Emissivity corrected pyrometry method

Abstract

The present invention relates to a method for coating a substrate (22) with at least one layer (23-31), wherein during the deposition of the layer (23-31), at least one optical measuring instrument (10, 11) measures a pair of successive measurements ({U E , n , U R , n}) are determined repeatedly on the layers (23-31), each of which corresponds to a measured radiant power at a light wavelength, and E ) and the reflectance value (U R ), and the measurement pair ({U E , n ,U R , n}) and a given correction value (γ), the actual value of the surface temperature (T C ) is calculated and used to control the temperature controllers (5', 6') to adjust the temperature of the substrate (22) relative to the desired value (s). To improve the determination of the correction factor, multiple (k) measurement intervals (t i ) with at least two pairs of measurements ({U E , n ,U R , n}) is measured, and each measurement interval (t i ) where the temperature dependence coefficient (C i (T i ) is determined and used to calculate a correction value (γ).

Description

[Technical field]

[0001] The present invention relates to a method for coating a substrate with at least one layer, in which during the deposition of the layer at least one optical measuring device repeatedly measures pairs of measurements on the layer, each comprising an emissivity value and a reflectance value, one after the other, and from the pairs of measurements and a predefined correction value, temperature values ​​of the substrate temperature are calculated, which can be used as actual values ​​with which a temperature control device is adjusted to temperature control the substrate relative to a setpoint.

[0002] The invention further relates to an apparatus comprising a computing unit programmable and programmed to compute a correction value. [Background technology]

[0003] US Patent No. 6,398,406 forms the technical background of the present invention. The emissivity-corrected pyrometry method described therein, also called emissivity-corrected, reflectance-corrected or emissivity-compensated, allows non-contact optical temperature measurement during thin film deposition where the optical properties of the measurement object are unknown and continuously changing. Pyrometry methods for non-contact temperature measurement utilize the relationship between the thermal radiation emitted by a hot measurement object and the object's temperature, which is described by the well-known Planck radiation equation, and in practice the object's emissivity is explicitly recorded by a corresponding prior calibration. The measurement object can be any optically accessible surface in a process chamber that is relevant for temperature monitoring or control. In the present invention, the object to be measured is in particular the surface of one or more substrates in a process chamber during a deposition process, which produces a semiconductor layer structure comprising a variety of approximately stoichiometric compounds of main group III (Al, Ga, In) and nitrogen.

[0004] The following publications are also state of the art: WG Breiland, Technical Report SAND2003-1868, June 2003 For example, here is the public release: https: / / www.osti.gov / biblio / 820889 or https: / / prod-ng.sandia.gov / techlib-noauth / access-control.cgi / 2003 / 031868.pdf (hereinafter referred to as "Breiland 2003")

[0005] The prior art includes DE 10 2018 106481 A1, which discloses a generic device.

[0006] DE 44 19476 discloses a method by which the radiation emitted and reflected from a substrate during the deposition of a layer can be measured.

[0007] DE 10 2020 111 293 A1 discloses emissivity-compensated pyrometry for minimization of residual vibrations.

[0008] Known emissivity correction methods are based on determining the missing unknown emissivity by measuring the reflectivity of the surface of the object to be measured. The emissivity is determined using Kirchhoff's law with ε=1-ρ for opaque substrates. The detection wavelength of the pyrometer is selected in such a way that the selected substrate (here silicon) is opaque for wavelengths at normal operating temperatures (T=600-1200 °C), i.e. values ​​in the range of 800 nm to 1000 nm. The method works with sufficient accuracy, since the reflectivity is measured at exactly the same wavelength as the thermal radiation. The light required for this is provided by a laser. In practice, pyrometers do not have a sharp measurement wavelength, but a wavelength interval (approximately ±10 nm, but can be narrower or wider). It is necessary that this interval width and the effective wavelength of the emissivity and reflectance measurements match as closely as possible. The reflectance is measured by emitting light of a given wavelength at the sensor position and reflecting it off the wafer surface at normal incidence, as far as possible at the same position as the pyrometer measurement. The reflectance is determined by means of a prior calibration from the signal strength of the measured reflected light. In practice, the thermal emission and reflectance of an object often cannot be measured simultaneously, but instead are measured separated in time so that the reflectance measurement does not interfere with the thermal emission measurement. See also the lock-in technique described in DE 44 19476.

