US 3316121 A
Abstract available in
Claims available in
Description (OCR text may contain errors)
April 25, 1967 B. LOMBOS ETAL 3,316,121
E PITAXIAL DEPOSITION PROCESS Filed Oct. 7:, 1965 3 Sheets-Sheet 1 POWER SOURCE BELA LOMBOS and THOMAS RAPHAEL SOWGYI y W, W M Attorneys April 25, 6 B. LOMBOS ETAL 3,316,121
EPITAXIAL DEPOS ITION PROCESS Filed Oct. 5-2, 1963 3 Sheets-Sheet P;
W W/ M 2 M; A
V DIRECTION OF GAS FLOW RUN NO.| STANDARD DEVIATION =04,
RUN NOZ STANDARD DEVIATION =O-4JL BROWN ISH DEPOSIT BELA LOMBOS and THOMAS RAPHAEL SOMOGYI By W Dra W eys April 25, 1967 B. LOMBOS ETAL 3,
EPITAXIAL DEPOSITION PROCESS Filed Oct. 2 1963 3 Sheets-Sheet 8 I700 f 9 cd b a TOK RESISTANCE RF HEATING HEATING A E 60 kCGI/IIIOIQ G d AE= 37 kcoI/mole b e I400 AE= 24 kcoI/mole c f I I I I I I 0 2 4 6 8 IO [2 I4 DISTANCE IN CM BELA LOMBOS and THOMAS RAPHAEL SOMOGYI United States Patent Ofiice 3,3121 Patented Apr. 25, 1967 This invention relates to a method and apparatus for the epitaxial growth of crystals and more particularly concerns the epitaxial deposition of silicon from a vapour onto a silicon substrate.
As is known, the epitaxial deposition of a substance onto a single crystal substrate of the same substance involves the oriented overgrowth on the substrate so that its surface provides through its lattice structure, preferred orientation for the deposited material. Epitaxial growth of semi-conductor crystals is of great commercial interest and utility in the electronics field.
In the past, difiiculty has been encountered in epitaxial deposition processes since the growth conditions of each specimen have to be carefully watched and controlled if uniform thicknesses of deposition on a single specimen and from one to the next are to be obtained. In particular, for example, when a number of specimens are prepared in a single reaction tube, or vessel in a moving vapour stream by the reduction of silicon tetrachloride by hydrogen, the depletion of the silicon tetrachloride from the gaseous phase leads to excessive deposition on the substrate specimens first encountered and insufficient deposition on those encountered later by the gas phase. To some extent, this trouble has been mitigated in trying to keep the concentration of reactants and products constant by arranging for the inlet of gas mixture at different locations in the vessel. It is not, however, entirely satisfactory and is extremely complex to achieve properly.
It is an object of the present invention to meet the difficulties of the prior art, and there is provided a method of epitaxial deposition of a solid from a gaseous phase comprising the steps of, controlling the equilibrium constant of the reaction so that when depletion of the reactants tends to alter the reaction rate, the equilibrium constant is adjusted to maintain a constant rate of deposition.
Now, Miller et al. report in the Journal of the Electrochemical Society, 109, p. 643 (1962), that the free energy change for the reaction:
(over the range 298-2000" K.)
Thus, we may deduce that the equilibrium constant K of the reaction will be given by:
It will be seen therefore that log K and hence K increases with increasing temperature within the practical range for single crystal growth. Now as mentioned above, in a uniform temperature reaction tube the yield of the reaction decreases in the direction of the gas flow because of changes in the concentrations of reactants and products. When hydrogen is in a very large excess over the incoming SiCl its concentrationmay be regarded as substantially constant. Qualitatively, then, there occurs a gradual drop in the mole fraction of SiCl, accompanied by a rise in the mole fraction of HCl as the gaseous mixture travels along the tube, resulting in a decreasing deposition rate. -We find that if the equilibrium of the reaction can be altered by adjusting the temperature profile of the furnace according to the desired change in equilibrium the deposition rate along the reaction tube can be kept substantially constant. Since the equilibrium constant increases with the temperature, the usable portion of the furnace is determined by the melting point of silicon on the one hand and the lowest temperature for single crystal deposition of satisfactory qualtity on the other hand (approx. 1400-1700 K. for silicon). We find that the flow rate has an effect on this applicable length too.
After a discussion of the theory, a description of an apparatus for carrying out the invention will be made and reference will be had to the drawings in which,
FIGURE 1 shows a partly schematic arrangement of an apparatus constructed in accordance with the teaching of the invention,
FIGURE 2 shows a view in detail of the furnace in FIGURE 1.
FIGURE 3 is a plan view of epitaxially deposited samples showing numerically the thickness of deposit,
FIGURE 4 is a series of graphs of distance alongthe reaction tube against temperature for various values of AB.
The evaluation of the temperature gradient for a given flow rate using a tubular reaction chamber can be made in the following way:
The deposition rate (say) can be expressed in the form where or is an experimental parameter involving vapor saturator temperature, geometry of apparatus etc. (but not x or T), x is the distance along the direction of gas how, and T is the absolute temperature.
tip-20 da aw+ dT where or, are variables, independent of each other and of x and T.
