Activation of Molecular Hydrogen by Electron Impact. Velocity Distribution of Electrons Issuing from Small Holes. Activation of Gases by Electron Impact Citation Dalton, Robert Hennah (1928) Activation of Molecular Hydrogen by Electron Impact. Velocity Distribution of Electrons Issuing from Small Holes. Activation of Gases by Electron Impact. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/1K4P-YQ61. https://resolver.caltech.edu/CaltechETD:etd-02242005-091511 Abstract No abstract. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: Chemistry Degree Grantor: California Institute of Technology Division: Chemistry and Chemical Engineering Major Option: Chemistry Thesis Availability: Public (worldwide access) Research Advisor(s): Noyes, Arthur A. Thesis Committee: Unknown, Unknown Defense Date: 1 January 1928 Other Numbering System: Other Numbering System Name Other Numbering System ID Gates Chemical Laboratory 114 Record Number: CaltechETD:etd-02242005-091511 Persistent URL: https://resolver.caltech.edu/CaltechETD:etd-02242005-091511 DOI: 10.7907/1K4P-YQ61 Related URLs: URL URL Type Description https://doi.org/10.1021/ja01400a009 DOI Article published as ch. 1 https://doi.org/10.1103/PhysRev.29.248 DOI Article published as ch. 2 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 730 Collection: CaltechTHESIS Deposited By: Imported from ETD-db Deposited On: 24 Feb 2005 Last Modified: 03 Oct 2019 23:01 Thesis Files Preview PDF - Final Version See Usage Policy. 1MB Repository Staff Only: item control page

AC'.I'IVATION OF MOLECULAR HYDROGEN BY ELECTRON Th!PACT VELOCITY DISTRIBUTION OF ELECTRONS ISSUING FROM SM.ALL ROLES ACTIVATION OF GASES BY ELECTRON IMPACT Thesis by Robert H. Dalton In partial fulfillment of the requirements for the degree of Doctor of Philosophy California Institute of technology Pasadena, California l92S l. [Reprint from the Journal of the American Chemical Society, 49, 58 (1927).] (CoNTRIBUTION FROM THlt GATICS CHF;MICAI. L\BORATORY, CAI.IFORNIA lNSTITUTit OF TlccHNowcY, No. 114) THE ACTIVATION OF MOLECULAR HYDROGEN BY ELECTRON IMPACT BY GEORGl:t Gt.oCKLBR, WARRitN P. B.'\XTJ:tR AND RoBitRT H. DAI.TON RBCJIIVJID SliPTBYBBR. 30, 1926 PtlBUSBBD JANUARY 11, 1927 lnjroduction In this research we have undertaken to study the activation of hydrogen molecules by electron impact under conditions in which we know the energies of the impinging electrons. The experill).ents of Carlo and Franck1 show that hydrogen molecules can be activated by excited mercury atoms by collisions of the second kind and that copper oxide and tungstic oxide can then be reduced. In their experiments. the mercury atoms receive energy of 4.9 volts from the light source, which is sufficient to dissociate hydrogen molecules, the heat of dissociation of hydrogen being 3 to 4 volts. Furthermore, it has been known for some time that in a discharge tube hydrogen will disappear when a discharge is passed. 2 Hughes, 3 in particular, has investigated the electrical clean-up of hydrogen and nitrogen, and finds a definite decrease in the hydrogen pressure at 13.3 volts and higher. He adopts Langmuir's conclusion,4 that hydrogen is dissociated under these conditions, and that the decrease in pressure is due to the freezing out of atomic hydrogen on surfaces cooled by liquid air. His results will be referred to later in connection with our oym experiments. In Hughes' investigations no copper oxide was present, and the minimum electron energy at which hydrogen disappears was not accurately determined. . A number of experimenters5 also have investigated the chemical reactivity of hydrogen activated by an electric discharge, but the energies of the impinging electrons were not known. We shall discuss four possible mechanisms by which electrons may be expected to activate hydrogen molecules. First, it might be that an electron having kinetic energy of 3 to 4 volts could transfer its energy to the hydrogen molecule and cause its dissociation into atoms. These in turn could then react with other substances. However, it is known that no kink occurs in current-potential curves of hydrogen near 4 volts, and it seems, therefore, that electrons having kinetic energy equal to the dissociation energy of hydrogen molecules 1 Carlo and Franck, Z. Physik, 11 1 161 (1922). An excellent discussion of these phenomena may be found in Dushman "High Vacuum," Gen. Electric Rev., 1922. · a Hughes, Phil. Mag., 41, 778 (1921). 'Langmuir, THIS JOURNAL, 34, 1310 (1912); 371 417 (1915). 6 Wendt and Landauer, ibid., 42 930 (1920); 44, 510 (1922). Bonhoeffer, Z. physik. 1 Chem., 113, 199 (1924); and others. 2 2. Jan., 1927 ACTIVATION OF HYDROGEN 59 cannot transfer their energy to these molecules. We then should expect to find 110 evidence of reaction when hydrogen is bombarded with 4~volt electrons, and our experiments· actually do give no indication of reaction. This is in agreement with the commonly held idea that the dissociation of molecules does not occur as the direct result of electron impact. Second, electrons may have to possess sufficient energy to resonate the molecule, which may then dissociate if its heat of dissociation is less than its resonance potential. The hydrogen molecule has, according to the latest results of spectroscopy, 0 a resonance potential at 11.6 volts. Electrons of this energy can raise hydrogen molecules into an upper quantum state. These activated molecules would ordinarily return to the normal state after a short time. However, they may either dissociate into atoms upon impact with other molecules and the atoms thus react, or the resonated molecules may act directly on any oxide or other substance on which they impinge. A third mechanism of causing activation by electron impact might be that where the impinging electron transfers enough energy to the hydrogen molecule to cause its dissociation and resonance of one of the atoms. This process may be expected at 13 to 14 volts, which is the sum of the heat of dissociation and the resonance potential of the hydrogen atom. This is the mechanism postulated by Hughes. 