[0009] For an accurate measurement of the temperature, two different calibration steps are necessary, which allow the determination of the calibration parameters that are incorporated to calculate the temperature from the measurement signal. This involves the calibration of the radiometry with a blackbody radiation source (blackbody furnace, special reference source), thereby establishing a correlation between the intensity signal and the measured temperature. During the measurement, the calibration parameters thus determined can be used to determine the so-called "raw temperature", which has not yet been corrected for the effects of the unknown emissivity. Since the calibration parameters are determined using an independent calibration step, each measured reflectance signal is assigned a reflectance value in the interval 0...1. This calibration step is performed on a substrate of known reflectance (or emissivity in the case of opaque substrates), such as silicon, immediately after the desorption process step (removal of the native oxide), at a known temperature and on an uncontaminated surface, before the start of the layer deposition.

[0010] When depositing thin films at constant growth rate, sinusoidally oscillating temperature measurements are observed unless known emissivity correction methods are used, which are related to interference effects (Fabry-Perot oscillations) in transparent thin films. In the special case of MOCVD deposition of GaN or AlGaN on silicon at temperatures in the range of 950-1100°C, the oscillations are up to ±30°C. The aim of the present method is to reduce this temperature oscillation to less than ±2°C, or more preferably to ±1°C.

[0011] When the temperature measurement method described in the prior art is performed as described, many errors occur, as described below. All of these sources of error lead to an incomplete or artificially over-performed emissivity correction. Incorrect emissivity correction appears as residual temperature oscillations (residual oscillations), the amplitude of which is larger than the desired degree of error.

[0012] In particular, the material system GaN on silicon (Al-GaN) has been found to be particularly susceptible to the above mentioned error sources, since the measured reflectance value R oscillates between near zero and 0.5 due to the refractive index values ​​of the layer and wafer materials and due to the contact between the transparent layer and the opaque substrate.

[0013] The observed error sources are thought to be the following causes, which also occur in practice: In reflectance calibration, the exact reflectance value of the object to be calibrated is unknown, so the reflectance value used in the calibration does not match the physical reflectance, and there is an error in the reflectance value used for emissivity correction. -Errors in the adjustment and setting of the measurement optics. During reflectance measurements, scattering at layer boundaries in semiconductor layer structures causes part of the light that is actually reflected to not be recorded. · Scattered radiation from hot surfaces in the process chamber reaches the measurement head via multiple reflections off the process chamber walls and the wafer surface.

[0014] In the manufacture of electronic devices such as transistors for integrated circuits for power conversion and high frequency amplification, the control and repeatability of the deposition process and the yield of usable devices per wafer are significantly limited in embodiments of known methods on which the present invention is based because the measured wafer temperature is used to control the temperature in a closed control loop. The temperature control adjusts the heating device so that the measured temperature always corresponds to a given set point, and then the physical temperature oscillates accordingly with a residual amplitude of the temperature oscillations that is not fully corrected, which represents a measurement artifact. The device has a multi-layer structure deposited on a substrate, the multi-layer structure having a first portion and a second portion, on which a transition layer, particularly AlGaN, and a buffer layer, particularly GaN, are deposited, on which an AlGaN barrier layer is deposited, such that a two-dimensional electron gas is formed in the layer boundary region between the GaN layer and the AlGaN barrier layer. The reduced reproducibility is particularly related to the fact that the device structure typically consists of a series of functional blocks consisting of a thin AlN seed layer on a Si substrate, a series of transition layers, a series of thick GaN buffer layers, and a relatively thin but temperature-sensitive buffer layer made of AlGaN or AlInN. In the final buffer layer, the deviation from the physical temperature setpoint will vary from run to run or wafer to wafer depending on the random phase position of the residual measured temperature oscillations, thereby driving different compositions of the barrier layers that are critical for device function.

