- Email: [email protected]
Lithium adsorption on the (112) face of tungsten
SURFACE
SCIENCE 34 (1973) 368-384 0 North-Holland
LITHIUM
ADSORPTION
V. K. MEDVEDEV, Institute
of Physics,
Publishing Co.
ON THE (112) FACE
A. G. NAUMOVETS
Ukrainian
SSR Academy
Received 20 May 1972; revised manuscript
OF TUNGSTEN
and T. P. SMEREKA of Sciences,
Kiev,
U.S.S.R.
received 31 July 1972
Lithium adsorption on the (112) surface of tungsten single crystals has been studied by the methods of contact potential difference (CPD) measurement and low energy electron diffraction (LEED). Dependences of the work function, adsorption heat and atomic structures of lithium films on the lithium surface concentration are obtained. A close correlation is found between the changes in the film structure occurring with coverage and features of the work function and adsorption heat concentration dependences. It is concluded that interaction between lithium adatoms on W(l12) is highly anisotropic. Order-disorder transitions in adsorbed lithium films are studied.
1. Introduction The relationship between the electronic properties of surfaces and their atomic structure is one of the oldest but hitherto unresolved questions in surface physics. Now due to progress in LEED and vacuum technology it has become possible to perform the necessary experiments under well defined conditions. Previously l) we measured by the field emission method the work functions rp of basic tungsten crystal planes as a function of lithium atom concentrations nLi. The curves cp (12~~)are found to have highly different shapes. For the plane W(112) with its trough atomic structure, the curve cp(Q) has a rather unusual step shape. At the singular points of that curve the lithium concentration is related to the tungsten surface atom concentration as small integers. This fact suggests that at these points the film structures could be coherent with the substrate structure21. The main purpose of the present paper is to study lithium film structures on W(112) by a direct method and to correlate them with the data on the work function and adsorption heat. 2. Experimental The structure of lithium films was examined using a planar geometry LEED apparatus. The intensities at the fluorescent screen were measured 368
LiADsORPTIONONW
369
with a telescopic photometer. The work function was measured by the Anderson’s retarding field method. The lithium atomic beam was calibrated using the dependence Cp(n‘i) measured in ref. 1. The samples of dimensions 10 x 3 x 0.26 mm3 were cut from a tungsten single crystal refined by zone melting. The surface of the plate was parallel to (112) within 6 min of arc. After electroerosive cutting and grinding the final step of the sample preparation was electrolytic polishing during which a layer at least 0.1 mm thick was removed. The sample was spot welded through four 0.2 mm tungsten wires to the molybdenum supports sealed into a Dewar foot. This mounting allowed the crystal to be cooled down to liquid nitrogen temperature or to be heated up to T< 1500 “K by passing current through the supports. Up to higher temperatures (as high as 2800 OK) the crystal could be heated by electron bombardment with a gun placed behind the crystal. A tungsten-tantalum thermocouple was used to measure the temperature. The crystal was carefully outgassed. It was also decarbonized by annealing at 2000 “Kin oxygen atmosphere (= lo- 6 torr) during w 20 hr. The lithium source was described elsewherer). The tubes were fitted with titanum getter pumps and field emission gauges. At working conditions the pressure of the gases active in adsorption did not exceed 10-“-10-‘2 torr. 3. Experimental results 3.1.
STRUCTURES
OF LITHIUM
FILMS
A series of diffraction patterns for the W(112)-Li system corresponding to successively increasing lithium coverages is shown in fig. 1. These patterns are also presented schematically in fig. 2. The lithium films on W(112) are substantially ordered when the crystal temperature during lithium deposition is as low as 77°K (lithium flux was N lOl2 atoms cm-2 see-’ ). Crystal annealing at 200-300 “K and subsequent cooling to 77 “K causes the intensity of additional diffraction spots to increase by 20-30x. It must be emphasized that at room temperature no additional beams could be observed throughout the coverages used (< 5 monolayers) and lithium deposition results only in an increase of the background intensity. Fig. 1 shows considerable differences in the intensities of the spots located above and below the axis k= 0.This is due to the fact that the (112) tungsten face does not possess a symmetry plane of the (111) type. Because of dynamical effects the substrate asymmetry results in an intensity asymmetry of both normal and additional beams (the latter have comparable intensities at electron energies < 20 eV).
370
V. K. MEDVEDEV,
A. G. NAUMOVETS
Fig. la-lf
AND T. P. SMEREKA
Li ADSORPTION
Fig.
ON
W
371
372
V. K. MEDVEDEV,
A. G. NAUMOVETS
Fig. lm-lr
AND T. P. SMEREKA
Li ADSORPTION
ON w
373
Fig. 2. (a-r) are spot arrangement schemes for the respective patterns of fig. 1. The lower halves of diffraction patterns are shown. The normal spots are shown by dark circles, additional spots by light circles and streaks by dotted lines.
The (112) tungsten plane consists of close-packed atomic rows with troughs in between. As will be shown below, the lithium atoms at first fill the troughs. The second lithium atomic layer falls into troughs existing in the first layer, etc. In accord with this observation we introduce the relative concentrations c.‘ = n”!/n w, where I$! is the absolute lithium surface concenLI tration in the ith layer and n, is the tungsten atomic concentration in the (112) plane equal to 8.2 x 1Or4 cm-*. The first changes in the diffraction pattern are found at the lithium con-
Fig. 1. Diffraction patterns: (a) clean (112) face W; the surface concentration of tungsten atoms nw = 8.2 x 1Or4cm-2. (b-r) are patterns of the W(112)-Li system. The concentration ratio n&w is: (b) 0.25; (c) 0.3; (d) 0.33; (e) 0.42; (f) 0.5; (g) 0.67; (h) 0.71; (i)O.8; (j) 0.87; (k) 1.0; (1) 1.52; (m) 1.7; (n) 1.9; (0) 2.2; (p) 2.6; (q) 3.3; (r) 3.9. The crystal temperature is 77°K; the adsorbed film was annealed at 300°K; the electron energy is 50 eV.