3 A fourth possibility may be considered. It may be necessary that an electron ionize a hydrogen molecule before the latter can be made to react. It was found by Anderson 7 and Storch and Olson 8 that nitrogen and hydrogen reacted to form ammonia when bombarded by 17-volt electrons. This voltage is near the ionization potentials of these molecules. We may state at once the results of our experiments. Electrons of 11.4 volts' energy can activate hydrogen molecules, for we find that there is a definite pressure decrease when the accelerating. voltage applied to our tube has this value. At the same voltage we also obtain a kink in the current-potential curves, using the Franck method. This research was aided financially by a grant made to Professor A. A. Noyes by the Carnegie Institution of Washington. Experimental Procedure The experimental arrangement is shown in Fig. 1. The filament F was made of platinum foil coated with calcium oxide. ThiS was surrounded with a platinum can having in the end a circular grid Gi, 0.5 cm. in diameter made from platinum gauze of 52 mesh. The next electrode consisted of a copper cylinder B, 4.5 cm. in diameter with a platinum grid ~. also of 52 mesh at the end. One millimeter behind this second grid was a platinum plate P. The distance between the 1 Witmer, Proc. Nat. Acad. Sci., 12, 238 (1926). 7 Anderson, Z. Physik, 10, 54 (1922J. s Storch and Olson, Tms JOURNAJ., 45, 1605 (1923). 3. 60 GLOCKU~R, BAXTICR AND DAI.TON Vol. 49 filam~nt and first grid was 1 mm., and between the two grids 4 cm. A copper cooling coil C, through which tap water cotild be circtilated, was introduced into the tube by means of copper disk seals, and made good electrical and thermal contact with the copper cylinder. The copper cylinder and cooling coil were surface oxidized in a Bunsen flame before being placed in the tube, thus providing a large surface of copper oxide. With the tube first used, which had no cooling coil, jt was found that with the filament lighted, but with no accelerating field on, there was a noticeable rate of decrease or pressure. Since this effect may have been due to reduction by the hydrogen of the copper oxide hl!l!-ted by conduction from the hot filament, the cooling coil described above was introduced into the tube, with the restilt that there was no longer any decrease in pressure, but instead a smaller rate of pressure increase. This increasing pressure may possibly be due to slow evolution of gas from the heated filament or from the metal surfaces bombarded by electrons. It was· too large to be accounted for by leaks, which were careftilly tested for. These zero rates, however, caused no diffictilty in determining the voltage at which reaction did begin. Fig. !.~Experimental tube and wiring diagram. The electrical connections are as shown. The total current, henceforth designated by 11, and the current to the second can, 12, were measured on the same galvanometer with a scale 1 meter long and 3 meters away. On the most sensitive shunt 8.16 X 10-~ amperes gave a deflection of 1 cm. Another galvanometer, giving a deflection of 1 cm, for .1.09 X 10-11 amperes on its most sensitive shunt, was used to measure the plate current Ia. The voltage applied between the filament and grid G1 was calctilated froni the equation Ri E1 = Ri + R2 [E - I1&] E is the voltage across the potentiometer, and R 1 and R2 are the respective readings of the two resistance boxes (the sum of whose resistances was kept at 10,000 ohms). The term I1R2 is introduced because of the electron current flowing from the filament to the other electrodes, hence through R2 and back to the filament, and is relatively important at the emissions used (O.l to 1 milliampere). One side tube led to a liquid-air trap and mercury cut-off, thence to a palla:dium tube for admitting hydrogen, a McLeod gage, another mercury cut-off, and pumps. Another side tube led to a.Pirani resistance gage which consisted of a thin tungsten wire 4. Jan., 1927 ACTIVATION OF HYDROGEN ' 61 9 cm. long enclosed in a narrow glass tube. It wa:s used according to the method described by Campbell,9 and showtt in the figure. By putting the gage in one arm of a wiieatstone .bridge, the other three resistances of which were fixed and each one-third greater than the Pirani resistance at room temperature, any change in pressure could be followed by means of the voltage necessary to obtain a balance. The voltage was calibrated against the.McLeod gage for a series of pressures of hydrogen up to 0.3 mm. of mercury. The Pirani gage was shielded from drafts, but no attempt was made to thermostat it. In order to follow small pressure changes, a more sensitive galvanometer was thrown aeross the bridge after a balance had been obtained, and readings were made with a telescope and scale 80 cm. away. With this set-up 1 cm. change in the scale reading corresponded to approximately 4 X 10- 5 mm. in a total pressure of hydrogen of 0.2 mm. The volume of the tubC up to the mercury cut-off was about 320 cc. It will be seen that with this apparatus the following objectives can be attained. (a) Electrons emitted from the heated .filament can be accelerated to a known velocity by the first grid and activate hydrogen molecules, which may then react with the copper oxide. (b) The water formed can be frozen out in the liquid-air trap with a resulting decrease in pressure. (c) The rate of decrease of pressure can be followed by means of the Pirani gage. (d) Mercury vapor is frozen out by the liquid air; thus, resonated mercury atoms cannot excite hydrogen molecules and cause reaction. (e) Current-potential curves can be taken by either the Franck or Lenard methods in the usual manner. 