[0015] In order to compensate for the above mentioned error sources, the following theoretical corrections of this measurement method have been described in the prior art, which are based on the mathematically derivable fact that the effects of a large number of error sources can be effectively compensated for by an additional correction value γ, thereby reducing the residual vibrations theoretically to zero.

[0016] A similar method is described in DE 102020 126597 A1. Summary of the Invention [Problem to be solved by the invention]

[0017] The starting point of the invention is the relationship between the measurement signal detected by a pyrometer due to the thermal radiation of the wafer surface and the temperature of the wafer surface, taking into account the emissivity of the wafer surface which differs from 1 due to the physical and optical properties which change during the layer growth.

[0018] This relationship is described by Planck's law of radiation in the Wien approximation, which is expressed here as:

[0019]

number

[0020] The symbols indicate the magnitude of the following values: U E : Measurement signal of thermal radiation on the wafer surface U R : Measurement signal of reflectance of wafer surface ε: Emissivity A, B: calibration parameters, where B<0 α: Measurement signal U R Calibration parameters for reflectance normalization that establish the relationship between the reflectance and the physical reflectance R (0≦R≦1). γ: A calibration parameter for scattered light correction that can also compensate for errors in the calibration parameter α for reflectance determination (up to α=1)

[0021] heat radiation U E and the reflectance of the wafer surface U R The measurement signals of are recorded as close as possible in time and space. The signal corresponding to the thermal radiation is the radiation intensity, which is detected by the detector of the pyrometer and converted into a temperature measurement by means of the calibration parameters A and B, which are determined, for example, by means of a blackbody calibration carried out before the test. The reflectance signal is generated by measuring the intensity of an optical signal emitted from the instrument, ideally with the same wavelength as the thermal radiation measurement, and reflected back by a detector on the reflecting wafer surface. Due to the optical thin film effect (Fabry-Perot effect), the reflectance of the wafer surface during thin film deposition varies approximately sinusoidally over time. Typical growth rates for GaN-on-Si processes are 0.5-5 μm / h, the wavelength used for the measurements is 950 nm, and the oscillation period is up to 10 min. For the further procedure of successively determining the scattered light calibration parameter γ, equation (1) is suitably rearranged.

[0022]

number

[0023] In a specific embodiment of a process in a flat reactor with several individual wafers, the pyrometer is fixedly mounted in an optical window in the ceiling of the process chamber, its line of sight directed to a spot on the surface of the substrate holder. For reasons of thermal averaging during coating, the substrate is rotated slowly around the center of the reactor. A typical rotation period is about 12 seconds, which corresponds to 5 revolutions per minute. However, the rotation speed can be higher or lower. At a particular spot on the wafer to be measured, the measurement signal vs. U E and U R is recorded every 12 seconds. 1 and 2 show the configuration used. The measurement point 13 may be a measurement zone in which several measurements are performed. The measurement signal pairs U E and U R corresponds to the average value over that zone. Measurement points 13 or measurement zones are located on each wafer 7 and may be located at the wafer center, the wafer edge, or anywhere in between. A prerequisite for the applicability of the method is that the temperature is sufficiently constant, or at least the temperature changes are sufficiently small, over a given, relatively small number of rotation periods, i.e. a period of several multiples of 12 seconds, but at least one rotation (12 seconds in the embodiment used). This relatively short time interval is referred to below as t i Let us denote by . Since there is thermal inertia of the overall system, this condition is met during layer deposition as long as one avoids intentional heating and cooling changes, for example when switching between individual layers. In particular, the temperature changes must not exactly match or compensate for residual oscillations in the temperature signal (which are the measurement artifacts to be compensated for) in terms of frequency and phase. This is important to prevent actual temperature changes from being automatically and erroneously corrected for by changes in emissivity. It has been found that a reliable and accurate determination of the calibration factor γ requires recording the raw signal over a time frame significantly longer than the time interval over which the wafer temperature is to remain constant within a tolerance interval. i This time interval, which is long compared to t k This will be referred to as the " The reason for this is that at least a quarter period of the thin film oscillation must pass for the determination of γ, and preferably one or more complete oscillation periods to achieve the required accuracy. Typical period lengths of the thin film oscillations are on the order of a few minutes for typical growth rates of GaN processes and detection wavelengths used in pyrometers. The oscillations arise as a result of alternating constructive and destructive interference of the reflectance signals of the pyrometer or in the deposited thin film during the growth of the layer.

