INFLUENCE OF TECHNOLOGICAL FACTORS ON THE STRUCTURE OF FILMS OBTAINED BY EVAPORATION IN VACUUM

A. N. RUDNEV, A. A. KHERUMENKO

(Presented by the scientific seminar of the Research Institute of High Voltages)

The properties of thin films obtained by evaporation in vacuum generally differ from the properties of these same materials in monolithic form [1]; however, the literature [2] contains information on thin metallic films obtained, some of whose properties repeat the properties of monoliths. The main factor determining the properties of thin deposited films is their structure. A crystalline or amorphous film structure is distinguished. A crystalline structure is characterized by the size of the crystals, their mutual orientation, and the continuity of the film at its various thicknesses. We shall characterize an amorphous structure by the continuity of the film. Depending on the deposition regime, one or another structure of the film is obtained. The structure of metallic films most often proves to be crystalline [3], whereas dielectrics may have both amorphous [4] and crystalline [5] structures.

The magnitude of the adhesion of a deposited film to the substrate is also of known interest.

The process of film formation on a substrate and its structure depend on a whole series of factors. An atom arriving at the surface of the substrate possesses a certain kinetic energy, which causes it to move over the surface of the substrate. The character of the atom’s motion over the substrate surface depends on the potential function of this surface, which in the ideal case consists of periodically recurring potential “wells” [3, 253]. The potential function of the substrate surface and the kinetic energy of the atom determine the surface mobility of the atom. The collision of an atom moving over the substrate with another atom of the same kind may lead to the formation of an atomic pair, whose mobility decreases, since the energy required for their union is absorbed [1; 6]. With sufficient supersaturation of the vapor of the condensing material, triple nuclei may arise on the surface of the substrate. With still greater supersaturation, four-atom nuclei may arise. In the initial stage of film growth, doublets, triplets, etc., may serve as centers of crystallization. In the general case these centers are distributed nonuniformly over the surface of the substrate, creating islands of the condensing material. The energy required to evaporate a metal atom from the surface of a clean substrate and from the surface of a metal is different. If the former is greater than the latter, a stable monatomic layer is obtained [3, 252]. However, in the majority of cases there occurs pro-

the process of island formation [1; 3, p. 257]. The nucleation of crystallization centers occurs only in the initial stage of condensation and is completed within a short time.

The importance of the influence of “surface mobility” on the structure of a film is obvious. In turn, surface mobility depends on many factors. Levinstein [7] showed that the velocity of atoms arriving at the substrate usually does not affect the structure of the film. However, this question is still a matter of debate [1; 8; 9]. The velocities of the atoms of the evaporated substance are determined by the temperature of the evaporator [4]. Dielectrics should be evaporated without their chemical decomposition, by correctly choosing the maximum temperature of the evaporator. With rapid evaporation of SiO, dense, partially reduced films are obtained. With slow evaporation \((\leq 15 \text{ Å/sec})\), the films are porous, but of more uniform composition [10]. Denser films contain more silicon than is present in the initial SiO. This is confirmed by electron-diffraction studies [4].

As the evaporation rate is increased, the mechanical stresses in SiO films change from tensile to compressive [11]. Consequently, it is possible to select an evaporator temperature at which the stresses will be minimal.

The surface mobility of particles is substantially affected by the temperature of the substrate. As it is increased, mobility rises. This leads to an increase in the number of crystallization centers and to more ordered growth of crystals and uniform growth of the entire film [12, 13]. However, as the substrate temperature is increased, the structure of metallic films becomes more coarse-grained [14, 15]. A high substrate temperature promotes evaporation from its surface of contaminants adsorbed on it and prevents the deposition of particles of the residual atmosphere in the vacuum as a result of the process of secondary evaporation, thereby increasing the adhesion of the film to the substrate. Heating the substrate (annealing) and the presence of oxygen atoms in the vacuum chamber promote the formation of an oxide layer between the film and the substrate, which also increases the adhesion of the film [16]. The difference between the temperature coefficients of linear expansion of the film and the substrate inevitably leads to the appearance of mechanical stresses in the film. The magnitude of the stress in the film may be sufficient for macroscopic cracks to appear in it. Thick dielectric films therefore tend to peel off [10]. Heating SiO films on aluminum mirrors to 200°C leads to a considerable decrease in the mechanical stresses in them [11].