10 The procedure adopted was as follows. After pumping out to io- 5 mm.; hydrogen was admitted by heating the palladium tube in a gas flame; and readings were taken on the McLeod gage and on the voltmeter of the Pirani gage after liquid air had been placed about the liquid-air trap. The filament was then lighted and its temperature adjusted to give the emission desired at a definite accelerating voltage. After conditions had become steady, the mercitry cut-off was closed and a series of readings with no accelerating field wa.S taken on the rate of deflection of the sensitive galvanometer in the Piratii set-up. ·These readings constituted a set of "zero rates" for that particular run. Then the electrons were accelerated with a definite potential, and a siniilar set of rate measurements was taken. From time to time the mercury cut-off was opened and closed, allowing the hydrogen to come up to its original pressure. The voltage was increased in steps of 0.4-0.5 v. This process was repeated until rates had been obtained covering the voltage range desired. At the same time readings were taken of the total current Ii. and of Iz and Ia. No field was applied between the two grids, but a retarding potential of 0.5 v. was maintained between the second grid and plate. In these runs. either the total current Ii or the sum of the currents 12 I3 was kept constan,t by adjusting the filament temperature. + 9 Campbell, Proc. Pkys. Soc. London, 33, 287 (1921). These methods are fully described in "Critical Potentiais," Compton and Mohler, Bull. Nat. Res. Council, 1924,.and elsewhere. 16 5. 62 VoL 49 GWCnsR, BAXTER ANP D.AI,TON Experunental Results Reaction Measurements.~Ca:refuJ investigation failed to give any evidence of reaction between 0 and 10 volts' accelerating potential. It was found, however, that reaction does begin around 11.4 volts. From 11.4 volts on, the rate of pressure decrease rises rapidly with the applied . voltage until the ionization potential is just passed and the glow discharge sets in. Above ionization it was found in most of the runs that tlie rates increase onJy slowly with the voltage, but depend primarily on the number of electrons passing through the first grid. Since electrons are not emitted from a heated platinum-calcium oxide surface until a certain accelerating voltage is applied, a corresponding correction must be applied to the voltage ,scale. This initial velocity correction was determined as follows. For a constant heating current, the accelerating voltage was increased in small steps from 0 to 5 v., and the total current 11 read. The foot of the curve so obtained correspond~ to the voltage at which the fastest electrons will just escape from the filament and gives the correction to be subtracted from the observed voltage at which reaction begins. Four determinations for different heating currents gave 1.12, 1.19, 1.20 and 1.21, average 1.2 v., In Table I are given the voltages (from which the correction of 1.2 v. has been subtracted) at which reaction is found to begin. The average of these values is 11.4 v. The table contairts all our experiments. TABI.lt I Vo1.TAG:it (CoRuci'~n) AT WH:rca ~CTION STARTS Expt. 1 2 3 4 5 6 Pressure of Ht, mm. of Hg Volts Expt. 0.262 .249 .168 .124 11.8 11.9 11.8 11.4 7 8 .103 11.9 .158 11.5 9 14 15 18 Pressure of Hi, mm.of Hg 0.203. .244 .175 .162 .193 .185 Volts 10.9 11.2 12.0 10.9 10.4 11.5 Rescmance Pote11.tial of Hydrogen.-In order to determine the significance of the value 11.4 v., Franck current-potential curves were .taken in hydrogen and in hydrogen-helium11 mixtures of the composition 0.1 mm. of hydrogen and 1.1 mm. of helium. In making these :tuns, 1~ was kept constant and readings taken of 13 • A retarding potential of 0.5 v. was applied between the plate and second grid, while the accelerating potential was increased in small steps. Prominent kinks occurred in the 13 curves taken in pure hydrogen around 13.5 and 19 v. uncorrected. In the curves for the hydrogen-helium mixtures similar kinks were' found, u The helium used was that prepared previously. (1926). Gloclder, ['hys, Re'IJ., 271 423 6. Jan., 1927 63 AC'l'IVATION OF HYDROG:ltN as also a new one at22 voltS, due, to the :first;resonam;e pptential of helium. By plotting the differences of successive Ia r_eadings the kinks were.brought out very. strongly. The voltages at which the kinks occurred were obtained from the maxima in these difference curves, which give the point , at which the rate of change of slope is the greatest, thus corresponding to the middle of the electron velocity distribution. Two determinations by the Franck method gave the helium kink at 21.88 and 21.75 v., and a . Lenard run, 21.90. Froni spectroscopy the value of the resonance potential of helium is 19.77 v., which has also been obtained by the method of electron impact. Therefore, by subtracting 19.77 from the above values the initial veloeity correction can be obtained. This comes out 2.11, 1.9$ and 2.13, average 2.08 v. Table II summarizes the values found for the first kink in hydrogen after the correction of -2.08 v. has been applied. It is seen that the two methods used to determine the initial velocity correction give identical results for the voltage at which reaction starts and for the resonance potential of hydrogen. TABL~II ~Sol'i'AN~ Po'l'~N'l'IAL oli' HYDROG~ Expt. No. 10 11 12 13 14 15 16 17 H:ipressureinmm. 