[0024] SUMMARY OF THE DISCLOSURE The object of the invention is to improve the above-mentioned method with regard to the accurate determination of the correction values. [Means for solving the problem]

[0025] This problem is solved by the invention as specified in the claims, the dependent claims representing not only advantageous developments of the invention as specified in the main claim but also independent solutions to the problem.

[0026] It is first essentially provided that in each of a number of measurement intervals, which may be located successively in time, directly successively in time, spaced successively in time, and / or partially overlapping in time, at least two measurement value pairs are respectively measured, each measurement value pair including an emissivity value measured by an optical measuring device and a reflectance value measured by the same optical measuring device or a different optical measuring device. Preferably, multiple pairs of measurements are taken within one measurement interval, at least two or three, and a calibration curve can be constructed using the points of emissivity values ​​plotted against reflectance values. Furthermore, it is proposed to calculate the coefficient of temperature dependence at each measurement interval. It may be assumed that the "true temperature" of the substrate, which is not measurable with the above-mentioned measurement method, changes over the entire measurement time, which lasts for a number of measurement intervals. The change can be up to 20° C. or more. The number of time intervals can be selected such that during one measurement interval the "true temperature" of the substrate changes only slightly, for example by a maximum of 2° C., so that the temperature within each time interval can be considered approximately constant, which means that the coefficients calculated for each measurement interval are different from each other. In the CVD reactor, there is a susceptor rotating about an axis of rotation, on which the substrates are arranged around the axis of rotation, and one particular substrate passes under a measurement spot at intervals of 5 to 20 seconds, where a pair of measurements is taken. The duration of one measurement interval is long enough for the substrate to pass under the measurement spot at least two times. Preferably, the substrate passes under the measurement spot more than twice during the measurement period. A number of layers are deposited on the substrate, the layers being characterized in that the emissivity and reflectivity values ​​change periodically during deposition due to reflections within the layers. The periodic length in time is preferably significantly longer than the measurement interval. The measurement interval may be shorter than one quarter or one tenth of the periodic length. The total measurement time, i.e. the sum of the times of the measurement intervals, is preferably longer than one quarter of the periodic length. It is particularly preferred that the total measurement time is longer than the periodic length. The measurement time may be longer than several times the periodic length. In that case, the total measurement time may be the sum of the times of all measurement intervals.

[0027] In this method, the emissivity-corrected wafer temperature T i The first calculation of i ) for a number of successive measurement intervals t i , t i+1 , t i+2 , ..., where the measurement intervals may overlap. For this calculation, the measured signal U E and U R and equation (3) is used. The result of this first calculation is the normalization coefficient C i (T i ), the emissivity-corrected wafer temperature T i , and the scattered light calibration parameter γ i is the estimate of C i (Ti ) and γ i is U R Measurement signal U as a function of E From the plot of T, the wafer temperature T can be determined by simple linear regression (best fit line) of equation (3), or by any suitable regression or optimization method, for example. i is expressed as C by equation (4). i (T i ) is given by

[0028] In a next step, from these values, the calibration parameter γ required for the most accurate possible scatter correction over all measurement intervals i is calculated. k The value of γ in the short measurement interval i is determined. i In the determination of t, the temperature is assumed to be sufficiently constant, but the number of data points is very small, which leads to large inaccuracies. Therefore, in the second calculation step, a much longer time interval t k >t i Using data from k In this case, the calculation of t k corresponds to at least one quarter of an oscillation period, but preferably several complete periods. The scattered light calibration parameter γ k This second calculation step used to determine may have the following embodiments:

[0029] In a first variant of the invention, a temperature-dependent coefficient is used as the normalization coefficient. k The temperature-independent value U' in E To obtain i In each of the normalization coefficients C i (T i ) to U E The value is normalized. From equation (3), the temperature-independent value U' E is obtained.

