest that the distinction between Hickmott's a and p states, andbetween Ricca et al.'s y2 and a2 states, arises largely because of a surface rearrange-ment, which enables weakly held hydrogen to migrate to more strongly adsorbing sites.We conclude, therefore, that there is good evidence for only one low-temperaturestate in the adsorption of hydrogen on Mo or W, and that it is largely desorbed by150°K. This state can probably be identified with the y1 peak of Ricca et uZ.andwith the onset of the a peak in Hickmott's work. This should not be understoodto preclude the many adsorbed complexes observed by Rootsaert, van Reijen andSachtler 8 in the field emission microscope, but merely to indicate that the heats ofadsorption of these complexes are not sharply defined, but vary with coverage andoverlap one another.NATURE OF THE LOW-TEMPERATURE STATE OF ADSORPTIONThis state is responsible for the slow redistribution process that occurs via thegas phase at 78°K when the flow of hydrogen to the molybdenum film is stopped.The lowest coverage at which it is observed is about 6-5 x 1014 molecules per geo-metric cm2 of film surface adsorbed at 78°K.However, if the film is first saturatedwith hydrogen at 300"K, this weakly held state can be detected after very smallquantities of hydrogen have been adsorbed at the low temperature. This behavioursuggests that the weakly held hydrogen can only exist above a layer of the stronglyheld p state. One possibility is that it is molecular and is held over single vacanciesin the atomic layer. This would explain why this state is less pronounced whenthe surface layer is equilibrated by temperature cycling (see fig. ll), as the singlevacancies would be largely eliminated. Similar behaviour was observed by Riccaet aZ.7 who found that their y1 peak was considerably reduced if the surface wassaturated at 300°K prior to the adsorption at 78°K86 ADSORPTION OF HYDROGEN ON MOLYBDENUMWe gratefully acknowledge the support of the Ministry of Aviation in carryingout this work and also in providing one of us (N. T.) with a grant, and other financialassistance from Shell Research N.V., Amsterdam.1 Wagener, Brit. J . Appl. Physics, 1950, 1, 225.2 Clausing, Trans. 8th Vacuum Symp., 1961, 1, 345.3 Goldstone, Rev. Sci. Instr., 1964, 35, 1265.4 Pasternak and Wiesendanger, J. Chem. Physics, 1960, 34,2062.5 Moore and Unterwald, J. Chem. Physics, 1964, 40,2626.6 Hickmott, J. Chem. Physics, 1960, 32, 810.7 Ricca, Medana and Saini, Trans. Faraday SOC., 1965, 61, 1492. * Rootsaert, van Reijen and Sachtler, J. CataZysis, 1962, 1, 416