0.128 0.28 0.16 0.162 0.162 0.175 0.1 4 0.1 4 Volts, corrected 11.28 10.93 11.27 12.09 11.58 11.73 10.83 11.33 Volts, averaged 11.38 a Plus 1.1 mm. of He. This experimental value of 11.4 v. agrees well with the value 11.6 de~ rived by Witm.er 6 from spectroscopic data for the first resonance potential of the hydrogen molecule. It is also seen to coincide with the voltage at which hydrogen begins to disappear and we may therefore conclude that the first step in the reaction at these voltages is the formation of resonated hydrogen molecules by collision with 11.4 electrons. The resonated hydrogen molecule may the11 react directly with the copper oxide or be dissociated by collision with other molecules and subsequently react. Of course, after the first resonance potential is passed, the process of activation may include higher energy levels, and when ionization is reached the mechanism of reaction may be considerably altered. · Further work on these points is. to be continued; we wish only to record here the fact that reaction does set in when the first resonance potential of the hydrogen molecule is reached. · In order to identify still further the beginning of reaction with the resonance potential,. several runs were made in which both rates and Franck current-potential curves were taken. Fig. 2 illustrates two of these runs. In each case, it was observed that the break in the Franck curve begins at the same poi1!-t as that at u11f,ich t/ie.hydroglm, pressunrs.tarts todrop. Vol. 4Q . ~--· Identification .of: the Rea:ction Product · . In ord~f 'to determine the niitu~e of the final reaction product, one of the riliis was ~o~tinued until an' of the hydrogen had reacted (pressure about 10-:-4 n11h:). ,on removing the iiqµid air the pressure rapidly rose. After 30 rllinutes · the liquid· air was replaced. The pressure immediately fell to jts original value. . .This is the opposite of the results of Hughes3 (in the absence of copper oxide) for he fotind that on removing the liquid air Expt. 18.--'High emission 0.185 mm. of H2. Expt. 14.-Low emission 0.162 mm. of H2. 2.0 1.7 8 10 12 14 16 10 12 14 volts -corrected. Fig, ·2.~Two runs illustrating the relation between the beginning of the break .in the Franck current-potential curve and the voltage at which reaction stii.rts. · · and replacing it the pressure did not fall to its original value due to the presence of a non-recondensable gas. The non-recondensable gas he assumes, with Langmtiir; 4 to be molecular hydrogen formed by the recombination of atomic hydrogen that bad previously been frozen out on the walls at liquid.:air temp.erature. To determine whether or not water was forn1ed in our experiment· the liquid air was again removed, the mercury cut-out opened, and. the presstire r~ad on the McLeod gage. The two s. Jan., ·ig27 ACTIVATION OF HYDR0GEN 65 columns maintained a nearly constant difference of height except near the closed end of the capillary. After correcting for the small pressure of uncondensable gases, the difference of height was found to be 2.6 cm. The vapor pressure of water at the same temperature is 2.65 cm. Summary Hydrogen molecules have been activated by collisions with electrons of known energy and made to react with copper oxide. It is found that reaction begins at 11.4 volts. At the same time Franck current-potential curves show a resonance point at 11.4 volts. This agrees with the recent value of 11.6 volts for the first resonance potential of the hydrogen molecule as obtained from spectroscopy. It is concluded, therefore, that the first step in the activation process at low voltages is the formation of a · resonated hydrogen molecule. No evidence is found that 3- to 5-volt electrons dissociate hydrogen molecules. PASADnNA, CALIFORNIA 9. FEBRUARY,' 1927 PHYSICAL REVIEW VOLUME29 THE VELOCITY DISTRIBUTION OF ELECTRONS ISSUING FROM SMALL HOLES BY ROBERT H. DALTON AND WARREN P. BAXTER ABSTRACT The ve!Ocity distribution of a beam of 50-volt electrons issuing from a hoie 0.022 cm in diameter in a copper plate 0.02 cm thick has been measured. Seventy percent of the electrons were found to retain approximately their initial velocity. By coating the sides and edg~s of the hole with lampblack 95 % of the electrons were transmitted without appreciable energy loss. Similar results were obtained with grids of 100 mesh copper gauze. JNTRODT,JCTION.-Preliminary to the construction of a magnetic electron velocity filter, some experiments were undertaken to determine the effect of passage through small holes or fine-mesh gauzes on the velocity distribution of an electron beam. Lehman and Osgood 1 found that with velocities of the primary beam corresponding to from 200 to 1000 volts less than 1% of the electrons passing through a hole 0.19 mm in diameter in a copper plate 0.135 mm thi.ck retain their initial velocity. Since in similar experiments with SO-volt electrons we were able by using non-reflecting surfaces to get through more than 90% of the electrons with their initial velocity, it seems worth while to publish a brief description of our experiments. ·Experimental arrangement. The apparatus is shown diagrammatically in Fig. 1. The source of electrons is a platinum filament F coated with calcium oxide operating on a potential-drop of 2 volts. Surrounding this is the cylindrical copper cage Ci, pierced by a hole 0.022 cm in diameter at a point opposite the center of the filament. The copper in 1 Lehman and Osgood, Proc. Cambridge Phil. Soc. 22, 731 248 (1925). 