[0030]

number

[0031] Long time interval t k U in R U' E From the plot of γ k is determined by simple linear regression or by another suitable regression or optimization method, for example as described in Breiland, supra.

[0032] In each of a number of measurement intervals i, a measurement time t i The reflectance value U R and the emissivity value U E Then, a predetermined number of measurement pairs for each i are measured. Then, the coefficient C i (T i ) is determined. By substituting this into equation (5), we obtain a system of equations whose number corresponds to the number of measurements in the measurement interval i. The system of equations has two calibration coefficients, α and γ, of which γ is unknown, by which the regression can be determined.

[0033] In a second variant of the invention, the temperature T i C i (T i ) and a fit is performed on this temperature curve, resulting in a temperature measurement of C i (T i ) until it matches the temperature curve from k is adjusted.

[0034] First, the coefficient C i (T i ) is determined for each measurement interval i. Next, the coefficient C i (T i ) is used to determine the temperature value. Therefore, the temperature is determined for each measurement interval i. From equation (2), the temperature T' at each measurement point can be determined. Temperature T' is calculated by i By varying the value γ to minimize these deviations, a correction value γ can be determined.

[0035] In a third variant of the invention, a relatively short interval t i The estimated value of γ i But at long intervals t k is averaged over

[0036] The method according to the invention in particular allows the correction value to be continuously modified or adjusted during the deposition of the layer. In particular, it is arranged that the correction value is adjusted with each new pair of measured values. The measurement duration or the number of measurement intervals used for the determination of the correction value can be kept constant or can be varied within a given range. This results in the intervals during which the pairs of measured values ​​necessary for determining the correction value are calculated varying over time. The duration of the intervals, i.e. the measurement duration, can be of a constant value. The result of the above-described method is a calculated temperature, which can be used as an actual value for adjusting the substrate temperature to a setpoint by means of a control circuit and a temperature control device, for example a heating device.

[0037] The invention further relates to an optical measuring instrument having a computing device capable of determining pairs of measurements and of calculating correction values.

[0038] The invention further relates to an apparatus for carrying out the method, for example a CVD reactor, having a gas inlet member arranged in the reactor housing, by means of which a process gas, which may consist of a hydride of main group III and an organometallic compound of main group V, can be supplied to the process chamber. The gas inlet member is a central gas inlet member around which the susceptor rotates. However, the gas inlet member may also be configured in the form of a showerhead. The susceptor, which forms the floor of the process chamber, can be heated by a heating device. This brings the susceptor up to the process temperature. The susceptor can have pockets, each of which accommodates a substrate holder that rests on a gas cushion. The substrate holder is rotated by the gas flow that generates the gas cushion. The height of the gas cushion can be varied in order to vary the heat flow from the susceptor to the substrate holder. The heating device and the means for varying the gap height between the susceptor and the substrate holder form a temperature control device controlled by a computing device, which may include a controller, to regulate the substrate temperature relative to a setpoint. To determine the actual value of the substrate temperature, one or more optical measuring instruments may be provided, capable of measuring emissivity and reflectance values. According to the invention, the computing device is programmed to determine correction values, according to the method described above, that allow the actual value of the substrate temperature to be calculated from pairs of measurements.

[0039] The correction value is preferably derived from a fixed number of measurement pairs. When a new measurement is added to the set of measurement pairs, the oldest measurement pair is preferably discarded so that the correction value is calculated only from the most recent measurement pairs. [Brief description of the drawings]

