Analysis of Rheed data from rotating substrates

Abstract

A video tracking system and a program employing frequency-domain analysis for extracting RHEED intensity oscillation data for film growth on rotating substrates. In initial experiments on GaAs growth, excellent (2%) agreement has been obtained between oscillation frequencies measured for static substrates and substrates with rotation rates as high as 10 rpm. The capability of performing RHEED analysis on rotating substrates could lead to improvements in the quality of complex epitaxial structures and interfaces for which interrupting rotation can have a deleterious effect.

Description

COPYRIGHT

Appendix A of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.

BACKGROUND ART

Measurement of film thickness during growth is required in a number of arts and especially in the growth of thin film semiconductors on substrates. Such films may comprise amorphous or polycrystalline or epitaxial films. Such films may be grown or deposited in a variety of reactors. One such reactor is a molecular beam epitaxy (MBE) reactor.

Recently, the examination of images formed by in situ reflection high-energy electron diffraction (RHEED) intensity data has become one of the more useful tools for the analysis of growth by MBE. Static RHEED images, obtained when growth is interrupted, yield detailed information on surface reconstruction. Analysis of dynamic RHEED images ("RHEED oscillations"), plotted as a function of intensity versus time, and obtained as growth is taking place, can be employed to determine epitaxial growth rates and, therefore, alloy compositions. (See U.S. Pat. No. 4,855,013 issued Aug. 8, 1989.) This method is applicable because epitaxial growth causes a variation in surface roughness on the atomic scale that, under favorable conditions, produces well-resolved oscillations in the RHEED intensity. The period of these oscillations is the time required for the growth of one complete monolayer. Frequently, however, the conditions used for epitaxial growth yield oscillations that are not sufficiently obvious to permit the period to be obtained directly from plots of intensity versus time. Another major disadvantage with current known RHEED oscillation analysis methods is that substrate rotation, which is generally employed in order to improve uniformity of the growing films and interfaces, must be stopped in order to obtain a fixed diffraction pattern from which oscillation data can be obtained.

A need exists, therefore, for a method and apparatus which permits analysis of RHEED or other oscillations under unresolved growth conditions and which permits the acquisition of RHEED oscillation data while the substrate continues to rotate.

SUMMARY OF THE INVENTION

In accordance with the present invention, digital signal processing and frequency-domain analysis is used to enable a growth oscillation frequency to be extracted from data obtained during substrate rotation.

In accordance with the invention, a layer of material being grown on a rotating substrate in a reactor is subjected to an energy beam. Oscillatory variation in the beam, resulting from beam incidence on the growing material, are imaged and converted into frames of picture elements (pixels). Each frame is processed to produce a time varying pattern of the beam variations. The time varying pattern is converted to a frequency varying pattern to disclose pattern characteristics not readily observable in the time domain. These characteristics may be used to control growth process parameters.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic diagram of a measurement system in accordance with the invention.

FIG. 2 is a plot of RHEED intensity versus time for a non-rotating sample.

FIG. 3 is a plot of the power spectrum, i.e., power versus frequency of the RHEED data of the data from the sample of FIG. 2.

FIG. 4 is a plot of RHEED intensity versus time for a sample rotating at 5 rpm.

FIG. 5 is a plot of the power spectrum of the data from the sample of FIG. 4.

FIG. 6 is a plot of the power spectrum of data for the sample of FIG. 4 rotating at 10 rpm.

FIG. 7A is a plot of amplitude versus time of a given pixel intensity profile illustrating the clipping effect caused by limitations in dynamic range.

FIG. 7B is a plot as in FIG. 7A in which the true peak amplitude A1 is extracted assuming a Gaussian intensity profile.

FIG. 8 is a flow chart of the program of Appendix A.

DETAILED DESCRIPTION OF THE INVENTION

The invention will now be described in detail in connection with the drawings. FIG. 1 is a schematic drawing of a preferred embodiment of the system of the invention, as illustrated by a RHEED measurement system, in accordance with the invention.

An MBE reactor 10, such as a Varian GEN II modular 3-in (MBE) reactor system 10 may be used for the growth of films 13 of various materials on various substrates 12. Typically such materials comprise Group III-V film materials and substrates. For example, GaAs substrates may be prepared by chemical cleaning and etching followed by mounting on In on Mo blocks (not shown) and in situ oxide desorption at about 600.degree. C. The substrate temperature during growth is maintained at about 600.degree. C. as monitored by thermocouple and optical pyrometer measurements (not shown). Typical growth conditions are employed, including an As.sub.4 /Ga beam equivalent pressure of about 10 and a growth rate of about 1 .mu.m/h.

