The deposition of hydrogen beams on tungsten

SURFACE

SCIENCE

11 (1968) 227-241 8 North-Holland

THE DEPOSITION

OF HYDROGEN

BEAMS

Publishing Co., Amsterdam

ON TUNGSTEN*

RALPH KLEIN National Bureau of Standards, Washington, D.C. 20234, U.S.A. Received 5 February

1968

The deposition and chemisorption of hydrogen on tungsten have been investigated with a molecular beam apparatus to which a field emission microscope is attached. The FEM is operated under liquid helium. The surface temperature of both the surface on which the deposition is made, and the beam, can be varied independently. Advantage is taken of the formation of a sharp boundary line after spreading on a shadowed emitter to evaluate the effect of beam temperature on the condensation process. There is a very marked effect which suggests that the deposition depends on a critical velocity for the hydrogen molecule in the beam. This effect is found only for the second layer where the adsorption forces are relatively weak. The course of the deposition in the first layer can be followed from the characteristics of the electron emission from the tip as coverage proceeds.

1. Introduction Chemisorption processes have been investigated through the use of several experimental techniques, such as field emission microscopy, flash desorption, and low energy electron diffraction, that can provide detailed information on the interaction of the adsorbed species with the substrate. The adsorption of hydrogen on tungsten is of particular interest because of the relative simplicity of the hydrogen molecule, and the absence of both diffusion into the lattice and compound formation. Several flash filament and field emission studies have been made on this systemr-*). It is the purpose of the present work to extend these investigations through the use of molecular or atomic beamss31°) with a field emission microscope. Hess and Drechslerrr) previously reported the application of such a method for barium on tungsten. A combination of beam with field emission offers several advantages. The surface temperature of the target, that is the field emitter, may be varied independently of the beam, over the entire temperature range up to the melting point of the emitter. The temperature of the beam is set by that of the effusion source. An atomic beam can be conveniently prepared by the thermal dissociation of the molecular species. A comparison between deposition from molecular and atomic beams is thus possible. That beams can * This work was supported

by the U.S. Atomic Energy Commission. 227

228

R. KLEIN

be produced with controlled intensity coupled with the detail with which surfaces may be seen with the field electron emission microscope permits rather extended

observations.

2. The apparatus The apparatus consisted of an effusion cell, a differentially pumped collimating chamber, a hydrogen inlet system and a field emission microscope. An outline diagram is shown in fig. 1. A large mobile furnace could be positioned around the apparatus to provide for baking up to temperatures of 450 “C. The effusion cell and dissociator was a hollow cylindrical tungsten rod 5 cm long with a wall thickness of 0.01 cm, and an effusion hole 0.028 cm in diameter. It was made by spark erosion of a 0.3 cm diameter tungsten rod. The cell could be heated resistively to any required temperature. It was not necessary however, to exceed 2500°C. Mounting was such as to provide for thermal expansion of the rod and water cooling of the mounting bracket. Temperature measurement was made with an optical pyrometer through a viewing port. Below incandescense a thermocouple was employed. The cell could be operated continuously for many hours at 2500°C. The hydrogen pressure in the cell was monitored with a diaphragm gauge. Hydrogen was introduced by diffusion through a heated palladium thimble. The pressure in the effusion cell could be continuously maintained to within 2% of its set value. To avoid the uncertainties of calculation of thermal transpiration effects between the heated cell and the point of the pressure measurement as well as the pressure drops through the connecting tubing, calibration of flow was made directly as a function of the diaphragm capacitance gauge reading. This was done by supplying hydrogen to the tungsten effusion cell through a precision leak valve from a calibrated volume. The pressure drop rate in the calibrated volume served as a measure of the total flow through the effusion orifice. The field emission microscope, operated under liquid helium, was attached to the beam apparatus by a palladium tube to enhance condensation of scattered hydrogen. A wide temperature variation for the observation of the adsorption of hydrogen was possible since the loop to which the emitter was attached could be heated resistively. The field emitter was fabricated from tungsten. The details of the assembly have been described previouslyl). TWO emitter mountings were employed. In the first, the emitter was positioned with its axis perpendicular to the beam as shown in fig. 2a. Since only the upper portion of the surface is exposed to the beam, the lower portion remained clean. In the second method of mounting, the emitter was aligned

DEPOSITION

OF HYDROGEN

ON TUNGSTEN

WINDOW

BlONDITRAP

II

,,--., T==r \

I

, _ I

BAFFLE

DIFFUSION PUMP

Fig. I.