[0040] Hereinafter, examples of embodiments of the present invention will be described with reference to the accompanying drawings. [Figure 1] FIG. 1 shows diagrammatically an apparatus for carrying out the method. [Diagram 2] FIG. 2 shows a schematic cross-section along line II-II in FIG. 1 on a susceptor 4 on which a substrate 7 is placed, and a measurement point 13 at which the emissivity value UE and the reflectivity value UR can be measured using a reflectance meter 11 and an emissivity meter 10. [Diagram 3] FIG. 3 shows diagrammatically the construction of a multi-layer structure 21 on a silicon substrate 22 . [Figure 4] FIG. 4 shows the temporal progression of the measured emissivity values ​​UE and the measured reflectance values ​​UR over time t, where tk denotes the total measurement time, during which measurement value pairs consisting of emissivity values ​​UE and reflectance values ​​UR are measured at a number of measurement intervals i having duration ti. [Diagram 5] FIG. 5 shows in dashed lines the time course of the "true temperature" over the total measurement time tk, which corresponds to the total duration of the five measurement intervals t1 to t5. [Figure 6]6 shows a schematic way in which the temperature-dependent coefficients Ci(Ti) are obtained by drawing a correction curve through the measurement points on the plot of emissivity values ​​UE against reflectivity values ​​UR, based on at least two, in an exemplary embodiment three, measurement pairs {UE,UR} for one measurement interval i. For this, the intersection of the correction curve with the vertical axis passing through the zero point of the horizontal axis is determined, and the parameters α and γ are obtained from a triangle with a slope indicated by hatching. As α is known and Ci(Ti) is obtained from the distance on the y-axis, γ can be calculated. [Figure 7] FIG. 7 shows diagrammatically a representation similar to FIG. 6 to illustrate a first exemplary embodiment, where all measured values ​​for all measurement periods i are plotted on a graph, such that measurement points for different measurement intervals i result in a correction curve with different slopes and different intersections with the vertical axis, and the parameter α is obtained from a triangle with a slope shown here as hatched. [Figure 8] FIG. 8 is a representation similar to FIG. 6 in which all measured values ​​for all measurement intervals are plotted on one graph, but instead of the emissivity values ​​UE, normalized emissivity values ​​U′E are used, calculated by dividing the measured emissivity values ​​UE by normalization coefficients Ci(Ti), and through the cloud of points so obtained a correction curve is drawn, which intersects the vertical axis at the value 1 and whose slope is the product of the calibration parameter α and the correction value γ. [Figure 9] FIG. 9 is a representation similar to FIG. 5 for illustrating the second embodiment, but instead of the time progression of the “true temperature” of the substrate, the temperature Ti calculated from the coefficients Ci(Ti) for each measurement period i is shown as in FIG. 6. [Figure 10] FIG. 10 is a diagram similar to FIG. 9, in which the midpoint of each time interval is assigned a calculated temperature Ti and a connecting curve, which may be a broken line, is drawn through the calculated temperatures Ti to show the deviations between the calculated temperatures T′n at each measurement point n and the calculated temperature Ti for the measurement interval i, which are minimized in the optimization method. [Figure 11] FIG. 11 is a flow diagram of the method. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0041] 1 and 2 comprises a reactor housing 1, a heating device 5 arranged therein, a susceptor 4 arranged above the heating device 5, and a gas inlet member 2 for introducing, for example, TMGa, TMAl, NH3, AsH, PH3, and H2. The susceptor 4 is rotated about a vertical rotation axis A by a rotary drive 14. For this purpose, a drive shaft 9 is connected to the rotary drive 14 on the one hand and to the underside of the susceptor 5 on the other hand.

[0042] A substrate 7 is placed on the horizontal surface of the susceptor 5 facing away from the heating device 5. A substrate holder 6 is provided on which the substrate 7 is placed. The substrate 7 is located radially outward of the axis of rotation A and is held in place by the substrate holder.

[0043] Two measuring instruments can be provided. The emissivity value measuring instrument 12 can be formed by a pyrometer. The reflectance value measuring instrument 11 can likewise be formed by a pyrometer. A beam splitter 10 can be provided, by means of which the input beam can be split into the two measuring instruments 12, 11. The light path enters the substrate 7 at a measuring point 13. Figure 2 shows that the measuring point 13 moves over all the substrates 7 during the rotation of the susceptor 4.