A RHEED image 11 is generated by subjecting growing film 13 in reactor 10 to an electron beam (E-Beam) with a standard Varian RHEED electron gun 2 operated at 8 kV with an emission current of 2-3 A. During growth, the vertically mounted substrate 12 is rotated by motor 18 coupled by shaft 19 to substrate holder 23. The resultant image is shown on the phosphor screen 17 of RHEED screen 15. The image shown is that of a 2.times.4 As-stabilized RHEED pattern at the region of specular reflection where the E-beam is exiting at a near angle of incidence. The pattern will move as indicated by arrow A, when the substrate is rotated in the direction of arrow B. The moving image is captured by a CCD TV camera 16 equipped with a macro lens 14 that permits the magnification of any part of the image.

TV camera 16 is a CCD camera with RS-170 video output. The video output is coupled to an 8-bit frame grabber board 22 capable of storing each frame imaged by the CCD camera, comprising 640.times.480 pixels per frame, for analysis by computer 24. The frame grabber board acquires an entire image in 1/30s, a capture rate that yields a flicker-free display on TV monitor 20.

Computer 24 may comprise an 80386-based desktop computer equipped with a math co-processor and a commercial software package (National Instruments Lab Window.RTM.), for data collection and analysis. The software package provides a tool for developing a digital signal processing program (see Appendix A) to accomplish such tasks as base line correction to compensate for drift, digital filtering to remove undesirable noise components, fast Fourier and fast Hartley transform analysis, power spectrum measurements and calculating RHEED oscillations for both stationary and rotational substrates by calculating the maximum intensity of the diffraction pattern image per sub-frame.

A significant task in developing the video analysis system was to devise a method for extracting useful RHEED oscillation data from the video images captured by the frame grabber. Initially, it was hoped that identifying and then tracking the diffracted intensity patterns associated with a specified set of substrate azimuths would make it possible to collect the needed oscillation information from selected points on these patterns as they swept across the phosphor screen during substrate rotation. At normal rotation speeds, however, the intervals during which no identified pattern was present on the screen were too great to permit satisfactory analysis. Fortunately, tracking the region of diffracted intensity generated by the specular reflection of the electron beam from the substrate surface permitted useful RHEED oscillation information to be extracted from the video data.

To track the region of specular reflection three approaches were tried. A single-pixel tracking method failed because of the magnitude of the movement of the reflected beam due to nonplanar bonding of the substrate, wobble in the mechanical components of the azimuthal rotation apparatus, and instabilities in the electron gun. In addition, the spatial extent of the image of the reflected beam, together with the limited dynamic range of the frame grabber board 22, made evaluation of the peak intensity difficult. As a second approach, from the full video image, a rectangular subframe was selected that included the entire specular reflection, and the reflection was tracked by moving this subframe. However, the computational overhead required to analyze the intensity data associated with the large numbers of pixels contained in the subframe was so high that the processing rate dropped to about 5 frames/s. This rate was too low to yield enough data for accurately determining the oscillation frequency for growth at a normal rotation rate of 10 rpm. In the third approach, which was successful, a combination of 64 pixels in a single horizontal row and 64 pixels in a single intersecting vertical column was selected to track the region of specular reflection. With these two lines of pixels, the position of the peak intensity of the specular reflection could be accurately tracked.

Because of the limited dynamic range of the frame grabber board, it may sometimes be necessary to evaluate the peak intensity from data for lines of pixels of which one or more may have a saturated intensity value, as shown in FIG. 7A. This task may be accomplished by fitting the intensity profile to a Gaussian function and then estimating the peak intensity Al in the absence of saturation, as shown in FIG. 7B. With the combination of tracking and intensity analysis performed by the computer 24 under the direction of the program of Appendix A, video data may be processed at up to about 20 frames/s. This rate yields an uncertainty of approximately 1% in a growth rate of 1 .mu.m/h.

FIG. 8 is a flow chart of the main steps in the program of Appendix A. First, the computer is initialized and control parameters are inputted (STEP 1). A mouse is then used to select the position of initial tracking using a real-time display on the TV monitor of the rotating RHEED data (STEP 2). The intensity profile of the data is then measured along horizontal (H) and vertical (V) lines and a determination is made as to whether or not the V & H target lines are centered on the region of maximum intensity (STEP 3). If they are, the intensity data of the vertical target line is stored (STEP 5). If not, the H & V target lines are translated to correspond to the TV centered position and the new position (STEP 4) displayed and STEPS 2 and 3 repeated with the new position.

