Critical potentials of hydrogen in the presence of …rspa.royalsocietypublishing.org/content/royprsa/110/754/...Critical Potentials of Hydrogen in the Presence of Catalytic Nickel

464

Critical Potentials of Hydrogen in the Presence of Catalytic Nickeland Copper*

By J. H ulton W o l fe n d e n , B.A., Lecturer in Chemistry, Balliol College, Oxford ; Jane Eliza Procter Visiting Fellow, Princeton University, N.J., U.S.A.

(Communicated by Dr. E. F. Armstrong, F.R.S.—Received December 8, 1925.)

A recent paper by A. W. Gaugerf opens the possibility of an interesting and novel method for gaining information on the mechanism of hydrogenation catalysis by nickel and other metals. Gauger bombarded with electrons a target of catalytic nickel in the presence of a low pressure of hydrogen, and observed the critical electron velocities at which the occurrence of characteristic radiation could be detected by the photo-electric current which it produced by falling on a collecting plate connected to an electrometer.

Gauger reported a number of critical potentials, several of which he associated with the first members of the Lyman series and the ionisation of atomic hydrogen respectively. The results were interpreted as indicating the presence of atomic hydrogen in the system comprising hydrogen and a nickel catalyst.

A closer examination of this very suggestive work seems to reveal certain defects of method which render rather equivocal any conclusions drawn from the results obtained. The following criticisms are suggested :—

(1) An examination of the specimen curve in Gauger’s paper (which was printed upside down relative to the co-ordinates, owing to a printer’s error) shows that all except two of the breaks shown are “ negative ” breaks ; that is to say, the radiation increases less rapidly with voltage above the critical potential than below it. Breaks of this type are comparatively rare,$ and have never been reported by any other investigator of the critical potentials of molecular or atomic hydrogen. The as yet unpublished work of 0. H. Thomas on soft X-rays from nickel also fails to show a negative break associated with any of the numerous critical potentials observed. This unusual form of the curve suggests that perhaps some of the observed breaks are fictitious.

* Contribution from the Laboratories of Physics and Physical Chemistry, Princeton University, Princeton, N.J., U.S.A.

f * J. Am. Chem. Soc.,’ vol. 46, p. 674 (1924).J Kurth, ‘ Phys. Rev.,’ vol. 18, p. 468 (1921).

on June 12, 2018http://rspa.royalsocietypublishing.org/Downloaded from

http://rspa.royalsocietypublishing.org/

Critical Potentials o f Hydrogen. 465

(2) The aggregate correction for initial electron velocity and contact potentials seems to have been applied wrongly. As far as one can determine from the paper, the correction should have been added and not subtracted. This would increase all the observed critical potentials by 0-4 volt.

(3) The magnitude of this aggregate correction seems to be incorrect for two reasons : first, because the method of application assumes that the relative sensitivities of the apparatus for detecting radiation and primary electrons are the same* ; secondly, because the overall contact potential corrected for is that between the filament (tungsten) and the plate (platinum) connected to the electrometer, whereas the contact potential involved in the main experiments is that between the filament and the target of catalytic nickel.

Summarising the criticism relative to the correction for initial*velocities and contact potentials, it is submitted that the critical potentials reported by Gauger are subject to a correction of

_[_ 0 • 4 — x + yvolts,where x depends on the difference in sensitivity of the apparatus for detecting radiation and for detecting primary electrons, and y is the contact potential between platinum and the catalytic nickel target employed.

(4) A final uncertainty arises as to the origin of the radiation measured.It is conceivable that hydrogen gas in the absence of the catalytic metal might have given breaks characteristic of atomic hydrogen, ov/ing to thermal dissociation by the hot tungsten filament. And even if this possibility is excluded, there are still three alternative sources of the radiation—namely, the mass of residual hydrogen gas, the hydrogen at the catalyst surface, and, finally, the catalytic metal itself. Blank experiments with no gas present negatived the last of these alternatives, but none of the experiments performed served to determine whether the effects observed originated in the body of the gas or at the catalyst surface.