[0044] 3 shows a multi-layer structure 21 deposited sequentially in several successive coating steps in a coating process. A nucleation layer 23 made of AlN or InN is first deposited on a silicon substrate 22. A first AlGaN layer 24 is then deposited on the nucleation layer 23, followed by a second AlGaN layer 25, followed by a third AlGaN layer 26. The three AlGaN layers 24-26 form a transition layer. The aluminum content of the transition layer may be gradually reduced.

[0045] A first buffer layer 27 made of GaN is then deposited on the transition layers 24-26. This layer may be doped with C. A second buffer layer 28, also made of GaN, is then deposited on the first buffer layer 27, which may be undoped.

[0046] When one of the layers 23 to 31 is deposited on the substrate 22, the reflectance value measuring device 11 and the emissivity value measuring device 12 are used to measure the measured value pair {U E,n ,U R,n} are measured respectively. These pairs of measured values ​​are saved in the memory of the computing device 15. The two measuring devices 11 and 12 are arranged in a common measuring head, and the optical measuring device thus constructed can have the computing device 15. The computing device 15 can be accommodated in a housing arranged outside the reactor housing. A light source can also be accommodated therein, the light of which is guided to the measuring head by a light guide. The measuring head can further have another light guide, by means of which the light is guided to the housing in which the optical measuring device is arranged, which performs the functions of both the reflectance value measuring device and the emissivity value measuring device.

[0047] The actual temperature T of the substrate 7 C can be determined by the Planck radiation law. For simplicity, the following Wien approximation is used for this purpose:

[0048]

number

[0049] Figure 4 shows the emissivity value U E and the reflectance value U R Due to reflections within the deposited layer, the measurement signal oscillates as the layer thickness increases, i.e. as time t progresses.

[0050] In order to compensate for residual vibrations due to the effects of scattered light, etc., the calibration parameter γ described above is inserted, resulting in the following relationship:

[0051]

number

[0052] Here, the element of formula (4) plays an important role in the present invention.

[0053] In the method according to the invention, during the deposition of one of the layers shown in FIG. 3, in particular one of the layers 23 to 28, a time t k 5 shows, as an example, a non-constant time course of the "true temperature" of the substrate, which increases slightly and then decreases slightly. k are measurement times t during which measurements are performed, e.g., t1, t2, t3, t4, and t5. i In each measurement interval i, i.e., during each of measurement times t1, t2, t3, t4, and t5, three pairs of measurements {U E,n ,U R,n Each pair of measurements is then converted into an emissivity value U E and the reflectance value U R Includes.

[0054] Figure 6 shows the relationship between the measurement pair {U E ,U R}, U R Above coordinate system U E From equation (2), it can be seen that the correction value γ embodies the unknown contribution to the slope of the line. From equation (3), it can be seen that the coefficient C(T) can be determined by the intersection of the correction curve with the vertical axis of the coordinate system.

[0055] Using linear regression, at each measurement interval i, we obtain the correction value γ i and coefficient C i Furthermore, we can find the coefficient C i Using equation (4), the temperature T i can be obtained (see Figures 9 and 10).

[0056] If all the measurement points of all the measurement intervals were plotted on a graph, as shown in Fig. 7, correction curves with different slopes and different intersections with the vertical axis would be formed. Fig. 7 shows the individual correction values ​​γ i and coefficient C i indicates that there is a

[0057] Emissivity value U E and the reflectance value U R In order to be able to evaluate the relationship in a common expression, in a first exemplary embodiment of the present invention, the measured emissivity values ​​U E , the coefficient C i Emissivity value U by dividing E The normalized emissivity value U' of all the measurement pairs thus obtained is E,i is shown in Figure 8. The normalized emissivity value U' E , the unified correction value γ, the reflectance value U R,n and emissivity value U E,n The formal relationship between is as follows:

[0058]

number

[0059] The linear relationship resulting from equation (5a) can therefore be represented by the cloud of points shown in Figure 8. The correction value γ can be determined from the slope of a correction curve obtained according to Figure 4 from measurements recorded over at least one cycle length divided into a number of measurement periods.