Molecular beam - field emission microscope system.

229

R. KLEIN

230

HYDROGEN

BEAM

HYDROGEN

EMIT

TING

BEAM

I

TIP

PHOSPHOR SCREEN

EMITTING

TIP

/ (a) EMITTER

EMITTER

SCREEN

ASSEMBLY

Fig. 2.

(b)

Ii0

PHOSPHOR

Emitter

position

ASSEMBLY

with respect to beam. (a) Emitter emitter parallel to beam.

perpendicular

to beam;

(b)

with its axis parallel to the beam. The phosphor screen of the emission tube was inclined at 45” as shown in fig. 2b. Work function-coverage measurements could be made without the necessity of heating to obtain uniform distribution of the adsorbate as would be the case in the arrangement shown in fig. 2a.

3. Deposition with the emitter axis perpendicular to the beam The hydrogen

effusion

orifice, 0.28 mm in diameter,

was 1 m from the

field emitter. The relationship of the beam to the field emission assembly was that shown in fig. 2a. The first experiments consisted in depositing molecular hydrogen at 300°K on an initially clean tungsten tip at 4.2 “K. Fig. 3 illustrates the field emission patterns obtained. Characteristically for very low coverages, the surface involved exhibits a mottled appearance. This is seen especially clearly where, because of the geometry of the deposition, the surface is ‘shadowed’. Deposition is only on the upper part of the approximately hemispherical tungsten surface. The contrast between the lower clean portion of the surface and the upper portion where deposition has occurred is evident even for low coverage. The mottling appears when as little as 0.05 of a monolayer has been deposited. As the coverage increases, the pattern appears more uniform. Comer et a1.l) first observed that the presence of a physisorbed layer of hydrogen on tungsten can be distinguished by the ease of its surface mi-

DEPOSITION

fig. 3c

OF HYDROGEN

231

ON TUNGSTEN

fig. 3d

Fig. 3. Deposition of molecular hydrogen on emitter whose axis is perpendicular to the beam. Beam density 2.4 x 1Ol2molecules/cm2 sec. (a) Clean tungsten.(b) Deposition time45 sec. (c) Deposition time - 2 min. (d) Deposition time - 4 min.

gration relative to the chemisorbed layer. In fact, the physisorbed layer is mobile well below 30°K. Mobility of the chemisorbed layer is not evident before about 170°K. In the deposition experiments with hydrogen on the tungsten tip perpendicular to the beam axis, a concentration gradient results because of the shadowing process. Fig. 3 shows a sequence of field emission patterns in which a beam of molecular hydrogen at 300°K strikes the target at a rate of 2 x 1OL2molecules/cm2 set the tip temperature being maintained at 4.2”K. As the exposure time increases, the emission decreases both because of work function and pre-exponential factor changes. Because of the approximately hemispherical form of the emitter, the coverage roughly follows a cosine distribution. This was verified from densitometer readings of the negative of the field emission pattern photographs of fig. 3. If the tungsten surface is exposed to the beam for a sufficient time, the chemisorbed layer is completed and a more weakly bound physically adsorbed layer forms. The presence of this layer can easily be detected. By heating