The magnitude of the microscopic irregularities of the substrate surface affects the surface mobility of atoms and, consequently, the structure of the film, since the character of the potential function of the substrate surface is determined by its roughness. To reduce the magnitude of the microscopic irregularities of a glass substrate, a thin layer of SiO should be deposited on its surface [17, 18].

The uniformity of the structure of deposited films depends on the type of evaporated material and the degree of its purity. The formation of an oxide layer is determined by the affinity of the evaporated material for oxygen. For electrophysical studies of deposited films, materials of high purity are required, since impurities give rise to various defects in the structure of the film. Impurities in the film may also arise because of contamination of the evaporator. Preliminary heating of the evaporator in vacuum reduces contamination of the evaporated material...

of the evaporator. However, a small amount of material from the evaporator, for example in the form of oxides, can enter the structure of the film [19].

The design and material of the evaporator also affect the amount of impurities in the film.

Vacuum strongly affects the structure of deposited films. The presence of vacuum-oil vapors in the chamber causes the formation of additional defects in the film structure. At a high substrate temperature, carbides may form in the film structure as a result of carbon being released from decomposing vacuum-oil molecules [4]. The time of formation of a film of residual gases on the substrate surface is also determined by the degree of vacuum [20]. This film dissolves in the condensing material, causing looseness of its structure, and worsens adhesion [1]. In addition, the partial pressure of residual oxygen in the chamber affects the oxidation of SiO films and metals [21].

An important factor is the degree of cleanliness of the substrate surface. There are many methods for cleaning a substrate. The most effective appear to be ultrasonic cleaning [22; 23; 24], flame polishing in a flame of pure hydrogen [26], heating in vacuum [3; 22; 26], and cleaning in vapors of isopropyl alcohol [10]. The method of cleaning the substrate by ion bombardment gives a high degree of cleanliness, but increases the roughness of the substrate. Quantitative evaluation of different cleaning methods is difficult, because the slightest differences in procedure or apparatus lead to different results [1].

Thin metallic films, in terms of their sublimation structure, are far from equilibrium. They consist of relatively perfect crystals of small size, grown from individual nuclei [27]. The independence of nucleation and growth of thin crystals leads to their strong mutual disorientation and to defects in the boundaries that form. This is explained by the fact that the rate of influx of atoms is so high that new layers nucleate before the formation of the preceding ones has time to be completed. Under such conditions, a large number of defects is quite probable. For example, when two steps moving toward each other, originating from neighboring nuclei, come into contact, twins, dislocations, packing defects, and other lattice imperfections may arise, which can substantially worsen the structure of the film. The presence in the film of atoms of another material may be regarded as the appearance of additional crystallization centers, which strongly influence film growth [28].

The presence of through holes in the film depends on the presence, in the stream of evaporated material, of macroparticles whose energy is sufficient to mechanically pierce through the already-deposited film. Evaporation of macroparticles is caused by the design of the evaporator (boats and spirals). A design of evaporator has been developed that makes it possible to eliminate the presence of such macroparticles in the stream of evaporated material [29].

Other factors affecting the structure of a film include the following. The crystallographic order of the plane onto which deposition is carried out affects the structure of the condensed film [1; 30]. Along with the orienting influence of the substrate surface, the structure of the nucleus affects the structure of condensate films. Three-dimensional nuclei lead to the formation of crystals with the [111] plane parallel to the substrate, while four-dimensional ones lead to the [100] plane, also parallel to the substrate.