10. VELOCITY DISTRIBUTION OF ELECTRONS 249 the. neighborhood of the hole is 0.02 cm thick, and the distance to the filament is about 0.1 cm. C2 is a similar cage 2 cm long with a hole 0.5 cm in dfametet in the end facing C1. The ends of the cages are discs which screw into place so that they n:iay readily be replaced by grids, in which case the copper plate Pis used to measure the current traversing the second of these. It is about 0.1 cm from C2. All the cages have an inner diameter of 3 cm. The cages, plate, and filament are mounted rigidly. with respect to one another and are suspended with the axis of the cylinders parallel to the earth's magnetic field in a tube closed by a ground-glass stopper (set in sealing wax) and connected with a liquid-air trap, a McLeod gauge, and vacuum pumps. Gi. G2, and Ga are galvanometers measuring the total current, the current from C2, and the plate current, respectively. A. c 50 Accalaratin§ O Volts "Ratardin_1 50 Fig. 2. The general procedure was as follows. The apparatus was pumped out to a pressure of 10-4 to 10-5 mm, and the pumps were kept running during all the measurements. An accelerating potential of 50 volts was established between C1 and the filament, and the temperature of the latter adjusted till a convenient emission resulted. The current through G2 was then measured with various retarding potentials .between C1 and C2. When the end of C2 was replaced by a grid, the plate-current was read on Ga with various retarding potentials between C2 and the plate. . Experimental results. The experimental curves are reproduced in Fig. 2. Curve A(1) was obtained with the hole 0.022 cm in diameter. The metal parts were baked out in vacuum at about 400° before 11. 250 ROBERT II. DALTON AND WARRRN P. BAXTER placing in the apparatus, but no attempt was made to reduce the surface layer of copper oxide. The current to cage C2 is plotted against the retarding or accelerating potential between C1 and C2. It is apparent from the curve that about 70% of the electrons (based on the current at zero volts) retain nearly their original velocity. To determine whether low velocity reflection and secondary _emission were affecting the distribution, the end of C1 was now coated with lampblack in a candle flame, since Gehrts 2 has shown that a lampblacked surface is a good absorber for electrons. Curve A (2) shows the result obtained using the same total emission as for curve A(1). About 95% of the electrons have their initial velocity. It is also interesting to note that the actual number of electrons retaining their initial velocity is nearly the same in the two cases, so apparently the only action of the soot is to remove from the beam all those electrons which strike even a glancing blow, thus preventing reflection and secondary emission, and eliminating the low-energy group previously present. An<?ther distribution curve was taken after sooting C1 and C2 completely; but the curve was practically the same as A(2), showing that C2 was an efficient Faraday cage. These are much better distributions than were obtained by Lehman and Osgood in the experiments above cited. The difference may be due to the higher voltage at which they worked, since it is known that the reflection of electrons from copper increases with the voltage up to 200-300 volts. The nature of the surface as determined by the previous treatment of the copper would also affect the reflection. In some. earlier experiments where a crude apparatus was used, in which the metal parts were not baked out, we obtained curves somewhat similar to theirs, but after sooting the copper surfaces the results were as good as with the apparatus described above. Since, in most of the experiments in electron impact work, the electrons traverse wire gauze grids, an experiment was tried to determine the effect of lampblack under these conditions. The hole in C1 was replaced by a circular grid of 100 mesh copper gauze 0.5 cm in diameter, while the end of C2 opposite the plate was replaced by a similar grid 2.5 cm in diameter. Curve B(1) is a distribution curve of the current reaching C2 in the absence of lampblack. In this case only about 60% of the electrons retained approximately their initial velocity, while the majority of these have suffered a small energy decrease. Curve B(2) shows the distribution after the grids and cages had been 2 Gehrts, Ann. d. Physik. 36, 995 (1911). 12. VEWCITY DISTRIBUTION OF ELECTRONS 251 · sooted. It 1s apparent that about 90% of the electrons have their initial velocity. The absolute magnitudes have no meaning, since the emission was much higher in the second case. To demonstrate further the effect of carbon, distribution curves of the plate current were taken first without and then with lampblack. These are reproduced in curves C(l) and C(2), respectively, in which the plate current is plotted against the accelerating or retarding voltage between C2 and P. C(l) is similar to curves obtained by Lawrence3 with a very short Faraday cage. C(2) shows a much better distribution, but it is poor compared to the curves above, which probably indicates that carbon is not a perfect absorbent. C(l) is plotted to a much smaller current scale than C(2), since the form of the curve and not the magnitude of the current is important .. The authors wish to express their thanks to Dr. G. Glockler for his help and advice; also to the Carnegie Institution of Washington for financial assistance received from a grant made to Prof. A. A. Noyes. GATES CHEMICAL LABORATORY, CALIFORNIA INSTITUTE OF TECHNOLOGY, PASADENA, CALIFORNIA, October 29, 1926. 1 Lawrence, Proc. Nat. Acad. Sci. 12, 29 (1926), 13. THE ACTIVATION OF GASES BY ELECTRON IMPACT INT:OODUCTION ·.me purpose of this research was to study the chemical behavior of gases activated by electron impact and to measure the energy necessary for their activation. The general plan was to use an appa- ratus similar to that employed in the study of the reaction between copper oxide and active hydrogen by Gleckler, Baxter, and Dal ton~ The gases were introduced at pressures of 0.05 - O. 3 mm. of mercury into a. four electrode tube connected with a liquid air trap and a ~irani pressure gauge. A four electrode tube was used in order that the energy loss of the electrons might be studied by making current potential curves 2 as done by Franck a.nd others. With this arrangement it was possible to follow the course of any reaction which involved a change in the n"Um.ber of molecules or produced a product condensible in liquid air. It waa of course not possible to use gases which would undergo reaction when ex... posed to the hot filament. The systems studied were: carbon monoxide and hydrogen, nitrogen and copper. oxygen a.nd oa.rbon, Only the first gave definite positive results a.nd for this reason it will receive the n:ost attention. The author wiehes to express his thanks to Professor Tolman, Dr. Gleckler and Mr. Baxter for their assistance in this research; also to the Carnegie Institution of Washington for financial aid received through a grant made to Professor A.A.Noyes. 