374
V. K. MEDVEDEV,
A, 0. NAUMOVETS
AND
T. P. SMEREKA
centration ~5 x 1013 cmm2. Diffuse streaks appear between the normal reflections along the k-axis, i.e. the [iil] direction. With nLi > 1 x 1014 cmm2 the streaks break up to a series of a dim spots aligned in the same direction. On further adsorption the number of spots diminishes and they become much sharper. At n~i’~2 x 1014 cmT2 (cr =a) a pattern shown in fig. lb is observed in which additional spots with k= _tb_, k$ and split spots at k= &$ are seen (fig. 2b). This pattern corresponds in Wood’s notations) to a lattice p(1 x 4) (fig. 3b). Splitting of the additional beams (h, ++) may be
Fig. 3. (a) is the structure of W(112) face; (b-k) are structures of the lithium adsorbed films; (b) is the first layer ~(1 x4); (c)is the first layer p(1 x 3); (d) is the first layer p(1 x2); (e) is the incoherent structure of the first layer; n~i/nw = 0.75; (f) is the first layer p(1 x 1;) (g) is the second layer ~(1x4); (h) is the incoherent structure of the second layer; (i) is the close-packed second layer; (j) is the third layer; (k) is the fourth layer; (1) is the W(112) along [Till.
Li ADSORPTION
ON
W
315
due to the existence of antiphase domains having the structure p( 1 x 4) and separated along the [Till direction by an odd number of lattice parameters a2 so that antiphase boundaries are parallel to [X10] direction [see, e.g., eq. (45) in the review paper, ref. 41. Progressive lithium deposition causes the fractional order beams of the ~(1 x 4) structure to move, as is shown schematically in fig. 2c, and finally to merge at points k = 5%; 2 3; ) 4; etc. The resulting pattern corresponds to a lattice p(1 x 3) (figs. Id, 2d, 3~). Based on the concept of refs. 5 and 6 this behavior can be explained by formation of a statistical mixture of the p(1 x 4) and p( 1 x 3) unit cells in the interval +< c1 < f_ A simple kinematical analysis (similar to that given in ref. 5, p. 440) supports this conclusion. Within 3 < ci <+ a transition from p(1 x 3) to the p(1 x 2) structure is observed. The nature of this transition is quite different however. The lithium atoms in excess of the stoichiometric concentration ci =3 are not randomly distributed but incorporate to form domains of the p(1 x 2) structure. This follows from the fact that the spots of both structures coexist over a wide coverage range (fig. le). In the coverage interval + cc1 < 1 the fractional order beams of the p( 1 x 2) structure transform to streaks smeared along the h-axis and shift outwards with coverage increase. Simultaneously weaker streaks, which are due probably to double electron scattering, are shifting to k = 0 (figs. 1g, h, i, j). Gerlach and Rhodin suggested’ss) that patterns of this type correspond to one-dimensionally incoherent structures. In this case the film consists of atomic rows arranged in the troughs and randomly shifted relative to each other. The interatomic spacing is the same in all the rows. It diminishes continuously with coverage increase and in the interval +
376
V. K. MEDVEDEV,
A. G. NAUMOVETS
AND
tmn
T. P. SMEREKA
c
0
--/
a
-2
0 0
5
/O
/&/O”C&
‘5
a
Fig. 4. The work function q, adsorption heat q and spot intensity Z versus the concentration n of adsorbed Li atoms on the W(112) face. (a) (1) r&z), CPD method, T=3OO”K; (2) q(n), CPD method T=77”K; (3) p,(n), field emission method l), T=77”K; (4) q(n). (b) Z(n): (1) additional spot (0, -4) of the structure p(1 x 2); (2) normal spot (0, -1); (3) additional spot (0, -$) of the structure p(1 x 4). The letters above mark the coverages corresponding to the various parts of fig. 1.
p(1 x 1). The next diffraction pattern, which is apparently associated with the second layer, has rather bright additional beams at k= A$, less bright at k= At and still less bright ones at k= ++ (h integer) (fig. 11). This pattern exists in a wide coverage range (see in fig. 4b the spot intensity dependence on nri). Assuming the multiple electron scattering we attribute this pattern to the structure p(1 x 4) in the second layer. The second layer atoms are believed to fall into troughs formed by the first layer. At c2 M 0.5 the k = 3 spots undergo some broadening in the [Till direction and at c z ~0.7 to 0.75 they practically disappear (fig. lm). At c2 20.7 the k= k$ spots begin to spread into streaks (of nonuniform intensity) moving to k= _t 1 with ca increasing; the k= k$ spots (due to double scattering) spread into similar streaks moving to k = 0 (fig. 1n). Thus t : second layer atoms at c2 > 0.75 also form one-dimensionally incoherent structures. Unlike
LiADSORPlTONON
the first layer the streaks
in the second
377
W
layer patterns
do not collapse
with
the normal k = f 1 spots. Instead they stop at c2 x 0.9 when the shortest interatomic spacing, within the error limits (several percent), equals the atomic diameter in lithium metal (fig. 3i; for simplicity, a model of the structure with rows in phase is shown, although in reality there is some disorder in the mutual position of the rows since the additional beams in fig. In are streaked along [ilo]). After that rather monotonous patterns are observed which gradually replace each other during the lithium deposition. As judged by doses of lithium, the patterns correspond to the third (figs. lo, p) and fourth (figs. lq, r) layers. In the third layer the lithium adatoms form a centred rectangular lattice (fig. 3j) with the atomic spacing along [iil] somewhat larger than in the close-packed second (fig. 3i) or in the fourth (fig. 3k) layer. The fourth layer lattice is again simple rectangular with the atomic spacing along [Till equal to the lithium atomic diameter (the interpretation of patterns fig. lo-r is also based on the double scattering model). It must be emphasized that the formation of the third and the fourth layers is accompanied by a rearrangement of underlying layers whereby all the film gets a new equilibrium structure. This can be deduced from the fact that in the diffraction patterns from the films 3 or 4 layers thick the spots characteristic of the second or third layer are absent, whereas spots characteristic of the tungsten substrate spacings are clearly visible. The rearrangement of the film is taken into account in drawing the models of the third and the fourth layers (figs. 3j, k). After the completion of the fourth layer no other ordered structures were observed. With subsequent lithium deposition, all spots weaken rather fast so that for the six-layer thick film they are hardly seen in the increased background. Under these conditions the ordered film cannot be obtained even after a careful annealing. 3.2. WORK
FUNCTION
The dependence of the work function c~ on absolute lithium coverage on W(l12) is shown in fig. 4a. Curves 1 and 2 were obtained by the CPD method at crystal temperatures 77 and 300”K, and curve 3 by the field electron emission method at 77”Kl). Note the close agreement of the data taken by different methods and the small influence of temperature on the curve Cp(%I. The intensities of spots of the p( 1 x 2), p (I x l), p (1 x 4) structures versus nLi are shown in fig. 4b. Bends and the minimum in the p(nLi) curve definitely correlate with the formation of these structures. The work function changes effectively terminate when the formation of the close-packed second layer
378
V. K. MEDVEDEV,
A. 0. NAUMOVETS
AND T. P. SMEREKA
is complete. It is important to mention that the structure analysis the concentration scale of the cp(11~~)curve calibrated in ref. 1.