232

R. KLEIN

the tip slightly (20°K is quite sufficient, for example) the weakly adsorbed layer becomes mobile. A hydrogen molecule migrates until it is chemisorbed on a ‘trap’ site and becomes immobilized. Since half of the emitter surface remains initially clean, migration can proceed until the physisorbed layer is depleted. If insufficient hydrogen was deposited to cover the entire tip after low temperature spreading, then a sharp line boundary between covered and clean regions results. This is shown in fig. 4. A sharp boundary can first be observed just beyond the grazing incidence line of the beam with the surface. Further deposition and spreading displaces the boundary until the entire tip is covered. A boundary line is never observed on the portion of the tip exposed to the beam. Since the sharp boundary is formed after spreading of the hydrogen in excess of a monolayer, it is most likely that a complete chemisorbed layer is formed up to the sharp line boundary. In the present experiments, the

fig. 4c Fig. 4.

Sharp boundary formed by second layer spreading. (a) Clean tungsten. (b) With deposit but before spreading. (c) After spreading.

DEPOSITION

OF HYDROGEN

233

ON TUNGSTEN

hydrogen is deposited on the emitter, which geometrically is rather like a hyperboloid of revolutionla). It is represented in fig. 5 as a spherical cap on a cone with the solid line indicating the deposit before, and the dotted line after, spreading. It is of interest to consider deposition on a sphere from a molecular beam (for ease of calculation, and to a rough approximation to

-

BEFORE -

-

- -

Fig. 5.

AFTER

SPREADING SPREADING

Deposition

profiles.

simulate deposition on the emitter). The maximum deposit occurs at the pole and clearly becomes zero at the equator, the grazing incidence boundary of the beam with the sphere. The concentration at the pole, in terms of number of layers, for the case where, after spreading, a monolayer tablished up to the equator, is obtained from the expression

would

be es-

tn (2) 2naz sin 0 cos 0 dB = 2xa2, .c 0

where Z is a factor expressing the coverage at 6 =O just before the spread. This gives Z=2. Thus, 2 layers must be deposited on the 0 =O surface to give a sharp line boundary at the equator of the sphere after spreading. For the actual emitter, the maximum coverage would probably not differ much from the two layers calculated for the hemispherical case. Calculation for the emitter is complicated by diffusion from the shank. The usual relationship found for sticking probabilities is that they remain constant for a fairly extended coverage range and then decrease at higher coverages4). The

234

R. KLEIN

sticking probability for second layer deposition is less than that for the first layer. It must be concluded that at the minimum time for establishment of the sharp line in the deposition of hydrogen on the tungsten tip, although there is more than a monoIayer present on the maxjmum covered regions, there is probably considerably less than 2 layers. A series of runs were made to determine the minimum time of deposition for the formation of a sharp boundary after spreading. The beam density was 2.4 x 10” molecules/cm2 set at the emitter. The effusing source was 300°K. The minimum time was found to be 16.5 min, reproducible to within 15 sec. A total of 2.4 x lOi mo~e~ules/cm~ impjnged on the target area. If the average surface density of tungsten is assumed to be 8 x lOi4 atoms~cmz, then 8 x 1014 motecules could be accommodated in the monolayer, considering the calcellation of the factor of 2 from the two sites per molecule of H, with the similar factor for the ratio of area of the spherical cap to projected area. This leads to an average sticking coeffcient of 0.3 for the process described. It is a lower limit for the sticking coefficient of hydrogen on clean tungsten at 4.2%. 4. Spontaneous spread The following observation was made when hydrogen at 300°K was deposited from the beam on the 4.2”K tungsten tip. With the same beam conditions as described, the beam was interrupted at times to observe the field emission pattern. The tip was not heated. For about 35 min there was a progressive darkening of the pattern toward the grazing incidence line as the deposition progressed. Then a spontaneous spreading commenced with a sharp delineation between the clean and hydrogen covered surface. With a few seconds further exposure to the beam, the spreading was completed over the whole tip. There was no stoichiometric re~ationshjp between the amount of additional surface coverage and the hydrogen deposited. The latter could account only for a negligible shift of the boundary. The spontaneous surface migration of llydrogen has been observed previously with field emission using a tungsten and a nickel tipl). However, with molecular beam deposition, the sharply defined condition and the remarkabIe instability of the spontaneous spread region can be demonstrated. The addition of an amount of hydrogen equivalent to perhaps +; of a monolayer was sufficient to extend the spreading to at least monolayer coverage over a clean surface corresponding to about $ of the tip area. Experiments showed, on the other hand, that when just su%cient hydrogen was deposited to give monolayer layer coverage up to the grazing boundary of the beam after low temperature spreading, further deposition of hydrogen corresponding to the amount described above would give only a small shift of the sharp boundary line