14. ACTIVATION OF OXYGEN 'Exper!Eiental Procedure: - The oxygen was obtained by heating mercuric oxide some of Which was contained in a bulb attached to the apparatus. The mercuric oxide had been previously prepared in the ap- paratus by boiling rr.ercury for some time with well dried tank oxygen • .Except in Experiment 2 (see Table I) the carbon was present in the form of lamp black deposited on grid, G2 , with a smoky gas-cqgen flame. In the case of Experiment 2 graphite which had. been previously ignited in vacuo was used. The apparatus used is shown diagramatically in Figure I. Within the tube all metal parts are of platinun:: except for the short length of tungsten wire used to bring the leads through the glass. The source of electrons is F, operating on a potential drop of about 2 volts. Around this is a cylindrical electrode, G,, with a circular grid. of 50 mesh gauze 0.1 cm. from the filament. The next electrode is also cJrlindrical and has in the end a circular grid, G2 , of the same mesh as G,. The dista.~ce from G1 to G2 is about 4 cm. One millimeter behind G2 is the plate, P. The electrical connectio~s are shown in the diagrem, I,, I 2 , I 3 and !~ are galvanometers. The potential between filament and G1 is established by means of a storage batter;y and resistance in parallel and is calculated by the expression P. D. = RR, + R2 (E1 - I1R2) 1 where R 1 and R2 are the resistances indicated in the diagram, E, is H I IL.J I I- - - Il---1vWNM 16. the potential of the battery and I 1 is the current between the filament and the first electrode. A connecting tube leads from the apparatus to the Pirani pressure gauge, A, which consists of a fine platinum wire about lS cm. long and 0.003 cm. in diameter. In the original experiments the wire was sealed in a straight glass tube O. 5 cm. in diameter. This form while quite sensitive was undesirable since when heated the wire became slack so that its position with respect to the walls and hence the heat conductivity and apparent pressure were variable. These dif- ficulties were avoided by the arrangement shown in Figure I in which the wire is stretched and centered by a small "Weight sliding in a glass tube. To maintain the walls at constant temperature the tube containing the wire was placed in a thermos bottle filled with water. The slight heating due to the Pirani wire did not raise the temperature of the water appreciably. The connections a.re shown in the figure and are those recommended by Campbell. 2 The Pirani wire is in one arm of a. Wheatstone bridge the other three resistances of which are fixed. The voltage, V~ necessary to balance the bridge is then a measure of the pressure. Over the pressure ranges for which the instrument was used (o. 01 -to O. 3 mm.) the pressure is a nearly linear function of V2 - V0 2 where Vo is the voltage to balance the bridge when the pressure is negligible. If absolute pressure measurements are desired the gauge mu.st be calibrated against a. BcLeod gauge with each gas used. .Another connecting tube leads from the apparatus to a. liquid air traj:>, a mercury cut-off 1 devices for introducing various gases, a McLeod gauge, another mercury cut-off 1 and the pumps. To prevent changes in pressure due to changes in temperature the electron tube was surrounded by a water jacket through which a stream of ·tap water was passed. The experimental procedure was in general the following. The apparatus was pumped out to about lQ':'.'5 mm. 1 the filament lighted, and the trap at the tube immersed in liquid air. The gas was introduced, pressures being read both on the McLeod and on the Pirani gauge. The mercury cut-off at the tube was then closed and the rate of pressure change determined with various accelerating voltages between the filament and first grid. These rates were followed by measuring the rate of deflection of the galvanometer I~ When the bridge was balanced. Us- ing a sensitive galvanometer changes of pressure of io-5 mrn. caused deflections of 1 cm. In the ·work on oxygen a great deal of difficulty was encountered in obtaining sufficient electron emission to produce measurable chemical effects. In some cases the emission of a filament was less than one hundredth of its value in the presence of hydrogen gas. Experimental Results: - The results of the experiments with oxygen were as follows. It was found that with the filament heated the pressure decreased steadily, even although no accelerating potential was applied. .Aiteiq;>ts to remove this zero rate by running for some time with high oxygen pressure and by cooling the tube failed, so it was decided to' correct the other rate readings by subtraction. This was not difficult since the zero rate remained nearly constant d.ruing the course of the run. This initial rate was, however, of the same magni- tude as the rate due to electronic excitation. From 0 up to 8 volt• there wer~ no important changes in the rate. Most experiments to be sure showed some increase between Z llil.d 5 volts but these effects were small compared to the changes above 8 volts and were not definitely reproducible. Around 9 volts in all the experiments (see Table I) the rate began to increase with the voltage, rapidly at first and then more slowly up to the highest potentials investigated. of the runs. Figure II shows two The voltages are plotted as abscissae and the rates of pressure decrease in arbitrary units as ordinates. . 