confirms
3.3. HEAT OF ADSORPTION To determine the heat of adsorption q we have measured the equilibrium work functions corresponding to various temperatures and a constant flux N of lithium atoms (fig. 5). The q (Y1Li)curves (fig. 4a) allow this relation
Fig. 5. (1,2) Dependences of the equilibrium work function for the W(112)-Li system on the crystal temperature: (1) N = 9.3 x 1011cm-2 set-l; (2) N = 6 x lOlo cm-2 set-I; (3) adsorption isobar N= 9.3 x lollcmm2 set-‘.
to be converted to an adsorption isobar nLi (T) (fig. 5, curve 3). q(nLi) was calculated from the isobar using a well-known expression from the absolute reaction rate theory”):
4 (nLi) = k,T In T,
where v is a frequency factor and k, is the Boltzmann constant. Independent data on the adsorption heat and frequency factor can be obtained without additional assumptions from two isobars at different lithium fluxes (fig. 5, curves 1 and 2). In this way we obtained vx 1013 set-’ which is consistent with the theoretical value k,T/h, (hp is the Plank constant) for mobile adsorption (ref. 11). The q versus nLi curve is presented in fig. 4a. At , we substituted into the expression for q the concentran LI.>82x10’4cm-2 . tion of lithium atoms in the second layer nlf,)=n,,-8.2 x 1014 cm-*.
Ii
ADSORPTION
ON
379
W
Thereby a natural assumption was made that the evaporation rate of lithium from the second layer is much greater than that from the first one. The initial heat of adsorption obtained here is in good agreement with a value of 2.72 eV obtained in ref. 12 where the field desorption of lithium from tungsten was studied. 3.4. ORDER-DISORDER TRANSITIONS IN
LITHIUM
FILMS
To obtain information about order-disorder transitions in the lithium films we measured the temperature dependences of spot intensities for various structures. In our experiments, only the small central part of a spot was projected onto the photomultiplier input. After deduction of the background intensity, the spot intensities were plotted on a relative scale Z/Z,, where I, is the corresponding intensity at 77°K (fig. 6). It is clear from fig. 6 that practically none of the lithium structures can be observed at room temperature. As in our preceding papers 931% we have investigated the influence I
I
IO
1
f 05
0
Fig. 6. Temperature dependence of spot intensity: (1) additional spot (1, -4) of the structure p(1 x 4); (2) additional spot (0, -4) of the structure p(1 x 2); (3) spot (0, -1) of the structure p(1 x 1); (4) additional spot (0, - $) of the structure p(1 x 9); (5) spot (0, -1) of the clean face. Electron energy is 50 eV, lithium concentrations correspond to the intensity maximum of additional spots. (6) Theoretical behaviour of the spot intensity for Ising’s modelzl). The transition temperature is chosen equal to 230°K.
of the order-disorder transition on the work function. The result is similar, i.e. the work function change accompanying the disordering does not exceed several hundredths of one ev. 4. Discussion
The W (112)-Li system has a great number of different structural states. The symmetry of the two-dimensional lattices indicates a rather complicated interaction between the lithium atoms. In the lattices p(1 x 4), p(1 x 3) and
380
V. K. MEDVEDEV,
p(1 x 2) the lithium
A. G. NAUMOVETS
atoms are arranged
AND T. P. SMEREKA
to have a maximum
spacing
along
the [iil]direction and a minimum spacing (equal to one between the adjacent troughs) along [ilO]. As is clear both intuitively and according to statistical calcul ationss), such structures indicate that there must exist repulsive interatomic forces along the troughs and attractive forces across the troughs. Such an anisotropic interaction seems to be caused by the highly anisotropic substrate structure. It may be assumed that the repulsive forces along the troughs are due mainly to be dipole-dipole interaction. Indeed the dipole moment at low coverages, evaluated as p=Acp/4nnLi (Acp is the work which function change) equals for this system to z 1.7 D. Some contribution is expected to be comparatively small at appreciable dipole moments can come fron long-range interaction via the substrate electron gas14). Along the [ilO] direction the adatoms are separated by rather strongly protruding tungsten atomic rows which may partly screen the fields of the charged lithium atoms and at the same time serve as mediators in their exchange interaction. As a result the chains -Li-W-Li-Ware formed along this direction. It is worthwhile to discuss a possible relation between the lithium film structures and the unsaturated orbitals on the substrate surface. This factor usually is ignored in alkali metal adsorption whereas in other cases the concept has proved to be extremely fruitfu115+ 16). Due to its small radius the lithium atom must be more sensitive to the surface potential relief than are other alkali atoms. Fig. 7 shows schematically the
L-ml
Fig. 7.
Scheme of orbitals on the surface of W(112) face.