The vertical target line data stored in STEP 5 is corrected in accordance with a Gaussion curve, as necessary in STEP 6, and this data is subjected to frequency domain analysis as disclosed in the parent application referenced above (STEP 7).

The intensity data extracted from the video system is then processed and analyzed by means of frequency-domain techniques described in the parent application incorporated herein by reference. For the rotational RHEED data, it is also necessary to perform digital filtering to remove spurious frequency components. A relatively wide bandpass filter, with a .+-.20% frequency cutoff, was found to be adequate for this purpose. Analyzed growth data from the computer 24 is then coupled to reactor controls 28 to control various growth parameters, such as flux ratios of the III-V elements, rotation speed, temperature, etc.

FIG. 2 shows the time-domain RHEED intensity data for a non-rotating GaAs sample obtained with the video tracking system operating at about 20 frames/s. There are 512 data points, corresponding to about 25 s of data acquisition. In this case, of course, the tracking was not necessary. FIG. 3 shows the power spectrum obtained by frequency-domain analysis of the data of FIG. 2, with a well-resolved peak at 1.17 Hz. FIG. 4 shows the time-domain data taken for the same sample, but with a substrate rotation rate of 5 rpm. No oscillatory behavior is apparent. FIG. 5 shows the power spectrum of the data from FIG. 4, with a peak at 1.15 Hz. FIG. 6 shows the power spectrum obtained by frequency-domain analysis of data taken for the same sample rotating at 10 rpm. For this case, a well-resolved central peak of 1.17 Hz is visible. Thus, for static, 5-rpm and 10-rpm experiments, the extracted RHEED oscillation frequencies agree to within 2%.

While we were to able extract RHEED oscillation data in experiments on GaAs samples rotating at rates above 10 rpm, the signal-to-noise ratio in the power spectrum was reduced. With faster hardware and improved software it is expected that well-resolved power spectra from data for samples rotating at rates approaching the current maximum of about 100 rpm may be obtained.

J. Zhang et al, Appl. Phys. A.42, 317 (1987) reported that the phase relationships of RHEED oscillations as a function of azimuthal angle for GaAs growth are quite complex because of the interaction of multiple scattering processes. It is surprising, therefore, that sufficient signal power of the fundamental oscillation frequency can be recovered to permit discrimination of this frequency from the other components that are present.

EQUIVALENTS

Those skilled in the art will know, or be able to ascertain using no more than routine experimentation, many equivalents to the specific embodiments of the invention described herein.

While the invention has been described in connection with RHEED oscillations, other applications are envisioned for this system. For example, any in situ method for generating an incident beam on a specimen during growth and detecting the resultant beam variations with time will benefit from the invention. Thus, light scattering, rather than diffraction phenomena, may be detected and used as the time domain data for determining growth oscillation. Photo-emissions or reflected beams may also generate detectable oscillations. The beam may comprise any suitable energy beam, such as an electron beam, photon beam, ion beam or X-ray beam.

All manner of growth chambers, including (PVD) or the like, are contemplated, as well as a wide variety of growth materials, such as superconductors, silicides and metal organic films of semiconductor material, without limitation.

These and all other equivalents are intended to be encompassed by the following claims. ##SPC1##

Claims

1. A method of determining growth properties of a layer of material as it is being grown on a substrate, comprising the steps of:

a) rotating the substrate upon which the layer is being grown;

b) subjecting the growing layer to a beam of energy impinging on the layer as the substrate is rotated to modulate the beam by the variations in the growth;

c) converting the beam modulations into a visual image consisting of frames of pixels;

d) tracking a selected set of pixels in a frame;

e) generating time varying data from said set of pixels of the oscillatory variations induced in said beam by said modulations during the growth process;

f) converting said time varying data to frequency varying data; and

g) utilizing said frequency varying data to determine growth properties.

2. A method of analyzing the growth process of a layer of material as it is being grown on a substrate, comprising the steps of:

a) rotating the substrate;

b) directing a beam of energy onto the growing layer as the substrate is rotated to modulate the beam by the growth process;

c) forming a visual image of the modulated beam, said image consisting of frames of pixels;

d) tracking a selected set of pixels in a frame;

e) from said set of pixels generating time varying data of the variations in the beam induced by said growth process; and

f) generating frequency varying data corresponding to said time varying data.