DEPOSITION OF HYDROGEN

235

ON TUNGSTEN

fig. 6a

fig. 6b

fig. 6c

fig. 6d

Fig. 6. Successive second layer deposition and spreading, (a) Clean tungsten. (b) Sharp boundary. (c) (b)exposed to intensity of 2.4 x lOlz molecules/cm2 set for 5 min and spread. (d) Five additional minutes on (cf and spread.

after spreading. This is apparent from fig. 6. The spontaneous spreading has characteristics of a melting process. The deposition time required to produce the spontaneous spread was about twice that for the completion of a monolayer to the grazing iine after spreading. In view of the previous analysis for the angular distribution of the deposit prior to the spread, it may be inferred that three to four layers of hydrogen are required on a shadowed tip as the condition for the spontaneous spread. 5. Depasition of atomic hydrogen By operating the dissociator at 2700°K with a hydrogen pressure of 30 microns, a beam of hydrogen may be produced that is -98.5% atomic if undiluted by background gas. Because the tip ‘sees’ the dissociator, it was necessary to establish that no significant heating effects at the tip were caused by radiation. This was accomplished by first depositing molecular hydrogen

236

R. KLEIN

on the tip in an amount sufficient to give an excess of a monolayer on the half of the emitter facing the beam. The hydrogen flow was then stopped and the dissociator heated to 2700°K. After 15 min, no field emission pattern change could be observed. Resistive heating of the loop of the emitter tip to about 20°K then gave the sharp line boundary pattern, confirming the presence of a multilayer. The radiation adsorbed by the emitter surface from the dissociator was insufficient to produce significant temperature changes. A clean tungsten surface was then exposed to the atomic hydrogen beam. The exposed half of the tip showed decreasing electron emission as the deposition proceeded (similar to fig. 3). This demonstrated that condensation on the surface, at least in the first layer, occurs. Indication of second layer formation, as evidenced by a sharp boundary after low temperature spreading, was absent. This was true even for exposure of the surface to three times the mass of hydrogen that would have been sufficient for second layer formation in the case of a 300°K hydrogen beam. A very low sticking coefficient of atomic hydrogen on a monolayer surface would account for this observation. A dependence of sticking probability on the beam temperature is an alternative possibility, and is indeed the correct interpretation. With the dissociator at lOOO”K, the second layer was not found even for long exposure times. At this temperature, and for the 30-50 micron pressure of hydrogen in the effusion cell, the fraction of atomic hydrogen in the beam was considerably less than one per cent. The pressure in the effusion cell was adjusted at each temperature so that the total mass flux rate was maintained constant. For the effusion cell temperature of 540 OK, the deposition time required for the formation of the sharp line boundary after spreading was approximately double that for 300°K.

6. Emitter parallel to beam The flux intensity per unit surface area is reasonably uniform over the surface with the emitter tip positioned with its axis parallel to the beam. The variation of flux over the tip is about 50”/,, based on a cosine distribution over a spherical surface. Observations on the relationship of total incident flux to both work function and the pre-exponential factor as obtained from Fowler-Nordheim plots are not very instructive because of the only moderate (0.4 eV) work function increase over the entire coverage range from clean surface to monolayer. However, decrease of the pre-exponential term and increase of work function with increasing coverage is observed. This is in accord with previous observations. A more sensitive indication of the progress of the deposition is obtained from the plot of the voltage for a fixed emission current as a function of time. This is shown in fig. 7a. The voltage is given by

DEPOSITION

I

5

I

IO

OF HYDROGEN

I

I

I

I

15

20

25

30

TIME.

