1966, No. 3/4 87 Microfractography of thin films J. M. Nieuwenhuizen and H. B. Haanstra 620.187.5 The special structure of evaporated thin films, which,are of particular interest to the electronies industry at the present time, has given rise to the concept of a fourth state of aggregation of the material. A surprisingly direct insight into this state has now been obtained from ''fractographs'' - photographs of fracture surfaces - which have been made, by means of the electron microscope.for aluminium films about 1micron thick. " Thin films of materials of various kinds showeffects and properties which do not appear, or are much less pronounced, in the bulk material. Thin films are therefore the subject of intensive investigation, and their properties have led to important applications. For several decades they have been used for example as getters in thermionic valves, and for coating glass surfaces to give them particular reflecting (or nonreflecting) properties. More recently, thin films have found application in electronic circuit components, especially for digital techniques [11, and the possibilities offered by thin film components in integrated circuits are of considerable practical importance. The thin films concerned are gener~lly produced by evaporation on to a substrate or carrier in -a vacuum, and their thickness may be anywhere between a few to many thousands of atoms. In the latter case the thickness of the film may be of the order of 1 micron. These relatively thick films can sometimes be given considerably different characteristics if in the evaporation process the molecular beam is directed not vertically upon the substrate but more or less obliquely. Obliquely deposited films of semiconductors such as silicon, gallium arsenide or tellurium, show a very marked photovoltaic effect, which is virtually absent in vertically deposited films. Obliquely deposited films of some substances are dichroic, vertically deposited films of the same substances are not. In obliquely deposited magnetic layers, for example of permalloy (80Ni20Fe), a strong magnetic anisotropy may occur, in which' - at least if the angle of incidence is not made too great - the preferred direction of magnetization is perpendicular to the plane of incidence of the atomic beam. The magnitude of the mechanical J. M. Nieuwenhuizen and H. B. Haanstra are with Philips Research Laboratories, Eindhoven. stresses (if any) may also depend on the angle from which the film is evaporated [21. All these differences can only be explained on the assumption that for different evaporation angles films of different structure are formed. It has now proved possible to make these differences in structure directly visible with the electron microscope. In the following we shall briefly describe the procedure and show some results. For "thin" thin films, where the processes 'of interest are nucleation and the growing together of islands of the evaporated material to form an unbroken coating, the electron microscope has previously been employed with success as a research tool [31. In this case the films themselves are placed as specimens in the electron beam of the microscope. The "thick" thin films with which we are concerned here, however, are not sufficiently transparent to electrons. As with bulk metals, only the surface can be made accessible to electronmicroscopie investigation, viz. by making a transparent replica. Conclusions about the internal structure of the film can then only be drawn in so far as this structure appears in relief on the surface. This information is very limited and inadequate, [1] See W. Nijenhuis and H. van de Weg, Developments in the field of electronic computers during the last decade, Philips tech. Rev. 26, 67-80, 1965. [2] On the photovoltaic effect, see: B. Goldstein and L. Pensak, J. appl. Phys. 30,155, 1959; E. 1. Adirovich, V. M. Rubinov and Yu. M. Yuabov, Sov. Phys, Solid State 6, 2540, 1965 (No.10). On dichroism: A.Kundt, Ann. Physik u. Chemie27, 59,1886. For magnetic anisotropy: D. O. Smith, M. S. Cohen and G. P. Weiss, J. appl. Phys. 31, 1755, 1960... Fot mechanical stresses: J. D. Finegan and R. W. Hoffrnan, J. appl. Phys. 30, 597, 1959. [3] G. À. Bassett, J. W. Menter and D. W. Pashley, in: Structure and properties of thin films, Proc. int. Conf. Bolton Landing, New York 1959, page 11; J. van de Waterbeemd, Physics Letters 16, 97,1965 (No. 2); J. van de Waterbeemd, Philips Res. Repts. 21, 27, 1966 (No. I).
PHILlPS TECHNICAL REVIEW 88 and we therefore looked for a technique in which a cross-section of the film could also be reproduced in the replica. It is not sufficient just to break off a piece of substrate with its film, for if a replica of such a piece is made that extends over the surface and the fracture face, the part in which we are interested is to be found at the edges of the replica, which usually curl up and are not so suitable for good observations. This can be avoided by applying an intermediate film which is afterwards dissolved [4J, but the finest details are then likely to be less distinct, and there is the possibility that spurious details may be introduced by the structure of the intermediate layer. VOLUME 27 Etching away parts of the film before making the replica did notproduce useful results either, the transition from the original surface to the substrate was then too gradual; cross-sections of the layer providing clear information about the internal structure are not obtained in this way. A method that did produce useful results was to make a scratch in the thin :film with the point of a razor blade. On each side of the scratch the :film crumbles away here and there; fragments of the film become detached from the substrate, and the piece of film that remains shows a well-defined fracture surface, extending from the surface of the film to the substrate. ti Fig. 1. When the surface of a thin film is scratched with the point of a razor blade, a fairly clean fracture surface b is obtained at some places. A continuous replica can now be made of the film surface I, the fracture surface and the exposed surface s of the substrate. The step-like replica can be detached without damaging it. Fig. 2. Electron-micrograph of a replica of an aluminium film about I micron thick, obtained by the method described. The picture of the fracture surface shows that the aluminium film is built up from parallel crystal columns, inclined at a certain angle. It was verified from stereornicrographs made of all the specimens that the replica had retained the form illustrated in fig. 1.