unsaturated bonds in the tungsten (112) face. It can be seen that the geometry of the orbitals favours adatom arrangement into linear chains along the [ilO] axis, as i; observed experimentally. From this point of view one also can explain the lack of coherence in the positions of neighbouring rows when the interatomic spacing along [iil] smoothly diminishes ($
Li
Now let us discuss
ADSORPTION
ON
381
W
in more detail the transitions
between
different
first
layer structures. When the interatomic distances along troughs are larger than 2a2, the atomic repulsion is not strong enough to cause incoherence between the film and substrate periods at unstoichiometric Li concentration. That is why the p (1 x 4)+p (1 x 3) transition occurs via mixing of the cells 5s6, and the p (1 x 3)-p (1 x 2) transition by a growth of the p (1 x 2) domains. As could be expected in the coverage range 3 < ci < 4 the measured adsorption heat only weakly depends on the concentration (fig. 4a). The possibility of first order phase transitions in systems with dipole interaction [in our case corresponding to the condensation of excess atoms to produce a more dense phase p(1 x 2)] was substantiated theoretically by Bol’shov17). In the p(1 x 2)+p (1 x 1) transition the film lattice parameter along [iil] diminishes gradually showing an increased role of the repulsive forces at agrees also with a the interatomic distances a,
382
V. K. MEDVEDEV,
A. G. NAUMOVETS
AND T. P. SMEREKA
that ,u,~(I x 1) = + ,u=~(1 x 2). As far as we know, it is a rare case when the mutual adatom depolarization (due not only to Coulomb but also to other types of interaction) compensates precisely the barrier change which could occur due to a concentration increase. One layer having a p (1 x 1) structure is far less than the lithium coverage which, according to Gerlach and Rhodin’s terminologys), is reasonably to be assigned to the physical monolayer. As the second layer is filled, the work function decreases again and reaches a minimum on formation of the structure p (1 x 4) when 8~0.8 (e= 1 corresponds to the physical monolayer which is the close-packed two-layer film). Further packing of the second layer leads to a work function increase and at 8 = 1 it becomes practically constant. In this respect the W(112)-Li system is similar to other related systemsS-10,13,19,30
>*
The work function is weakly dependent on long-range order (see section 3.4). The barrier properties seem to be determined rather completely by short-range order in the adatom arrangement. This behaviour may result from a small coherence width of electrons overcoming the surface barrier. Proceeding to disordering of the lithium films, let us note that in the first layer the same tendency is observed as in other related systems g*13): temperature of disordering increases with the film packing indicating an increased energy of atomic interaction. Fig. 6 shows a theoretical curve according to Ising’s two-dimensional model 21) I = 1,(1 - T/T,)‘, where Tc is the disordering temperature which for the sake of illustration is chosen to be 230°K. From fig. 6 it is clear that this relation does not agree with the experiment. This could be caused by the fact that the theory neglects long-range interaction which in our case is rather significant. Indeed, for the system W (100)-H, for which the long-range interaction is less essential, the experiment agrees quite well with Ising’s modelz2). It must be noted that the structures of the third and forth layers disorder at rather low temperatures (- 150-180°K). These values are far less than the melting point of lithium metal (453°K). It has been shownss) that such a strong lowering of the melting temperature in thin films is associated with the essential contributions from the free energy of the surface and the interface. It is interesting to compare the properties of lithium and sodiumlo) films on the W (112) face. Similar structures are formed for both systems, although owing to a larger atomic diameter the structure p( 1 x 1) cannot be realized in the sodium film. It seems that for this reason the work function
Li ADSORPTION
minimum
for the W (112)-Na
layer, as in the case of lithium,
ON
383
W
system is not achieved but on formation
on filling
the second
of the first close-packed
layer (in this case coverage is also less than the physical monolayer). At the same time the behaviour of Na, K and Cs films on the other substrate with the trough structure - the Ni (110) face - is essentially different. In particular, at low coverages not every trough contains adatoms. Gerlach and Rhodin concluded that in these films the repulsive interatomic forces act along the troughs as well as across them being less along the troughs. Further investigations are necessary to elucidate the cause of this difference. 5. Conclusion The results presented demonstrate the complicated anisotropic character of lithium atomic interactions on a substrate with a strong potential relief anisotropy. Due to their small size the lithium adatoms are far more sensitive to the atomic structure of the adsorbent than other alkali adatoms. That is why for lithium the variety of realized structures is richest and differences between adatom electronic states in various structures are greatest as evidenced by the peculiar work function changes.
References 1) V. M. Gavriliuk and V. K. Medvedev, Fiz. Tverd. Tela 8 (1966) 1811. 2) V. M. Gavriliuk, A. G. Naumovets and A. G. Fedorus, Zh. Experim. Teor. Fiz. 51 (1966) 1332. 3) E. A. Wood, J. Appl. Phys. 35 (1964) 1306. 4) P. J. Estrup and E. G. McRae, Surface Sci. 25 (1971) 1. 5) L. H. Germer, J. W. May and R. J. Szostak, Surface Sci. 7 (1967) 430. 6) G. Ertl and J. Kuppers, Surface Sci. 21 (1970) 61. 7) R. L. Gerlach and T. N. Rhodin, Surface Sci. 10 (1968) 446. 8) R. L. Gerlach and T. N. Rhodin, Surface Sci. 17 (1969) 32. 9) A. G. Fedorus and A. G. Naumovets, Surface Sci. 21 (1970) 426. 10) J. M. Chen and C. A. Papageorgopoulos, Surface Sci. 21 (1970) 377. 11) S. Glasstone, K. J. Laidler and H. Eyring, The Theory of Rate Processes (McGrawHill, New York, 1941). 12) V. K. Medvedev, Izv. Akad. Nauk SSSR Ser. Fiz. 33 (1969) 528. 13) V. K. Medvedev, A. G. Naumovets and A. G. Fedorus, Fiz. Tverd. Tela 12 (1970) 375. 14) T. B. Grimley and S. M. Walker, Surface Sci. 14 (1969) 395. 15) Z. Knor and E. W. Muller, Surface Sci. 10 (1968) 21. 16) P. W. Tamm and L. D. Schmidt, J. Chem. Phys. 51(1969) 5352. 17) L. A. Bol’shov, Fiz. Tverd. Tela 13 (1971) 1679. 18) J. Matthews, in: Epitaxie-Endoruxie (VEB Deutscher Verlag fur Grundstoffindustrie, Leipzig, 1969) p. 135. 19) R. L. Gerlach and T. N. Rhodin, Surface Sci. 19 (1970) 403. 20) D. A. Gorodetsky, Yu. P. Mel’nik and A. A. Yas’ko, Ukr. Fiz. Zh. 12 (1967) 649. 21) M. E. Fisher, Rept. Progr. Phys. 30 part 2 (1967) 615.
384 22) P. J. Estrup,
V. K. MEDVEDEV,
A. G. NAUMOVETS
AND
T. P. SMEREKA
in: Structure and Chemistry of Solid Surfaces, Ed. G. A. Somorjai (Wiley, New York, 1969) p. 19-1. 23) N. T. Gladkich, R. I. Zaichik, V. P. Lebedev, L. S. Palatnik and V. I. Khotkevitch, in: Poverkhostnaya Difiziya i Rastekaniye (Nauka, Moscow, 1969). p. 222.