Fig. 7.

231

ON TUNGSTEN

3! 5

minutes

Voltage for 1O-7 A emission as a function of deposition 2.4 x 1Ol2molecules/cm2 sec.

time. Beam intensity

the Fowler-Nordheim relationship, is determined by the work function and pre-exponential terms as well as the factor relating voltage and field. For hydrogen deposition, the pre-exponential factor is the most decisive in establishing the voltage variation with coverage. It is interesting to note that there is an initial rapid voltage increase but that after a time the voltage approaches a steady or only a very slowly increasing value. In the figure shown most of the voltage increase occurred in the first twelve minutes of deposition. A surface held at elevated temperatures exposed to the hydrogen beam accumulates a surface layer until a steady state condition is attained. The coverage characteristics depend both on the temperature of the surface and the intensity of the beam. In deposition on a 4.2 “K surface, partial coverage of the tip followed by heating to 250°K indicated some loss of hydrogen from the emitter surface by diffusion down the shank, but the effect was very small. When the tip is maintained at elevated temperatures, desorption is balanced by deposition from the beam. The steady state concentration is, of course, shifted toward lower values with increasing temperature. At

238

R. KLEIN

sufficiently high temperatures, at a fixed beam density, the concentration on the surface remains virtually zero. With a beam density of 2.4 x 1OL2 molecules/cm2 set this temperature is above about 525 “K. Hydrogen adsorption on tungsten involves several binding states, as has been shown by flash filament experiments4). The manifestation of these states in the beam experiments is by the attainment of an equilibrium coverage independent of surface temperature and beam intensity within certain temperature limits. Fig. 7b shows deposition at a tip temperature of 425°K. A steady state surface concentration is attained after about twenty minutes. A similar experiment at 300°K showed no difference in the steady state condition, but it was attained in a shorter time. Confirmation of the energy states can also be made by observing the voltage changes as a covered tip is heated. The graph of fig. 8 shows this. Hydrogen was deposited on the tip at 4.2”K to monolayer coverage. The

5500

0

I 50

I 100

I 150

I 200

I 250

TEMPERATURE,

Fig. 8.

Resorption

I 300

I 350

I 400

I 450

I 500

“K

from a hydrogen monolayer on tungsten surface as followed by emitter voltage measurements.

tip was heated for 15 set successively at the temperatures indicated. After each heat treatment, the voltage required for 10m7 A emission from the tip was measured. Three binding states are clearly indicated. The lowest energy state is desorbed in the region O-250°K. The intermediate state is desorbed between 250°K and 375 “K, and most strongly bound above 375 “K. This is in agreement with the results of Ricca et al.“) who find three binding states for hydrogen on tungsten, although they further delineate two substates for

DEPOSITION

each of the three states.

OF HYDROGEN

Mimeault

239

ON TUNGSTEN

and Hansens)

report

only two binding

states. Moore and Unterwald?) report only one state but then the initial adsorption temperature of 300°K was high and the percentage of the strongly bound state is sufficiently small that it probably escaped detection,

7. Beam temperature and its effect on condensation The time to the formation of the sharp boundary line was found to be strongly dependent on the beam temperature. The sticking coefficient of the first or strongly chemisorbed layer appears to be insensitive to beam temperature. The emission characteristics of the tip exposed to equal hydrogen fluxes at various beam temperatures are virtually the same with respect to voltage for a given emission current or to pre-exponential factor, provided monolayer coverage is not exceeded. As has been noted, the appearance of the sharp boundary after spreading is associated with greater than monolayer coverage on the areas of maximum coverage. In contrast to the insensitivity of the sticking coefficient to beam temperature for less than monolayer coverage, the time required for the sharp boundary formation (after spreading) depends on the temperature. Table 1 illustrates this. Chubb and TABLE