The present work aimed at a repetition and extension of Gauger’s w'ork, modifying the method in such a way as to eliminate the features which make his results of an equivocal character. In particular, the correction for initial velocity and contact potentials has been applied in such a way as to put the

* Smyth, ‘ Phys. Rev.,’ vol. 14, p. 409 (1919).



466 J. H. Wolfenden.

absolute values of the critical potentials on a surer basis. Furthermore, by comparison with results obtained when the preponderating effect was due to a body of plain molecular hydrogen, evidence was afforded that the effects measured were associated with the hydrogen at the catalyst surface. The work has been extended to compare the effect of catalytic nickel with that of plain nickel and also that of catalytic copper.

Materials and Apparatus.Catalytic nickel was prepared by the reduction of the oxide by hydrogen

at 370° C. The oxide was prepared by the ignition of a thin film of chloride-free nickel nitrate solution spread over a thin rectangular plate of massive nickel.

Catalytic copper was prepared in an exactly similar manner except that the reduction was carried out at 200° C. In both these cases and in the case of the plain nickel target, the metal was heated to redness vacuo by an induction furnace before the application of the catalytic layer.

The apparatus is illustrated in fig. 1. The drawing is roughly to scale, the diameter of the central bulb being 9 cm. and the distance between target and collecting disc being about 2-5 cm. The electrical connections are essentially the same as those employed by Gauger, except for the modification necessary to admit of the measurement of ionisation currents as well as radiation effects. Electrons from a notched platinum strip coated with barium and strontium oxides are accelerated towards the target of catalytic metal, which rests on a water-cooled copper butt-end. The radiation produced falls on a platinum disc connected with a Compton electrometer (whose sensitivity remained very steadily at about 2,300 mm. per volt). Between the target and the collecting disc are placed three platinum gauzes whose voltages (illustrated on the diagram) are arranged to prevent any ions or electrons from reaching the collecting disc. Thus, with the voltage of the middle gauze at 52 volts, the electrometer readings measure the photo-electric current due to the radiation from the target and from the gas between filament and target. By means of a switch, the voltage of the middle gauze could be changed to 8 volts, under which circumstances positive ions are accelerated towards the collecting disc, so that the combined effects of ionisation and radiation are observed in the electrometer current.

The targets are mounted along two parallel thin nickel wires connected at their ends to two cylinders of soft iron. By the use of an electromagnet it is possible to bring any one of the targets in front of the electron source and in contact with the water-cooled butt-end.



Critical Potentials of Hydrogen. 467

The leads to the filament are introduced through a tube which enters the central bulb perpendicular to the plane of the paper. All ground-glass joints are made tight with de Khotinsky cement, which is disposed so as to reduce the vapour pressure of the cement within the apparatus to a minimum.

The experimental tube is connected through two liquid air traps to a hydrogen generator (in which hydrogen was generated by the electrolysis of baryta solution), a McLeod gauge and a mercury diffusion pump. Before use the

-TO ELECTROMETER

TO PUMPFILAM ENT

[SOFT]IirqnI-

TARGETS

W ATER-COOLEDCOPPER BUTT-END

B R A SS-G L A SS GROUND JOINT

W ATERF ig. 1.—Apparatus.

Voltage Scheme : Collecting Plate ............................................... 0 volts.Grid (1) ................................................................ 6 volts.Grid (2 )............................................................... 8 or 52 volts.Grid (3 )................................................................ 9 volts.Filament ........................................................... 16 volts.Target ............................................................... 16 + VA .

tube was baked out in vacuo for about ten hours at 370°. Higher temperatures would have de-activated the catalysts.




The differences between this experimental tube and that employed by Gauger consist essentially in the use of an oxide-coated filament with correspondingly smaller chance of thermal dissociation in the device, which permits of the use of several targets without opening the tube, and in a general arrangement by which the collecting disc subtends a considerably greater angle at the target with a correspondingly greater sensitivity for detecting radiation and by which a closer spacing between filament and target accentuates the effects from the catalyst surface relative to those originating in the body of the gas.