6 vo ,./ .," '. "'" / c \I ---u f\_ Q 1......-- "" Vo ~ v - I ~ Ct . r, _/ ct 4 ' a J ....." ~ 0 I~ ~ s ~ / JD D- ~ I/ LP" 10 20 Volt& FIGURE II 20. The numerical results of the experiments in oxygen gas are summarized in Table I. TABLE I Pressure of 0 2 Experiment Number mm l 2 0.09 '0.09 ~ 0.09 0.09 7g 0.24 0.22 0.23 0.12 g Corrected critical voltage s.7 10.l 10.S 9.2 s.5 0.09 9 9.3 9.2 9.2 8.6 Weigh tad mean 9.2 volts This table contains all the experiments except one in which w~re the ":S" batteries 'eeeeme discharged. Experiments 6 to 9 are the most accurate and have been given twice the weight of the others in determining the mean. since they were made with the imprned Pi rani and with the tube water-cooled. Also in the earlier runs the voltage of the "B" battery was not always accurately known. The values of the corrected critical voltages were obtained as follows. The difference between rate at a given voltage and the zero rate was divided by the total emission to compensate for the increase of this with increasing voltages. The values so obtained were plotted against the corrected voltage. The voltage applied to the tube is not a true measure of the energy of the electrons since some energy is required to remove the electrons from the filament. This initial 21. velocity correction was determined by teking the voltage at the foot of the cu.rVe ·obtained by plotting the values of the total emission against the voltage in the region from 0 to 3 volts. The average of three de- terminations gave O. 85 vol ts for this correction•. Subtracting this value from the applied potential gives for practical purposes the maxi- mum voltage of the electrons and is spoken of as the corrected voltage. The maximum value is of interest rather t": ;;m the average, which is about a volt lower,since the critical voltage is taken a.a the point where the rate first starts to change, which is the point at which the fastest electrons just attain the c~itical value. In Figure II the rate of pressure decrease in arbitrary units is plotted against the corrected voltage. ments 4 and 6. The data are those of Experi- In curve 6 the foot of the curve is very sharp and the potentie.l at which reaction starts can be e'.1sily determined. In case 4 where this is not true the foot was determined by extending the rapidly rising nearly- linear part of the curve till it intersected the voltage axis. The Reaction Product: - It is natural to suppose that the product of the reaction is carbon dioxide, since carbon monoxide will not be frozen out by liquid air at these pressures. ments substantiated this supposition. The following experi- The pro J.uct from a. number of runs was allowed to accumulate in the liquid air trap, the residual oxygen was pumped out, the mercury cut-off a.t the pumps closed, and the liquid air removed. A series of pressure readings was then made on the McLeod 22. gauge with the mercury at different heig~ta. The product behaved as a perfect gas up to pressures of 18 cm. a.r.d so cannot have been water vapor. There was no permanent change in pressure on lighting the filament Which eliminates ozone as a possibility. Finally some of the product was collected in a small side tube, cooled in li~uid air, sealed off, and opened under barium hydroxide solution. A small quantity of a. white precipitate of barium carbonate was obtained.. Thus the product must consist at least in part of carbon dioxide. Discussion: - It is interesting to cornpare the value 9.2 volts at which reaction starts with the known energy levels of oxygen. band spectra of oxygen have been only partially w:>rked out. electronic 1 evels at about l. 7 vol ts and at 6 vol ts. The Tb.ere a.re 4 It is thought that the convergence limit of the oscillational states of the 6 volt electronic level corresponds to the energy for dissociation into two normal atoms. This gives about 7 volts for the work of dissociation of the normal molecule. Above 6 volte there are no levels given till ionization into 0 2 + between 14 and 15 volts. ·I'hue spectroscopically there are no energy levels yet given in the neighborhood of 9 volts thoU;gh it is probable that such exist. The work of various investigators on the determination of the critical potentials of oxygen by electron impact methods is not in good. agreement. For the first critical potential, values varying from 7.9 I:; to 9.2 volts have been reported.~ Bowever the later work which is probably the more reliable gives values close to 8 volts. A second critical potential is found by most observers at about 15.5 volts and is accepted as the ionization potential.6 It is seen that except for some early, rather unreliable work none 'of the above levels a.re in agreement with the potential a.t which reaction was found to start. To be sure this potential is only l volt greater than the best value of the first resonance potential as determined by electron impact experiments, but it seems improbable that the error .could be that large. Furthermore i f we accept the spectrosco- pie value of 7 volts as the energy necessary for dissociation into nor- mal atoms, the process cannot consist of dissociation into excited atoms since there a.re no suitable atomic levels to take up the 2 volt excess, the first resonance of the oxygen atom being -9.1 volts. If the 2 volt excess could go into kinetic energy we might have dissociation as the primary process. However, it is most probable that we are dealing here with a previously unrecorded critical potential of the molecule. Work is now being started on current potential curves in oxygen and this will aid in interpreting the process. One mechanism may be advanced which will account for the critical reaction