SCIENCE 34 (1973) 368-384 0 North-Holland
LITHIUM
ADSORPTION
V. K. MEDVEDEV, Institute
of Physics,
Publishing Co.
ON THE (112) FACE
A. G. NAUMOVETS
Ukrainian
SSR Academy
Received 20 May 1972; revised manuscript
OF TUNGSTEN
and T. P. SMEREKA of Sciences,
Kiev,
U.S.S.R.
received 31 July 1972
Lithium adsorption on the (112) surface of tungsten single crystals has been studied by the methods of contact potential difference (CPD) measurement and low energy electron diffraction (LEED). Dependences of the work function, adsorption heat and atomic structures of lithium films on the lithium surface concentration are obtained. A close correlation is found between the changes in the film structure occurring with coverage and features of the work function and adsorption heat concentration dependences. It is concluded that interaction between lithium adatoms on W(l12) is highly anisotropic. Order-disorder transitions in adsorbed lithium films are studied.
1. Introduction The relationship between the electronic properties of surfaces and their atomic structure is one of the oldest but hitherto unresolved questions in surface physics. Now due to progress in LEED and vacuum technology it has become possible to perform the necessary experiments under well defined conditions. Previously l) we measured by the field emission method the work functions rp of basic tungsten crystal planes as a function of lithium atom concentrations nLi. The curves cp (12~~)are found to have highly different shapes. For the plane W(112) with its trough atomic structure, the curve cp(Q) has a rather unusual step shape. At the singular points of that curve the lithium concentration is related to the tungsten surface atom concentration as small integers. This fact suggests that at these points the film structures could be coherent with the substrate structure21. The main purpose of the present paper is to study lithium film structures on W(112) by a direct method and to correlate them with the data on the work function and adsorption heat. 2. Experimental The structure of lithium films was examined using a planar geometry LEED apparatus. The intensities at the fluorescent screen were measured 368
LiADsORPTIONONW
369
with a telescopic photometer. The work function was measured by the Anderson’s retarding field method. The lithium atomic beam was calibrated using the dependence Cp(n‘i) measured in ref. 1. The samples of dimensions 10 x 3 x 0.26 mm3 were cut from a tungsten single crystal refined by zone melting. The surface of the plate was parallel to (112) within 6 min of arc. After electroerosive cutting and grinding the final step of the sample preparation was electrolytic polishing during which a layer at least 0.1 mm thick was removed. The sample was spot welded through four 0.2 mm tungsten wires to the molybdenum supports sealed into a Dewar foot. This mounting allowed the crystal to be cooled down to liquid nitrogen temperature or to be heated up to T< 1500 “K by passing current through the supports. Up to higher temperatures (as high as 2800 OK) the crystal could be heated by electron bombardment with a gun placed behind the crystal. A tungsten-tantalum thermocouple was used to measure the temperature. The crystal was carefully outgassed. It was also decarbonized by annealing at 2000 “Kin oxygen atmosphere (= lo- 6 torr) during w 20 hr. The lithium source was described elsewherer). The tubes were fitted with titanum getter pumps and field emission gauges. At working conditions the pressure of the gases active in adsorption did not exceed 10-“-10-‘2 torr. 3. Experimental results 3.1.
STRUCTURES
OF LITHIUM
FILMS
A series of diffraction patterns for the W(112)-Li system corresponding to successively increasing lithium coverages is shown in fig. 1. These patterns are also presented schematically in fig. 2. The lithium films on W(112) are substantially ordered when the crystal temperature during lithium deposition is as low as 77°K (lithium flux was N lOl2 atoms cm-2 see-’ ). Crystal annealing at 200-300 “K and subsequent cooling to 77 “K causes the intensity of additional diffraction spots to increase by 20-30x. It must be emphasized that at room temperature no additional beams could be observed throughout the coverages used (< 5 monolayers) and lithium deposition results only in an increase of the background intensity. Fig. 1 shows considerable differences in the intensities of the spots located above and below the axis k= 0.This is due to the fact that the (112) tungsten face does not possess a symmetry plane of the (111) type. Because of dynamical effects the substrate asymmetry results in an intensity asymmetry of both normal and additional beams (the latter have comparable intensities at electron energies < 20 eV).
370
V. K. MEDVEDEV,
A. G. NAUMOVETS
Fig. la-lf
AND T. P. SMEREKA
Li ADSORPTION
Fig.
ON
W
371
372
V. K. MEDVEDEV,
A. G. NAUMOVETS
Fig. lm-lr
AND T. P. SMEREKA
Li ADSORPTION
ON w
373
Fig. 2. (a-r) are spot arrangement schemes for the respective patterns of fig. 1. The lower halves of diffraction patterns are shown. The normal spots are shown by dark circles, additional spots by light circles and streaks by dotted lines.
The (112) tungsten plane consists of close-packed atomic rows with troughs in between. As will be shown below, the lithium atoms at first fill the troughs. The second lithium atomic layer falls into troughs existing in the first layer, etc. In accord with this observation we introduce the relative concentrations c.‘ = n”!/n w, where I$! is the absolute lithium surface concenLI tration in the ith layer and n, is the tungsten atomic concentration in the (112) plane equal to 8.2 x 1Or4 cm-*. The first changes in the diffraction pattern are found at the lithium con-
Fig. 1. Diffraction patterns: (a) clean (112) face W; the surface concentration of tungsten atoms nw = 8.2 x 1Or4cm-2. (b-r) are patterns of the W(112)-Li system. The concentration ratio n&w is: (b) 0.25; (c) 0.3; (d) 0.33; (e) 0.42; (f) 0.5; (g) 0.67; (h) 0.71; (i)O.8; (j) 0.87; (k) 1.0; (1) 1.52; (m) 1.7; (n) 1.9; (0) 2.2; (p) 2.6; (q) 3.3; (r) 3.9. The crystal temperature is 77°K; the adsorbed film was annealed at 300°K; the electron energy is 50 eV.