Experimental Procedure.The filament current was switched on, about 6 amperes being sufficient to

raise the filament to a dull red heat which gave the desired thermionic emission. Hydrogen was admitted to a pressure of about 10 mm. of mercury and allowed to remain in contact with the target for five minutes. Pumping was then started, and as soon as the pressure had fallen to 2 X10-4 mm., the first readings wrere taken.

As the accelerating voltage was raised in steps of 0 • 25 volt, observations were made of the electrometer current and the thermionic current between filament and target (measured on a micro-ammeter) corresponding to each value of the accelerating voltage. The experimental curves were obtained by plotting accelerating voltage (VA) as abscissa, and the quotient of electrometer current divided by thermionic current as ordinate. At the lower voltages the electrometer current was measured by “ rate of deflection ” ; at the higher voltages, shunts were introduced, and the “ constant deflection ’’ method employed.

Results.Radiation Effects with Plain and Catalytic Nickel.—The first experiments

consisted in a repetition of Gauger’s work, employing a tungsten filament and measuring the photo-electric current due to radiation only. Determinations were made with both plain nickel and catalytic nickel as targets. Many curves were obtained, of which curves (1) and (2) in fig. 2 are typical exam ples.

Curve (1) is for catalytic nickel and curve (2) is for plain nickel.* Except for minor irregularities, which the study of a large number of curves showed to

* In these curves and in those that follow, the ordinates are of arbitrary magnitude, furthermore, in order to get curves for comparison in juxtaposition on the same diagram, the horizontal zero axis has been shifted up and down for the various curves on the same diagram. All voltages are uncorrected observed voltages.



Critical Potentials 469

be absolutely random in their occurrence, it is clear that the curves are smooth ones without any readily located discontinuities, and also that catalytic and plain nickel show no perceptible difference from one another.

It Avas c o n c lu d e d t h a t n o r e s u l t s b e a r in g a n y s im p le i n t e r p r e t a t io n w o u ld

C U R V EC U R V E 2

9 10Va v o l t s

F ig. 2.—Radiation only.—Curve (1): Catalytic nickel. Curve (2 ) : Plain nickel.

be obtained by this method. The decision was therefore made to measure ionisation currents instead of radiation effects alone in the subsequent experiments. The study of ionisation rather than radiation has several advantages :—

(1) The currents to be measured are considerably greater.(2) The results are more readily interpretable. While the radiation

potentials of both atom and molecule are numerous, and not entirely of uncontroversial origin, the ionisation potentials of the atom at 13-4 volts, and of the molecule at 16 volts are strongly marked and unambiguous.

(3) By the simple device of carrying out several runs at a pressure of 10~2 mm.of hydrogen, it is possible to make an unequivocal determination of the aggregate correction for initial electron velocity and contact potentials. Under such conditions the effects observed are due almost exclusively to molecular hydrogen, so that the location of the single strong break in the resultant curves determines the value of the observed voltage corresponding to a true electron velocity of 16 volts.




Arguing from these considerations, all subsequent experiments were carried out by measuring ionisation currents and employing an oxide-coated filament.*

Ionisation Measurements with Plain and Catalytic Nickel.—Measurements were carried out with plain and catalytic nickel at 10~4 mm., and at 10"2 mm. pressure of hydrogen. Specimen curves are shown in fig. 3. Curve (3) shows the result with a pressure of 10-2 mm.

CURVE 3CURVE 3

CURVE 4 ,

CURVE 4

CURVE 5CURVE 5

9 10

Va voltsFig. 3.—Ionisation and radiation.—Curve (3): Hydrogen at 10 ~2 mm. (plain nickel target).

Curve (4): Catalytic nickel and hydrogen at 10“ 4 mm. Curve (5): Plain nickel an< hydrogen at 10“ 4 mm.