Oriented growth of a film on a substrate is possible if the substrate is single-crystalline and the difference in their lattice constants does not exceed 10–15%. When Ge is deposited on an amorphous substrate with tem—

with a temperature up to 370°C an amorphous condensate is formed, whereas when it is deposited on a ZnS cleavage, a crystalline condensate is obtained already at 250°C, since the lattice periods of Ge and ZnS are close (5.63 Å and 5.42 Å). An aluminum film deposited on NaCl becomes single-crystalline at 300°C, although the ordering process begins already at 250°C [13; 1].

For electrophysical studies of films, the greatest interest is a fine-grained single-crystal structure with the least defectiveness. This imposes special requirements on the technological regime. The rate of condensation of the substance must be sufficiently high (hundreds of Å per sec). Proceeding from the degree of contamination, adhesion, and mechanical stresses in the film, the grain size of the film structure, and the formation of oxides in it, the substrate temperature must be chosen as optimal. The maximum temperature of the evaporator must be such as to exclude thermal decomposition of the dielectric being evaporated.

REFERENCES

1. Behrndt, “Vacuum,” 1963, 13, No. 9, pp. 337—347.
2. Reynolds and Stilwell, “Phys. Rev.,” 1952, 88, p. 418.
3. L. Holland. Deposition of Thin Films in Vacuum. GEI, Moscow–Leningrad, 1963.
4. Hass, “J. Amer. Ceram. Soc.,” 1950, 33, p. 353.
5. Wolfgang, “Naturwissenschaften,” 1963, 50, No. 17, 566.
6. Frenkel, “Z. Phys.,” 1924, 26, p. 117.
7. Lewinstein, “J. Appl. Phys.,” 1959, 20, p. 306.
8. Aziz and Scott, “Canad. J. Phys.,” 1956, 34, p. 731; 1951, 35, p. 985.
9. Holland, “Canad. J. Phys.,” 1957, 35, p. 697.
10. Bruce and Balmer, “Electro-Technology,” 1960, 66, No. 5, pp. 157—159.
11. Novice, “Brit. J. Appl. Phys.,” 1962, 13, No. 11, pp. 561—563.
12. Sennet and Schott, “J. Opt. Soc. Amer.,” 1950, 40, p. 203.
13. Thun and Hass, “Vac. Symp. Trans.,” 1958, 58. Pergamon Press. Sondon (1959).
14. Hass and Scott, “J. Opt. Soc. Amer.,” 1949, 39, p. 179.
15. Reimer, “Z. Naturforsch.,” 1956, a, p. 649.
16. Benjamin and Weaver, “Proc. Roy. Soc.,” 1961, A261, p. 516.
17. Behrndt and Maddocks, “J. Appl. Phys.,” 1959, 30, p. 276.
18. Bertelsen, “J. Appl. Phys.,” 1962, 33, p. 2026.
19. V. Kohl. Technology of Materials for Electrovacuum Devices. GEI, Moscow–Leningrad, 1957.
20. Gert, “Ann. N. J. Acad. Sci.,” 1963, 101, No. 3, pp. 722—755.
21. York, “J. Electrochem. Soc.,” 1963, 110, No. 4, pp. 271—275.
22. Putner, “Brit. J. Appl. Phys.,” 1956, 7, p. 332.
23. Koontz and Amron, “A. S. T. M. Special Techn. Publ. No.” (1958), 22, “Ultrasonic News” (3), (1959), 8.
24. Olaf, “Acustica,” I (1960), 253.
25. Nielson, “Vac. Symp. Trans.,” (1960), 253.
26. Fray and Nielson, “Brit. J. Appl. Phys.,” 1961, 12, p. 603.
27. Reports of the USSR Academy of Sciences, 1963, vol. 1, II No. 3.
28. Zehender, “Optic,” 1950, 7, p. 200.
29. Vergara, Greenhouse and Nicholas, “Rev. Sci. Instrum.,” 1963, 34, No. 5, pp. 520—522.
30. Mayer, “Physik duenner Schichten,” 2, Wissenschaftliche Verlagsgesellschaft, m. b. H. Stuttgart (1955).