374
V. K. MEDVEDEV,
A, 0. NAUMOVETS
AND
T. P. SMEREKA
centration ~5 x 1013 cmm2. Diffuse streaks appear between the normal reflections along the k-axis, i.e. the [iil] direction. With nLi > 1 x 1014 cmm2 the streaks break up to a series of a dim spots aligned in the same direction. On further adsorption the number of spots diminishes and they become much sharper. At n~i’~2 x 1014 cmT2 (cr =a) a pattern shown in fig. lb is observed in which additional spots with k= _tb_, k$ and split spots at k= &$ are seen (fig. 2b). This pattern corresponds in Wood’s notations) to a lattice p(1 x 4) (fig. 3b). Splitting of the additional beams (h, ++) may be
Fig. 3. (a) is the structure of W(112) face; (b-k) are structures of the lithium adsorbed films; (b) is the first layer ~(1 x4); (c)is the first layer p(1 x 3); (d) is the first layer p(1 x2); (e) is the incoherent structure of the first layer; n~i/nw = 0.75; (f) is the first layer p(1 x 1;) (g) is the second layer ~(1x4); (h) is the incoherent structure of the second layer; (i) is the close-packed second layer; (j) is the third layer; (k) is the fourth layer; (1) is the W(112) along [Till.
Li ADSORPTION
ON
W
315
due to the existence of antiphase domains having the structure p( 1 x 4) and separated along the [Till direction by an odd number of lattice parameters a2 so that antiphase boundaries are parallel to [X10] direction [see, e.g., eq. (45) in the review paper, ref. 41. Progressive lithium deposition causes the fractional order beams of the ~(1 x 4) structure to move, as is shown schematically in fig. 2c, and finally to merge at points k = 5%; 2 3; ) 4; etc. The resulting pattern corresponds to a lattice p(1 x 3) (figs. Id, 2d, 3~). Based on the concept of refs. 5 and 6 this behavior can be explained by formation of a statistical mixture of the p(1 x 4) and p( 1 x 3) unit cells in the interval +< c1 < f_ A simple kinematical analysis (similar to that given in ref. 5, p. 440) supports this conclusion. Within 3 < ci <+ a transition from p(1 x 3) to the p(1 x 2) structure is observed. The nature of this transition is quite different however. The lithium atoms in excess of the stoichiometric concentration ci =3 are not randomly distributed but incorporate to form domains of the p(1 x 2) structure. This follows from the fact that the spots of both structures coexist over a wide coverage range (fig. le). In the coverage interval + cc1 < 1 the fractional order beams of the p( 1 x 2) structure transform to streaks smeared along the h-axis and shift outwards with coverage increase. Simultaneously weaker streaks, which are due probably to double electron scattering, are shifting to k = 0 (figs. 1g, h, i, j). Gerlach and Rhodin suggested’ss) that patterns of this type correspond to one-dimensionally incoherent structures. In this case the film consists of atomic rows arranged in the troughs and randomly shifted relative to each other. The interatomic spacing is the same in all the rows. It diminishes continuously with coverage increase and in the interval +
376
V. K. MEDVEDEV,
A. G. NAUMOVETS
AND
tmn
T. P. SMEREKA
c
0
--/
a
-2
0 0
5
/O
/&/O”C&
‘5
a
Fig. 4. The work function q, adsorption heat q and spot intensity Z versus the concentration n of adsorbed Li atoms on the W(112) face. (a) (1) r&z), CPD method, T=3OO”K; (2) q(n), CPD method T=77”K; (3) p,(n), field emission method l), T=77”K; (4) q(n). (b) Z(n): (1) additional spot (0, -4) of the structure p(1 x 2); (2) normal spot (0, -1); (3) additional spot (0, -$) of the structure p(1 x 4). The letters above mark the coverages corresponding to the various parts of fig. 1.
p(1 x 1). The next diffraction pattern, which is apparently associated with the second layer, has rather bright additional beams at k= A$, less bright at k= At and still less bright ones at k= ++ (h integer) (fig. 11). This pattern exists in a wide coverage range (see in fig. 4b the spot intensity dependence on nri). Assuming the multiple electron scattering we attribute this pattern to the structure p(1 x 4) in the second layer. The second layer atoms are believed to fall into troughs formed by the first layer. At c2 M 0.5 the k = 3 spots undergo some broadening in the [Till direction and at c z ~0.7 to 0.75 they practically disappear (fig. lm). At c2 20.7 the k= k$ spots begin to spread into streaks (of nonuniform intensity) moving to k= _t 1 with ca increasing; the k= k$ spots (due to double scattering) spread into similar streaks moving to k = 0 (fig. 1n). Thus t : second layer atoms at c2 > 0.75 also form one-dimensionally incoherent structures. Unlike
LiADSORPlTONON
the first layer the streaks
in the second
377
W
layer patterns
do not collapse
with
the normal k = f 1 spots. Instead they stop at c2 x 0.9 when the shortest interatomic spacing, within the error limits (several percent), equals the atomic diameter in lithium metal (fig. 3i; for simplicity, a model of the structure with rows in phase is shown, although in reality there is some disorder in the mutual position of the rows since the additional beams in fig. In are streaked along [ilo]). After that rather monotonous patterns are observed which gradually replace each other during the lithium deposition. As judged by doses of lithium, the patterns correspond to the third (figs. lo, p) and fourth (figs. lq, r) layers. In the third layer the lithium adatoms form a centred rectangular lattice (fig. 3j) with the atomic spacing along [iil] somewhat larger than in the close-packed second (fig. 3i) or in the fourth (fig. 3k) layer. The fourth layer lattice is again simple rectangular with the atomic spacing along [Till equal to the lithium atomic diameter (the interpretation of patterns fig. lo-r is also based on the double scattering model). It must be emphasized that the formation of the third and the fourth layers is accompanied by a rearrangement of underlying layers whereby all the film gets a new equilibrium structure. This can be deduced from the fact that in the diffraction patterns from the films 3 or 4 layers thick the spots characteristic of the second or third layer are absent, whereas spots characteristic of the tungsten substrate spacings are clearly visible. The rearrangement of the film is taken into account in drawing the models of the third and the fourth layers (figs. 3j, k). After the completion of the fourth layer no other ordered structures were observed. With subsequent lithium deposition, all spots weaken rather fast so that for the six-layer thick film they are hardly seen in the increased background. Under these conditions the ordered film cannot be obtained even after a careful annealing. 3.2. WORK
FUNCTION
The dependence of the work function c~ on absolute lithium coverage on W(l12) is shown in fig. 4a. Curves 1 and 2 were obtained by the CPD method at crystal temperatures 77 and 300”K, and curve 3 by the field electron emission method at 77”Kl). Note the close agreement of the data taken by different methods and the small influence of temperature on the curve Cp(%I. The intensities of spots of the p( 1 x 2), p (I x l), p (1 x 4) structures versus nLi are shown in fig. 4b. Bends and the minimum in the p(nLi) curve definitely correlate with the formation of these structures. The work function changes effectively terminate when the formation of the close-packed second layer
378
V. K. MEDVEDEV,
A. 0. NAUMOVETS
AND T. P. SMEREKA
is complete. It is important to mention that the structure analysis the concentration scale of the cp(11~~)curve calibrated in ref. 1.