3. The method of claim 1 wherein the selected set of pixels consists of a row and column of such pixels.

4. The method of claim 1 wherein the beam of energy is taken from the group comprising electron, photon or X-ray energy and the time varying data relates to data from the group comprising scattered, diffracted, reflected or transmitted beam variations.

5. The method of claim 1 wherein the time varying data is processed to determine the maximum intensity of each frame.

6. The method of claim 1 wherein the growth properties determined are used to control further growth of layers.

7. The method of claim 2 wherein the beam modulation is from the group comprising scattered, diffracted, reflected or transmitted beam modulation.

8. The method of claim 2 wherein the time varying data in claim 2 is processed to determine the maximum intensity of selected pixels in each frame.

9. The method of claim 2 wherein information contained in the frequency varying data is used to control the growth of the layer.

10. Apparatus for controlling the growth of a layer of material grown on a substrate during the growth process, comprising:

a) apparatus for rotating the substrate;

b) an energy source for subjecting the growing layer to a beam of energy to produce variations in the energy beam caused by the growth process;

c) apparatus for detecting such variations in the energy beam caused by the growth process and generating a visual image of such variations, said visual image comprising frames of pixels;

d) tracking means for tracking a set of such pixels and generating a frequency varying pattern corresponding thereto; and

e) apparatus for varying the growth of the layer in response to information contained in said pattern.

11. Apparatus for analyzing the growth of a layer of material during growth on a rotating substrate comprising:

a) apparatus for rotating the substrate;

b) apparatus for directing a beam of energy at the growth layer as the substrate is rotated to produce variations in the energy beam induced by the growth process;

c) imaging means for producing an image of said variations comprising frames of pixels;

d) tracking means for tracking a set of such pixels; and

e) apparatus for generating a frequency varying beam pattern from said tracked pixels corresponding to variations in the beam induced by said growth.

12. A method of controlling the growth of a layer of semiconductor material as it is being grown on a rotating substrate, comprising the steps of:

a) rotating the substrate;

b) subjecting the growing layer to a beam of electron energy to obtain a diffraction pattern;

c) generating a visual image of said diffraction pattern which is comprises of frames of pixels;

d) tracking a set of pixels in each frame;

e) from said tracked pixels generating frequency varying data corresponding to oscillatory variations induced in said diffraction pattern; and

f) utilizing said frequency varying data to control said growth process.

13. A method of analyzing the growth of a layer of semiconductor material as it is being grown on a rotating substrate comprising the steps of:

a) directing a beam of electron energy onto the growing semiconductor layer forming on said rotating layer;

b) imaging a time varying diffraction beam pattern of frames of pixels corresponding to the variations in the electron beam induced by said growth;

c) tracking a set of pixels in said frames; and

d) from said set of pixels generating a frequency varying beam pattern corresponding to said time varying beam pattern.

14. Apparatus for controlling the growth of a layer of semiconductor material on a substrate during the growth process, comprising:

a) apparatus for rotating the substrate;

b) an energy source for subjecting the growing layer to a beam of electron energy to produce a diffraction pattern;

c) apparatus for detecting variations in said electron energy beam caused by the growth process and generating a visual image in response thereto said image comprising frames of pixels;

d) tracking means for tracking a set of pixels in each frame;

e) apparatus for processing a set of pixels to produce a frequency varying pattern corresponding thereto.

15. Apparatus for controlling the growth of a layer of semiconductor material being grown on a rotating substrate during the growth process, comprising:

a) apparatus for rotating the substrate;

b) an energy source for subjecting the growing layer to a beam of electron energy to produce a diffraction pattern;

c) apparatus for imaging variations in said electron energy beam caused by the growth process and generating a time-varying pixel pattern in response thereto;

d) apparatus for processing said time varying pixel image pattern to produce a frequency varying pattern corresponding thereto comprising:

(i) apparatus for forming multiple frames of pixels of said image pattern;

(ii) apparatus for selecting a portion of the pixels from said frames;

(iii) apparatus for determining the peak intensity of the pixels selected and for storing data representing said peak intensity in the form of time varying data;

(iv) apparatus for determining the means of the time varying data;

(v) apparatus for determining the baseline trend of the time varying data by fitting a polynomial to the data and subtracting the baseline from the means of the data;

(vi) apparatus for processing the baseline of the time varying data by performing a Fourier transform of the baseline of the time varying data to generate said frequency varying data; and

(vii) apparatus for determining the centroid of such frequency varying data.

16. The apparatus of claim 14, including a filter for selectively removing undesired frequency components from said frequency varying pattern.