,1966, No. 3/4 MICROFRACTOGRAPHY OF THIN FILMS 89 At such places carbon can be deposited in the usual way to make a replica. A continuous replica is then obtained of the surface of the :film, the fracture surface and the exposed surface of the substrate; seefig. 1. A difficulty with such a replica, which contains a kind of step with two sharp bends, is that the sharp edges easily break, and to prevent this from happening the carbon layer is removed extremely slowly. In this investigation we worked mainly with thin films of aluminium evaporated on glass at room temperature. The aluminium film under the replica can be removed by dissolving it in very dilute hydrofluoric acid (0.05 % HF in water); the replica then comes away from the glass at the same time. The dissolving process may in some cases last a whole week, but the result is that the replica remains intact [5]. Electron-photomicrographs of such replicas now give clear pictures of the fracture surface of a thin :film. A "microfractograph" of this type is shown in fig. 2. The step in the replica is clearly visible (although the edge of the step was not straight here, as in normal steps, but a zigzag line). The micrograph shows that the film is built up from columnar crystals, with parallel longitudinal axes inclined at a certain angle to the normal to the substrate. To learn more about this angle, which we shall call CPk and which is obviously of importance in the structure of the film, we made use of the following procedure. It is usual to make the relief structure of electronmicroscopie specimens more visible by shadowing, that is to say by evaporating a heavy metal such as platinum on to the specimen from an oblique angle: at places where no platinum atoms have been deposited, the electrons in the beam pass through the specimen with relatively little scattering, and at such places the micrograph shows a high density. We carry out this shadowing procedure for our evaporated thin films in the same vacuum vessel in which the films were produced, and we evaporate the Pt atoms on to the scratched aluminium :filmfrom exactly the same angle (in fact from the' same source location) as used for the Al atoms of the substrate, moreover making sure that when the :filmis scratched [6] the position of the substrate in the vacuum vessel is not changed, The length of the resultant Pt shadow thus as it were marks the angle cpa from which the aluminium is evaporated on to the substrate. It follows immediately from the fact that the Pt shadow in the photograph in fig. 2 does not have zero length that the columnar crystals of the layer have not grown exactly in the direction of the incident aluminium atoms (see fig. 3): they are more upright (cpk < cpa). The columns are however always in the plane of incidence of the Al atoms; we have been able to I k I a~' Fig. 3. The columns of the film are at an angle 'Pk to the normal to the substrate. This angle proves to be smaller than the angle (/la at which the beam of aluminium atoms (and the beam of platinum atoms in the shadowing process) is incident on the film. The columns are however in the plane of incidence of the atoms. The length of the Pt shadow is a minus k. establish this in experiments with five different angles of incidence, up to about 80. The angle CPk is easily calculated from the known cpa and the lengths k and a indicated in fig. 3, which can be measured on the micrograph. The values thus found are plotted in fig 4 against cpa. The relationship can be represented very well by the equation [7]: tan cpa = 2 tan CPk. 900'.------------r---------- ~-----------, i [4] R. Ya: Berlaga arid M. I. R~denok, Sov. Phys. Solid State 3, 458,. 1961. [5] Mrs. J. Andreas-Bosdijk did a large part of the experimental work for the development of this technique. ' [6] The reader will notice that the shadowing is done before the, carbon replica is made. When the carbon layer is detached, the deposited Pt atoms adhere to it. [7] Our attention was drawn to this by Drs. G. W. van Oosterhout. 60 '-% 'Fig. 4. Relationship found experimentally between the angles 'Pa and 'Pk of the direction of evaporation and the orientation of the columns. The relation is given to a very good. approximation by the curve shown which represents the equation: tan 'Pa = 2 tan (/Ik.
90 PHILlPS TECHNICAL REVIEW VOLUME 27 Fig. 5 shows another clear micrograph of a fracture surface. These and similar micrographs show details in the columns that provide some foundation for a probable explanation of the mechanism underlying the growth of the columns; we shall not, however, go experiment, incidentally, provides conclusive proof that the columnar structure in micrographs such as those in figs. 2 and 5 cannot be a result of the method of preparation itself - a possibility that couid not a priori be excluded.) To obtain a pronounced picture Fig. 5. A particularly detailed fractograph obtained by the method described. into this subject here rsi, Finally, fig. 6 shows the fracture surface of a composite film, built up by successive evaporation of aluminium from two directions. In the second part of the film the columns are accordingly inclined in a different direction, giving the cross-section a kind of herringbone structure. (This of the herringbone structure, the second layer must be given the chance to nucleate independently, e.g. on an amorphous oxide layer, produced by admitting air for a moment into the vacuum vessel betweyn the two evaporation processes. If this is not done, the columns do not abruptly change their direction of growth, and
1966, No. 3/4 MICROFRACTOGRAPHY OF THIN FILMS 91 Fig. 6. Fractograph of a film grown by the successive evaporation of aluminium from two directions. Air was admitted into the vacuum vessel between the two evaporation processes. the change in the fracture surface is brought out much less distinctly. The information supplied by the structure patterns obtained in the fractographs takes us a stage nearer to an explanation of the relationship, which we mentioned at the beginning of this article, between various properties of thin films and the direction from [8] See: C. Kooy and J. M. Nieuwenhuizen, Structural effects in thin films observed by electron microscopy of the film cross-section, shortly to be published in Proc. ColI. on basic problems of the physics of thin films, Clausthai 1965. which the films are evaporated. This information can moreover contribute towards a better technological control of evaporated films. Summary. The characteristics of some evaporated thin films depend to a marked extent on the angle at which the atoms arrive at the substrate. Replicas of fracture surfaces of relatively thick films of aluminium (thickness approx. I micron) have been made successfully by means of a simple technique. Photographs of these taken with the electron microscope show clear pictures of the cross-section of such a film. From these "fractographs" conclusions can be drawn concerning the structure of the films and the way in which the structure is influenced by the evaporation angle.