The condition for uniform deposition is:
2 0 ln f H J;
@T at Finally, integrating Equation 3:
T a in y o- 111 f :60
where R is the gas constant and AE the activation energy. When this equation is integrated without fixing upper and lower limits, and therefore introducing an arbitary constant of integration, we have (assuming AE contant) Now, if all parameters except temperature are fixed, Equation 1 will read:
i.e. g may be substituted for [L in Arrhenius equation. Hence, the integrated form will become:
AE=60 kcal./mole AE=37 kcal./mole AE=24 kcal./mole respectively +const.
We also determined the function in f(x) experimentally by employing both a resistance heated and an RF. heated furnace using a constant temperature range and obtained two different results.
To obtain the function ln f(x) slices of silicon were placed in a uniform temperature section of a reaction tube at 1215 C. and the deposition rate as a function of distance was measured. A similar experiment was conducted in an R.-F. heated reaction chamber resulting in a different set of values for In f(x). The two sets were fitted to curves assuming a linear relationship between In ,u. and x. The functions representing those curves were found to be:
. ln =6t5307 1O -2.4804 l x (Standard Deviation=3.9l X
for the resistance heated tube, and
In ,(L=1.5048-5.2600' 10" x (Standard Deviation: 3 .5 42 10- for the RF. heated tube.
These results combined with the above two values for the activation energy give a total of six distinct curves (a to f) for the temperature profile (see FIG. 4). In practice, the most uniform deposition with our resistance heated furnace was achieved with a temperature gradient closest to curve a of FIG. 4.
An apparatus for obtaining substantially uniform deposition of silicon will now be described having reference to FIGURE 1. A furnace 1 heated by source 10 and in a manner which will be described later includes a reaction tube 2 within which are placed specimens 3 of single crystal silicon upon which epitaxial deposition is to take place. Hydrogen is introduced into this tube through a line 4, whence it passes to an electrically operated routing and shutoff device shown diagrammatically as a valve 5. The hydrogen then passes into a line 6 whence it passes into a container 7 surrounded by liquid nitrogen 8. This nitrogen serves to cool the hydrogen and any gases boiling at a temperature higher than nitrogen are trapped by the molecular sieves 9 within the container 7. The cleaned hydrogen leaves vessel 7 through a conduit and is then :allowed either to pass into pipe 16 or pipe 17 depending upon the setting of valves 18 or 19 respectively. The ;gas passing through either of these tubes also encounters a flow meter 20 or 21 respectively, containing a ball 22 or 23 which is supported by the gas and is thus free to indicate rate of flow of gas through the tube concerned. Gas passing through tube 17 enters tube 25. From this the gas may pass into a silicon tetrachloride saturator by valve 27 or through a bypass line 28. The saturator 26 comprises a liquid containing vessel Within which is placed a liquid that will not freeze at the temperature at which the silicon tetrachloride is to be held for evaporation by the hydrogen. A mixture of ethylene glycol and water is suitable. The mixture is cooled by refrigeration coils 40. Within the liquid 36 is suspended a filter vessel 37, evaporator 38 and further filter vessel 39. These vessels are interconnected consecutively, the silicon tetrachloride containing vessel being so arranged that the hydrogen is introduced into the tetrachloride through a perforated disc and allowed to bubble up through it thereby mixing with and carrying off the vapour. After the mixture leaves the filter 3 it passes through the two way valve 29 and into the reaction tube 2.
A two way valve 5 is also arranged so that hydrogen may be excluded and helium introduced through a pipe 45. The helium may thus pass through the saturator and/or the bypass, depending upon the setting of valves 18 and 19.
Gas leaving the reaction tube 2 in a commercial establishment would be reclaimed, but in a laboratory setup it may be burned to avoid contaminating the atmosphere with explosive hydrogen as shown by the gas flame 46. This gas contains a certain amount of hydrogen chloride which should therefore be disposed of by venting into a fume cupboard in the conventional manner.
Turning now to FIGURE 2, in the furnace 1, gas enters tube 2 as shown by the arrow 50 and passes over the specimens 3 spaced out in a lengthwise direction along the tube on supports 51. If the tube 2 is of considerable width as shown here, more than one specimen may be placed in the cross section. Alternatively there may be a single holder 51 extending along the tube on which the silicon specimens 3 are placed. The furnace is heated by a coil 52 shown as a resistance heating device which may be embedded in firebrick 55. By providing a greater number of coils at the end of the tube at which the gas exits than at the entrance to the tube, an increasing temperature profile along the tube can be developed. The profile may be altered by changing the spacing of the coils, and the temperature at any cross section is dependent upon current through 52. Surrounding the coil is a tapered tube of insulating material 53 which again may be varied in thickness as required so as to alter the quantity of heat escaping and to cause more or less heat to be transferred from the heating coil 52 to tube 2. The temperature along the length of the tube may be determined by thermo-couples such as 54 inserted into it.