velocity of 9.1 volts. If, either by electron impact or by the glowing filament, oxygen is dissociated, the atomic oxygen so formed would be activated by 9.1 volt electrons. This process seems highly improbable since it assumes that a sufficient quantity of atomic P"' oxygen would be fcrmed and that it would not itself react at an appreciable rate. 24. CARBON MONOXIDE JU.ID HYDROGEN Marshal:l showed that hydrogen gas activated by collisions of the second kind with mercury in the 23 P 1 state reacted with carbon monoxide to :fbrm formaldehyde. It seemed probable foat activation of hydrogen or carbon monoxide by electron impact would lead to the same reaction, 'but this did not prove to be the case. The carbon monoxide was prepared by the action of formic acid on sulfuric acid. It was passed through potassium hydroxide containing some potassium pyrogallate to renove carbon dioxide and oxygen, and then through concentrated sulfuric acid and over phosphorus pentoxid.e for drying. The gas was admitted to the apparatus through stop cocks. The hydrogen was introduced by heating a palla.ditml tube at- tached to the apparatus to bright redness in a gas flame. The tube used was the same as that described previously in the work with oxygen. It was not water cooled and the old form of Pirani gauge was used. Be- fore commencing this work the tube was run at a high potential with hydrogen for some time to remove adsorbed oxygen. similar to that ueed with oxygen. The procedure was The gases were studied at pressures of about 0.05 mn. No reaction was found below the potential at which an a.re appeared (25 to 30 volte). At this potential carbon monoxide alone under- went a pressure decrease so that we can only say that reaction does not occur at the lower resonance potentials of either hydrogen or carbon monoxide. .. ACTIVE NITROGEN 'An attempt was made to produce chemically active nitrogen by electron :l.mpaet. It wae hoped that the value of the critical po- tential might shed some light on the question of the nature of active nitrogen. The object of the investigation was not accomplished since at the highest electron currents obtainable with the apparatus used, mto.tUi"able quantities of chemically active nitrogen were not produced at interpretable potentials. The results are, however, of sufficient interest to justify a brief mention of the experiments. It was first necessary to find a suitable reaction. For this ·purpose various substances were introduced into a tube connected with a Pirani gauge and containing nitrogen gas at about 1.2 nun. pressure. Ordinary cylinder nitrogen was used after drying over phosphorus pentoxide. It was admitted to the apparatus through stop cocks. e. gen was then activated by an electrod}ess discharge. The nitro- The production of active nitrogen was proved by the appearance of the afterglow. When copper foil was present under these conditions the pressure fell to about lo-3 mm. in a few minutes. There was no appreciable increase in pressure after standing several days. Thus active nitrogen forms with copper a stable compound and this furnished a very convenient reaction to study. The experimental arrangement was that used by Glockler, :Baxter, and Dalton3 except that the copper surfaces were cleaned in acid instead of being oxydized. Pressures of nitrogen from 0.1 to 0.3 mm. were used. There was practically no change in pressure, however, till the accelerating -potential reached the value at which a l'Ulllinous discharge s·et in, 26. which was at 20 to 30 volts deyending on the pressure. Thus-for example 1iJ'l an experiment in which the pressure of nitrogen was 0.17 mm. a run of 2 hours with 20 volts accelerating potential and 1 m.a. emission produced a change in pressure of only 5 x 10~3 mn. This is too small to If.i\fnit rate mea:surements of the kind we had been making>and might have been accounted for simply by temperature changes which occurred over that period of time. The fact that a pressure decrease started vihen a luminous discharge set in gives no information as to the energy of active nitrogen except that thil is less than 25 vol ts, since the arcing-potential depends on the pressure and geometry of the tube and also since a great many energy states are then present. 1 27. SUMMARY ·The results of the experiments on the h,.vdrogen-copper oxide reaction and on the effect of passage through small openings on the velocity distribution of an electron beam are sumnarized on pages 8 and 12 respectively• •Jn active form of oxygen capable of reacting with carbon to form carbon dioxide is produced when oxygen gas is subjected to bombardment with 9. 2 volt electrons. The primary process in the activati:'n is probably the excitation of an electronic level in the molecule. Mixture• of carbon monoxide and hydrogen when subjected to electron izn:pact failed to i:i.ndergo a measurable reaction till the arcing po tent ial was reached. It was found that active nitrogen reacts with copper to form a non-volatile, stable compound. An attempt was made to produce active nitrogen by electron impact using the formation of this compound as an indication of its presence. dence for its formation was found. Below the arcing potential no evi- 28. BEl''EREl~CES l. Thia Thesis, 3. 2. For a complete discussion of these method.a see Franck and Jordan, "Anregung von Q;Uantensprungen durch St8sse", ~pringer, Leipsig, (1927). 3. Campbell, Proc.Phys.Soc.J.. ~mdon, J}, 287 (1921). 4. Birge; Bulletin of the National Rssearch Council, go. 5. Foote and 1'.•bhler, Bureau of Stanr.e.rds Publicati<::n, No. 400 (l~.20). Boucher, Phys.Rev.,~' 202 (1922). Franck and Hertz, Verb. d. r. Phys.Gee. I 15, 34 (1913). Hughes and Dixon, Phys. Rev., 10, 502 (1~17). 6. Sm;vth, H.D., Proc.Roy.Soc.L0ndon, J.: 105, 116 (1924). 7. Marshall, J.?hys.Cbem., JQ., 1634 (1926). 57, 244 (1926).