* It was found that, presumably owing to impurities in the tungsten, the tungsten filament gave a copious positive ion emission. The oxide-coated filament was free fromthis defect, and also minimised the possibilities of thermal dissociation.




As one would expect, there is only one well-marked discontinuity, namely, the very strong one due to ionisation of the hydrogen molecule. This occurs at an observed voltage of 17-6 volts, indicating a correction of — 1 - 6 volts to be applied to all observed voltages in experiments with a nickel target. Curve (3) was obtained with plain nickel, but exactly similar curves were obtained with catalytic nickel; this was to be expected, since nearly all the effects observed at such a pressure must have been due to the body of gas between the filament and the target.*

Curves (4) and (5) were obtained with catalytic and plain nickel respectively at a gas pressure of 10"4 mm. ; that is to say, under circumstances where the predominating effects originate in the gas at the metal surface. It is at once clear that the form of the curve is widely different from that obtained with plain molecular hydrogen. The strong breaks are observed at corrected voltages of (13-0— 1-6), or 11-4 volts, and at (15-0— 1-6), or 13-4 volts. A weaker break is observed near 18 volts, but the change of direction is so slight as to make exact location difficult, and the break most probably corresponds to the strong one at 17-6 volts in curve (3). In some of the unpublished curves this last break is almost impossible to detect at all, although the two lower breaks are just as well defined as ever.

The corrected voltages for the two lower breaks on a series of three curves for each target are given below :—

Catalytic Nickel. Plain Nickel.First break ......................... 11-3, 11-3, 11-3 11-2, 11-4, 11 -6Second break ..................... 13-4, 13-4, 13-4 13-4, 13-4, 13-4

The observed facts are, then, that when conditions are changed by lowering the pressure so that the effects observed are due to hydrogen plus catalyst, rather than hydrogen alone, the 16-volt break due to molecular hydrogen almost disappears and is replaced by two breaks at 11*4 volts and 13*4 volts respectively, of which the latter is the stronger.

The 13-4-volt break can scarcely be other than that due to ionisation of the hydrogen atom. The interpretation of the 11 *4-volt break is less simple. There are four possibilities : the break may be associated with the hydrogen molecule, with the hydrogen atom, with the nickel target, or with a hydrogen-

* It will be observed, that this hydrogen molecule break is not as sharp as the others. This would be expected because of electron distribution (Olmstead, ‘ Phys. Rev.,’ vol. 20, P- 623 (1922)), and also because the arrangement of the electrodes is such that ionisation can take place at any point in the path of the accelerated electron. When effects from the metal target are measured, on the other hand, all electrons fall through the same potential drop and hence the breaks are sharper.




nickel complex. The last two alternatives are ruled out by the fact that a break at about the same voltage occurs with a target of catalytic copper • that it is not associated with the target alone is also confirmed by the observation described below, that effects due to gas-free nickel are negligible compared with the currents measured in the above determinations. This leaves the two alternatives of the hydrogen molecule and the hydrogen atom. The fact that it does not appear in the high-pressure measurements suggests that it is not associated with the molecule. The break is almost certainly to be identified with that observed by Olmstead* at the same voltage. Olmstead adduces evidence to show that the break is due to ionisation of the hydrogen molecule. Olmstead’s curves, however, show that the 11-4-volt break does not appear in his “ Grid off for Ionisation ” curve (where dissociation is negligible), whereas it makes its appearance, together with the 13-4-volt break in the “ Grid on for Ionisation ” curve, where the incandescent grid is introduced to produce thermal dissociation of the hydrogen. It is significant that the “ Grid off ” (no dissociation) curve is closely similar to curve (3) of the present work, while the “ Grid on ” (thermal dissociation) curve is like curves(4) and (5).

Thus, whatever be the interpretation of the 11 *4-volt break, it seems to be clearly indicated that the presence of nickel (both in the catalytic and unactivated condition) brings out strongly the ionisation potential of atomic hydrogen, and, in general, has a similar effect to that obtained by Olmstead in the use of an incandescent grid to produce thermal dissociation.