In one experimental apparatus, the furnace 1 was of firebrick with a Kanthal Al heating element embedded in it. The reaction tube was of fused quartz about centimeters long and 40 millimeters in diameter.
An alternative heating method is by electro-magnetic induction. The coil 52 may carry a high frequency current and since it will develop a more intense field where there are a larger number of loops a desired temperature profile within the tube can again be developed, by applying a suitable material for coupling (such as graphite) to the surface of the tube. The temperature may be measured by inspection through a portion of the tube with a radiation pyrometer.
In the prototype apparatus of FIGURES 1 and 2, operation was begun by, arranging on a quartz holder a single row of slices of silicon horizontally over a distance of 8 cm. along the length of the tube switching on the heating element 52. The apparatus was first purged by passing helium for fifteen minutes through both lines 16 and 17. The saturator 38 was bypassed by the helium by allowing it to fiow through line 23. After that the deposition of silicon was allowed to take place by altering the position of valve 5 to allow hydrogen to enter through pipe 4 and to pass through line 16 and saturator 38. A flow rate of about 0.2 litres of hydrogen per minute was arranged through the saturator whose temperature was kept at 18 C. by means of the bath liquid 36, and 6 to 8 litres of hydrogen per minute was passed through the bypass line 16.
The highest temperature that could be reached with the Kanthal Al wire furnace was just over 1300 C. and the slices were in the range of temperature between 1200- 1300 C. Two runs were made under the following conditions:
Deposition thickness was measured by an infra-red interference fringe method disclosed by W. G. Spitzer and M. Tanenbaum in the Journal of Applied Physics 32, page 744 (1961) and the results for two deposition runs are shown in FIGURE 3. The figures represent thickness of deposit in microns and it can be seen that the Standard Deviation of the deposited layer thickness over all the four slices of any particular run did not exceed 0.4 .t.
The reason for finding curve a of FIG. 4 closest to the temperature gradient producing the most uniform deposition for our resistance heated furnace can be explained in terms of diffusion control of the reaction. The experimental values of AE for our R.F. heated furnace and found by Theuerer were determined using slices of silicon with their planes perpendicular to the direction of the flow whereas in our resistance furnace they were parallel to the flow. We find the effect of the flow rate on the structure of the adsorption layer on the silicon is more significant for perpendicularly oriented slices than parallel ones.
In our resistance heated furnace a predeposition of silicon was formed on the hot wall of the reaction chamber, altering the concentrations of the gases arriving at the slices. The value of the activation energy determined experimentally in a constant temperature region of the resistance furnace under the above mentioned conditions was 60 kcaL/mole as stated before and using this value the desired temperature gradient along the furnace, producing the most uniform deposition was found (curve a, FIGURE 4). With the RF. furnace and having the slices with their planes perpendicular to the gas flow the value of 24 kcal./mole would give the temperature gradient with the most uniform deposition. In other instances where a furnace with other characteristics might be used, the value of AE found for that furnace for a given orientation of the silicon specimens in the gas stream and flow velocity would yield the temperature profile for most uniform deposition.
1. The method of depositing a solid from a gas phase epitaxially onto a solid phase wherein a gas thermally convertible to solid is passed over and along a crystalline surface of a solid upon which epitaxial deposition is to be made, the rate of deposition of solid from said gas being temperature dependent, said surface extending in the direction of flow of said gas, said deposition reducing the concentration of active elements in said gas whereby altering the rate of deposition of solid as the gas progresses along the surface, which comprises the step of, varying the temperature of said surface, in the direction of gas flow, for altering the equilibrium constant for the gas/ solid reaction in a direction to increase the rate of deposition, for a given concentration of active elements in the gas, as the concentration of said elements in said gas is reduced.
2. The method as defined in claim 1, the temperature of said surface being varied in the direction of said gas flow for maintaining a substantially constant rate of deposition of solid from said gas onto said surface.
3. The method of uniformly depositing a solid from a gas phase epitaxially onto a solid phase wherein a gas thermally convertible to solid is passed over and along a crystalline surface of a solid upon which deposition is to be made, said surface extending in the direction of flow of said gas, said deposition reducing the concentration of active elements in said gas, whereby altering the rate of deposition of solid as said gas progresses along the surface, which comprises the steps of, estabilshing a chosen uniform temperature over aid surface, passing said gas over said surface for a predetermined time interval, measuring the thickness of deposit for chosen distances along said surface in the direction of gas flow, and modifying the temperature of said surface progressively in the direction of gas flow to achieve a uniform thickness of deposit over the entire surface.
References Cited by the Examiner UNITED STATES PATENTS 2,880,117 3/1959 Hanlet 117-106 2,877,138 3/1959 Vodonik 118-491 X 3,031,338 4/1962 Bourdeau 117-107.2 X 3,168,422 2/1965 Allegretti et a1. 23223.5 X 3,201,101 8/1965 Jacques 118-49.1 X
ALFRED L. LEAVITT, Primary Examiner. A. GOLIAN, Assistant Examiner.