Time to formation Temperature of beam source (“IQ _____. 300 435 470 540

1

of sharp boundary line

Time (min) --. 16.5 18 25 32

Pollard’s), in studies of hydrogen condensation on surfaces at 3.6 to 3.9X, demonstrated an effect of gas temperature on the sticking coefficient. They found a value of approximately 1.0 for a gas temperature of 100°K. This decreased to about 0.7 for 600°K. The requirement for trapping a molecule impinging on the surface of a solid has been discussed by Zwanzigr4) with a simplified model using a linear chain and a truncated potential for the interaction of the impinging particle and the chain end. This model was elaborated by McCarroll and Ehrlichrs). The solution of the equations of motion lead to a critical kinetic energy such that below that energy all impinging particles are trapped and above that energy, all particles are reflected. The model, being one dimensional, is highly simplified. This,

240

R. KLEIN

of course, is well recognized. The idea of a critical velocity of trapping, can be utilized for the consideration of second layer condensation. The beam intensity in terms of velocity distribution is

The product of deposition time to the formation of the sharp boundary and the beam Aux of molecules with velocities up to the critical velocity vOshould be a constant. That is .=*{I--(I+$)exp(-$)j. where t and T are time and temperature and K and b are constants. The data are not extensive enough to provide a critical test of this relationship, but by using the data of table 1 with the estimate that ten minutes are required to establish the monolayer on the portion of the surface whose normal forms the smallest angle with the beam, an equivalent temperature of v~,ib=500”IS is derived. It is emphasized that this is not a simple, onedimensional situation. The atcual process is one in which molecular hydrogen is condensing on a tungsten surface covered with a monolayer of hydrogen. Comparison with the computed values of critical kinetic energy of McCarroll and Ehrlichr5) would be quite unrealistic. Nevertheless, the existence of a critical velocity of trapping does appear to be supported by the present experimental results. Acknowledgements

Thanks are expressed to Mr. L. Leder (now with Beach, California) who contributed greatly to the of the beam apparatus. Mr. J. Pararas guided the many valuable suggestions. Mr. J. W. Little aided operation of the equipment.

Philco-Ford in Newport design and construction construction and made in large measure in the

1) R. Gamer, R. Wortman and R. Lundy, J. Chem. Phys. 26 (1957) 1147; R. Wortman, R. Comer and R. Lundy, J. Chem. Phys. 27 (1957) 1099. 2) W. J. M. Rootsaert, L. L. van Reijen and W. M. Ii. Sachtier, J. Catalysis l(1962) 416. 3) J. A. Becker, in: Proc. Second Inrern. Congr. on Catnlysis, Vol. II (Editions Technip, Paris, 1961) p. 1777. 4) F. Ricca, R. Medana and G. Saini, Trans. Faraday Sot. 61 (1965) 1492. 5) J. Eisinger, J. Chem. Phys. 29 (1958) 1154. 6) T. W. Hickmott, J. Chem. Phys. 32 (1960) 810. 7) G. E. Moore and F. C. Unterwald, J. Chem. Phys. 40 (1964) 2626. 8) V. J. Mimeault and R. S. Hansen, J. Chem. Phys. 45 (1966) 2240.

DEPOSITION

OF HYDROGEN

ON TUNGSTEN

241

9) G. W. Sears and J. W. Cahn, J. Chem. Phys. 33 (1960) 494. 10) R. T. Brachman and W. L. Fite, J. Chem. Phys. 34 (1961) 1572, reported on deposition of hydrogen from molecular beams. 11) E. Hess and M. Drechsler, in : IV. Intern. Kongress fiir Elektronenmikroscopie, Berlin 19.58, Verh. I(l960) p. 811. 12) M. Drechsler and E. Henkel, Z. Angew. Physik 6 (1954) 341. 13) J. M. Chubb and I. E. Pollard, Vacuum 15 (1965) 491. 14) R. W. Zwanzig, J. Chem. Phys. 32 (1960) 1173. 15) B. McCarroll and G. Ehrlich, J. Chem. Phys. 38 (1963) 523.