Ionisation Measurements with Catalytic Copper.—Similar experiments were carried out with copper, except that the absence of a plain copper target precluded any measurements with unactivated copper. However, the close similarity between catalytic and plain nickel suggested that experiments with plain copper would be superfluous. Determinations were therefore made with catalytic copper as a target at hydrogen pressures of 10-2 and 10~4 nun. respectively. Typical curves are shown in fig. 4.

Curve (6) was obtained at the higher pressure and shows one strong break at 18 • 6 volts, indicating a correction of —2-6 volts to be applied to all observed voltages. Curve (7) was obtained at the lower pressure and shows breaks at corrected voltages of 11 -4 and 13-3 volts with a much weaker effect at 16 volts. The figures for the location of the breaks on a series of three curves at each pressure are as follows (voltages being corrected):

* ‘ Phys. Rev.,’ vol. 20, p. 013 (1922).




Pressure of 10 4 mm. 11-4, 11-3 and 11-4.13-2, 13-2 and 13-4.Break at 16, weak and diffi

cult to locate exactly.

C U R V E 6C U R V E S

CU R V E 7$C U R V E 7

12Va volts

Fig. 4.

The results are thus closely similar to those obtained with catalytic and plain nickel, and suggest that in the case of catalytic copper also atomic hydrogen is present in appreciable quantity. It is reassuring to observe that,

v o l . c x . — a . 2 i

Pressure of 10~2 mm.No break observed at 11*4. No break observed at 13-3. 16-1, 15-9 and 16-1.




in spite of the large initial velocity and contact potential correction, the position of the breaks agrees so closely with the observations in the case of nickel, where different corrections are applied.

Evidence that the Effects observed at 10“4 mm. Pressure are associated Hydrogen at the Catalyst Surface-There are two pieces of evidence that, in experiments at 10“4 mm. pressure, we are observing effects originating near the catalyst surface and not in the body of the gas. The first is the fact that the form of the current-voltage curve is so different at the lower pressure from that observed at the higher pressures. The second is afforded by comparing the ionisation current corresponding to a given voltage at 10-4 mm. and at 10-2 mm. pressure. If no surface effect were introduced at the lower pressure we should expect, on the same principle as that of the ionisation gauge, that current would be proportional to pressure. Observation showed, on the contrary, that the current at the lower pressure was anywhere from five to ten times that corresponding to pressure-proportionality. This was true of both plain and catalytic nickel and shows that in both cases a strong surface effect was being measured at the lower pressure.

Evidence that the Metal of the Target Itself does not contribute appreciably to the Effects observed.—Such a test was necessary to lend conclusiveness to the foregoing experiments. It had been found impossible to do a blank experiment of the kind described by Gauger, because in the presence of the adsorbent metal targets and the cold (and unbaked) copper butt-end, the last traces of gas were tenaciously retained and the necessary freedom from gas in the tube could not be obtained. The targets were therefore removed and the butt-end replaced by a disc of plain nickel supported on a thin nickel stem screwed into the brass-glass joint below. The target had been “ de-gassed ” by treatment in the induction furnace and the experimental tube was baked out for a longer period and at a higher temperature than the presence of heat-sensitive catalysts had previously permitted.

With such a target it was possible to measure the effects due to gas-free nickel alone. A run was carried out with no hydrogen admitted and with the gauzes arranged to measure radiation only. The much smaller currents observed were of such magnitude as to show that radiation from the target itself could not have made any appreciable contribution to the previous

measurements.

Discussion of Results.The series of experiments described above is a demonstration of the presence

of atomic hydrogen in the systems comprised by hydrogen in contact with




catalytic nickel, plain nickel and catalytic copper respectively. The demonstration has several limitations. In the first place, it is only qualitative ; strictly speaking, it tells us nothing of the proportion of atomic to molecular hydrogen in the adsorption layer, although the strength of the 13 *4-volt break precludes the possibility of atomic hydrogen being present merely in minimal quantity. In the second place, the experiments do not exclusively correlate the presence of atomic hydrogen with catalytic activity. For this reason it was proposed to extend the experiments to a metal of very small catalytic activity like lead ; it is regretted that circumstances prevented this.