confirms
3.3. HEAT OF ADSORPTION To determine the heat of adsorption q we have measured the equilibrium work functions corresponding to various temperatures and a constant flux N of lithium atoms (fig. 5). The q (Y1Li)curves (fig. 4a) allow this relation
Fig. 5. (1,2) Dependences of the equilibrium work function for the W(112)-Li system on the crystal temperature: (1) N = 9.3 x 1011cm-2 set-l; (2) N = 6 x lOlo cm-2 set-I; (3) adsorption isobar N= 9.3 x lollcmm2 set-‘.
to be converted to an adsorption isobar nLi (T) (fig. 5, curve 3). q(nLi) was calculated from the isobar using a well-known expression from the absolute reaction rate theory”):
4 (nLi) = k,T In T,
where v is a frequency factor and k, is the Boltzmann constant. Independent data on the adsorption heat and frequency factor can be obtained without additional assumptions from two isobars at different lithium fluxes (fig. 5, curves 1 and 2). In this way we obtained vx 1013 set-’ which is consistent with the theoretical value k,T/h, (hp is the Plank constant) for mobile adsorption (ref. 11). The q versus nLi curve is presented in fig. 4a. At , we substituted into the expression for q the concentran LI.>82x10’4cm-2 . tion of lithium atoms in the second layer nlf,)=n,,-8.2 x 1014 cm-*.
Ii
ADSORPTION
ON
379
W
Thereby a natural assumption was made that the evaporation rate of lithium from the second layer is much greater than that from the first one. The initial heat of adsorption obtained here is in good agreement with a value of 2.72 eV obtained in ref. 12 where the field desorption of lithium from tungsten was studied. 3.4. ORDER-DISORDER TRANSITIONS IN
LITHIUM
FILMS
To obtain information about order-disorder transitions in the lithium films we measured the temperature dependences of spot intensities for various structures. In our experiments, only the small central part of a spot was projected onto the photomultiplier input. After deduction of the background intensity, the spot intensities were plotted on a relative scale Z/Z,, where I, is the corresponding intensity at 77°K (fig. 6). It is clear from fig. 6 that practically none of the lithium structures can be observed at room temperature. As in our preceding papers 931% we have investigated the influence I
I
IO
1
f 05
0
Fig. 6. Temperature dependence of spot intensity: (1) additional spot (1, -4) of the structure p(1 x 4); (2) additional spot (0, -4) of the structure p(1 x 2); (3) spot (0, -1) of the structure p(1 x 1); (4) additional spot (0, - $) of the structure p(1 x 9); (5) spot (0, -1) of the clean face. Electron energy is 50 eV, lithium concentrations correspond to the intensity maximum of additional spots. (6) Theoretical behaviour of the spot intensity for Ising’s modelzl). The transition temperature is chosen equal to 230°K.
of the order-disorder transition on the work function. The result is similar, i.e. the work function change accompanying the disordering does not exceed several hundredths of one ev. 4. Discussion
The W (112)-Li system has a great number of different structural states. The symmetry of the two-dimensional lattices indicates a rather complicated interaction between the lithium atoms. In the lattices p(1 x 4), p(1 x 3) and
380
V. K. MEDVEDEV,
p(1 x 2) the lithium
A. G. NAUMOVETS
atoms are arranged
AND T. P. SMEREKA
to have a maximum
spacing
along
the [iil]direction and a minimum spacing (equal to one between the adjacent troughs) along [ilO]. As is clear both intuitively and according to statistical calcul ationss), such structures indicate that there must exist repulsive interatomic forces along the troughs and attractive forces across the troughs. Such an anisotropic interaction seems to be caused by the highly anisotropic substrate structure. It may be assumed that the repulsive forces along the troughs are due mainly to be dipole-dipole interaction. Indeed the dipole moment at low coverages, evaluated as p=Acp/4nnLi (Acp is the work which function change) equals for this system to z 1.7 D. Some contribution is expected to be comparatively small at appreciable dipole moments can come fron long-range interaction via the substrate electron gas14). Along the [ilO] direction the adatoms are separated by rather strongly protruding tungsten atomic rows which may partly screen the fields of the charged lithium atoms and at the same time serve as mediators in their exchange interaction. As a result the chains -Li-W-Li-Ware formed along this direction. It is worthwhile to discuss a possible relation between the lithium film structures and the unsaturated orbitals on the substrate surface. This factor usually is ignored in alkali metal adsorption whereas in other cases the concept has proved to be extremely fruitfu115+ 16). Due to its small radius the lithium atom must be more sensitive to the surface potential relief than are other alkali atoms. Fig. 7 shows schematically the
L-ml
Fig. 7.
Scheme of orbitals on the surface of W(112) face.
unsaturated bonds in the tungsten (112) face. It can be seen that the geometry of the orbitals favours adatom arrangement into linear chains along the [ilO] axis, as i; observed experimentally. From this point of view one also can explain the lack of coherence in the positions of neighbouring rows when the interatomic spacing along [iil] smoothly diminishes ($
Li
Now let us discuss
ADSORPTION
ON
381
W
in more detail the transitions
between
different
first
layer structures. When the interatomic distances along troughs are larger than 2a2, the atomic repulsion is not strong enough to cause incoherence between the film and substrate periods at unstoichiometric Li concentration. That is why the p (1 x 4)+p (1 x 3) transition occurs via mixing of the cells 5s6, and the p (1 x 3)-p (1 x 2) transition by a growth of the p (1 x 2) domains. As could be expected in the coverage range 3 < ci < 4 the measured adsorption heat only weakly depends on the concentration (fig. 4a). The possibility of first order phase transitions in systems with dipole interaction [in our case corresponding to the condensation of excess atoms to produce a more dense phase p(1 x 2)] was substantiated theoretically by Bol’shov17). In the p(1 x 2)+p (1 x 1) transition the film lattice parameter along [iil] diminishes gradually showing an increased role of the repulsive forces at agrees also with a the interatomic distances a,
382
V. K. MEDVEDEV,
A. G. NAUMOVETS
AND T. P. SMEREKA
that ,u,~(I x 1) = + ,u=~(1 x 2). As far as we know, it is a rare case when the mutual adatom depolarization (due not only to Coulomb but also to other types of interaction) compensates precisely the barrier change which could occur due to a concentration increase. One layer having a p (1 x 1) structure is far less than the lithium coverage which, according to Gerlach and Rhodin’s terminologys), is reasonably to be assigned to the physical monolayer. As the second layer is filled, the work function decreases again and reaches a minimum on formation of the structure p (1 x 4) when 8~0.8 (e= 1 corresponds to the physical monolayer which is the close-packed two-layer film). Further packing of the second layer leads to a work function increase and at 8 = 1 it becomes practically constant. In this respect the W(112)-Li system is similar to other related systemsS-10,13,19,30
>*
The work function is weakly dependent on long-range order (see section 3.4). The barrier properties seem to be determined rather completely by short-range order in the adatom arrangement. This behaviour may result from a small coherence width of electrons overcoming the surface barrier. Proceeding to disordering of the lithium films, let us note that in the first layer the same tendency is observed as in other related systems g*13): temperature of disordering increases with the film packing indicating an increased energy of atomic interaction. Fig. 6 shows a theoretical curve according to Ising’s two-dimensional model 21) I = 1,(1 - T/T,)‘, where Tc is the disordering temperature which for the sake of illustration is chosen to be 230°K. From fig. 6 it is clear that this relation does not agree with the experiment. This could be caused by the fact that the theory neglects long-range interaction which in our case is rather significant. Indeed, for the system W (100)-H, for which the long-range interaction is less essential, the experiment agrees quite well with Ising’s modelz2). It must be noted that the structures of the third and forth layers disorder at rather low temperatures (- 150-180°K). These values are far less than the melting point of lithium metal (453°K). It has been shownss) that such a strong lowering of the melting temperature in thin films is associated with the essential contributions from the free energy of the surface and the interface. It is interesting to compare the properties of lithium and sodiumlo) films on the W (112) face. Similar structures are formed for both systems, although owing to a larger atomic diameter the structure p( 1 x 1) cannot be realized in the sodium film. It seems that for this reason the work function
Li ADSORPTION
minimum
for the W (112)-Na
layer, as in the case of lithium,
ON
383
W
system is not achieved but on formation
on filling
the second
of the first close-packed
layer (in this case coverage is also less than the physical monolayer). At the same time the behaviour of Na, K and Cs films on the other substrate with the trough structure - the Ni (110) face - is essentially different. In particular, at low coverages not every trough contains adatoms. Gerlach and Rhodin concluded that in these films the repulsive interatomic forces act along the troughs as well as across them being less along the troughs. Further investigations are necessary to elucidate the cause of this difference. 5. Conclusion The results presented demonstrate the complicated anisotropic character of lithium atomic interactions on a substrate with a strong potential relief anisotropy. Due to their small size the lithium adatoms are far more sensitive to the atomic structure of the adsorbent than other alkali adatoms. That is why for lithium the variety of realized structures is richest and differences between adatom electronic states in various structures are greatest as evidenced by the peculiar work function changes.
References 1) V. M. Gavriliuk and V. K. Medvedev, Fiz. Tverd. Tela 8 (1966) 1811. 2) V. M. Gavriliuk, A. G. Naumovets and A. G. Fedorus, Zh. Experim. Teor. Fiz. 51 (1966) 1332. 3) E. A. Wood, J. Appl. Phys. 35 (1964) 1306. 4) P. J. Estrup and E. G. McRae, Surface Sci. 25 (1971) 1. 5) L. H. Germer, J. W. May and R. J. Szostak, Surface Sci. 7 (1967) 430. 6) G. Ertl and J. Kuppers, Surface Sci. 21 (1970) 61. 7) R. L. Gerlach and T. N. Rhodin, Surface Sci. 10 (1968) 446. 8) R. L. Gerlach and T. N. Rhodin, Surface Sci. 17 (1969) 32. 9) A. G. Fedorus and A. G. Naumovets, Surface Sci. 21 (1970) 426. 10) J. M. Chen and C. A. Papageorgopoulos, Surface Sci. 21 (1970) 377. 11) S. Glasstone, K. J. Laidler and H. Eyring, The Theory of Rate Processes (McGrawHill, New York, 1941). 12) V. K. Medvedev, Izv. Akad. Nauk SSSR Ser. Fiz. 33 (1969) 528. 13) V. K. Medvedev, A. G. Naumovets and A. G. Fedorus, Fiz. Tverd. Tela 12 (1970) 375. 14) T. B. Grimley and S. M. Walker, Surface Sci. 14 (1969) 395. 15) Z. Knor and E. W. Muller, Surface Sci. 10 (1968) 21. 16) P. W. Tamm and L. D. Schmidt, J. Chem. Phys. 51(1969) 5352. 17) L. A. Bol’shov, Fiz. Tverd. Tela 13 (1971) 1679. 18) J. Matthews, in: Epitaxie-Endoruxie (VEB Deutscher Verlag fur Grundstoffindustrie, Leipzig, 1969) p. 135. 19) R. L. Gerlach and T. N. Rhodin, Surface Sci. 19 (1970) 403. 20) D. A. Gorodetsky, Yu. P. Mel’nik and A. A. Yas’ko, Ukr. Fiz. Zh. 12 (1967) 649. 21) M. E. Fisher, Rept. Progr. Phys. 30 part 2 (1967) 615.
384 22) P. J. Estrup,
V. K. MEDVEDEV,
A. G. NAUMOVETS
AND
T. P. SMEREKA
in: Structure and Chemistry of Solid Surfaces, Ed. G. A. Somorjai (Wiley, New York, 1969) p. 19-1. 23) N. T. Gladkich, R. I. Zaichik, V. P. Lebedev, L. S. Palatnik and V. I. Khotkevitch, in: Poverkhostnaya Difiziya i Rastekaniye (Nauka, Moscow, 1969). p. 222.