If we assume such a correlation to exist, we have to explain the close similarity between the results obtained with plain and catalytic nickel. We might attribute it to the appreciable catalytic activity of plain nickel coupled with a possible partial de-activation of the catalytic nickel. In the absence of further experimental data, such speculation seems idle. One significant peculiarity of the critical potential method for exploring the catalyst surface should, however, be emphasised at this point. The electron bombardment method employed takes, as it were, a two-dimensional view of the catalyst, so that differences due to amount of surface as distinct from quality of surface are eliminated in the effects observed. This eliminates from our results one of the two factors governing catalytic activity as recorded by chemical observation.

The experiments, therefore, offer no conclusive hypothesis of catalytic action, but the presence of atomic hydrogen in the adsorption layer is probably of contributory significance in the light of such work as that of Taylor and Marshall.*

Summary.

Several criticisms are advanced which seem to render invalid Gauger’s measurements of the critical potentials of hydrogen in the presence of catalytic nickel.

A modified apparatus is described in which measurements of ionisation from hydrogen in the presence of catalytic nickel, plain nickel and catalytic copper were made. By comparison of these results with those obtained when molecular hydrogen was the preponderating source of ionisation, it is concluded that substantial quantities of atomic hydrogen are present at the metal surface in each of the three former cases. I t is shown that a similarity exists between the effect upon the form of the ionisation curve of the presence of the catalyst metal and that of an incandescent grid (as determined by Olmstead). The

2 KVOL. CX.— A.

* ‘ J. Phys. Chem.,’ vol. 29, p. 1140 (1925).



476 A. Fowler.

results are of significance in connection with the mechanism of catalyt^hydrogenation.

In conclusion, the author wishes to express his gratitude to Prof. K. T. Compton and Prof. H. S. Taylor for their constant encouragement, advice and

workers in Palmer Physical Laboratory, Princeton, whose friendly help was so generously given at all times.

By A. F owler, F.R.S., Yarrow Research Professor of the Royal Soch

“ compound line ” and “ elementary line ” spectra, assigned by Schus 1879* on the supposition that complex and simplified molecules or “ mol groupings ” were respectively involved in their production. The poss

Schuster, in collaboration with Roscoe,f in the observation of a line at ] when a condensed discharge was passed through oxygen in a short and n capillary tube. Additional lines of this third type were observed lat Lunt,J and further investigated by Fowler and Brooksbank.§

In accordance with present views as to the origin of spectra, the three s; are attributed to neutral, singly-ionised, and doubly-ionised atoms, ar designated 0 I, 0 II, 0 III, or 0, 0 +, 0 ++. Other spectra repres< higher degrees of ionisation are theoretically possible, and evidence of th duction of 0 IY and 0 V in the spectra of vacuum sparks has been obi * * * §

assistance. He also welcomes the opportunity to thank all the research

The Spectrum o f Ionised Oxygen ( I I ) .

Imperial College, South Kensington.

(Received December 23, 1925.)

Two different line spectra of oxygen have long been known under the :

of a further modification of the line spectrum was also foreshadows

* 6 Phil. Trans.,’ vol. 170, p. 41 (1879).f c Mem. Phil. Soe. Manchester,’ 3rd Series, vol. 7, p. 82 (1880).J c Annals of the Cape Observatory,’ vol. 10, pt. II, p. 26b (1906).§ 6 Monthly Notices R.A.S.,’ vol. 77, p. 511 (1917).



Documents

Critical potentials of hydrogen in the presence of …rspa.royalsocietypublishing.org/content/royprsa/110/754/...Critical Potentials of Hydrogen in the Presence of Catalytic Nickel