[0060] This correction value γ is used to calculate the temperature T from the relationship of equation (2) or equation (3), and the temperature T is used for control as the actual value of the substrate temperature for depositing, for example, one of the layers shown in FIG. 3, in which case U R and U E are the current values ​​of the emissivity and reflectance values.

[0061] A second exemplary embodiment of the present invention will now be described with reference to FIGS.

[0062] FIG. 9 is a diagram similar to FIG. 5, but instead of the “true temperature” which is technically difficult to measure, the temperature T i The temperature Ti is plotted as follows: i is determined directly from

[0063]

number

[0064] Next, the temperature T that passes almost through the midpoint of the measurement interval i i In Fig. 10, this correction curve is shown as a dashed straight line between the midpoints of the measurement intervals i. i (T i ) the average temperature T i is the temperature T' calculated according to the following formula: n It is clear that it deviates from

[0065]

number

[0066] For example, the value of the correction value γ is calculated by the following optimization calculation: i But, C i (T i ) is matched as accurately as possible to the temperature curve given by

[0067]

number

[0068] In the exemplary embodiment shown in FIG. i A broken line may also be placed as the correction curve through . In another exemplary embodiment, an exponential curve, a sinusoidal curve, or a combination of such curves may be used.

[0069] In a third exemplary embodiment of the present invention, as described above with reference to FIG. 6, for each measurement interval i, first each correction value γ i Then, the average value is calculated from these correction values ​​γi.

[0070] Further embodiments related to the above-mentioned methods of the first, second or third exemplary embodiment may have the following characteristics.

[0071] One particular data point U Ei (U Ri ) for one time interval t i From C i (T i In addition to calculating Ti, two or more time intervals t i , t i+1 , ... to C i (T i ) and Ti can also be determined. In that case, the effective values ​​C i (T i ) and T i is the average value of the calculation over multiple time intervals.

[0072] In a preferred variant or exemplary embodiment of the invention, the number of measurement intervals or the measurement duration is kept constant. In this variant, when calculating the correction value γ, only the most recent measurement pair is selected. When a new measurement pair is added to the set of measurement pairs, the oldest measurement pair is removed from the set. This procedure creates a measurement window that moves over time, within which the measurements required for the correction value are determined. The correction value γ is updated with each measurement.

[0073] Furthermore, U' E In order to determine as accurately as possible the values ​​used in the calculation of i (T i) a sliding approximation can be applied (possibly weighting the most recent measured data). Depending on the type of temperature change expected, this approximation can be linear, polynomial, exponential, sinusoidal or a combination of these functions. Depending on the signal format, an appropriate function may be automatically selected.

[0074] Time interval t i In order to reduce noise in the data resulting from the calculation over T, particularly erroneous data can be eliminated. i and γ i Both can be used as figures of merit. i.T i t i >t x >t k If, then the time interval t x The temperature comparison can be compared with the average temperature within the time interval t (which may have been measured with an inaccurate γ value). If the deviation is too large, e.g., greater than 10°C, the corresponding data point is rejected and not used in further calculations. x This can also be done using the maximum temperature measured at, which is usually closer to the actual temperature than the average value. ii. Usually, γ k It is known from experience what range lies within. For example, it can be a value between 0.95 and 0.98. i If empirical values ​​are not available, the classically calculated γ k,klass and γ calculated as described here k By comparing γ i Maximum allowable offset from |γ k,klass -γ k The allowable offset is |γ k,klass -γ k It can also be a multiple of |.

[0075] A longer time interval t kWhen performing an approximation over , values ​​with particularly large deviations can be eliminated. This is possible in the first, second and third exemplary embodiments.

[0076] The data points with minimum and maximum reflectance (or maximum and minimum emission signals) can be excluded from the calculations. These data points are particularly susceptible to noise due to small changes in reflectance and emissivity.

[0077] A further possibility to improve the noise is to use more measurement points or measurement zones (Fig. 2). In general, the measurement zones should be chosen to be small enough, since with different layer thicknesses on the wafer, the residual oscillations in the temperature signal are averaged out, but the resulting signal will be inaccurate. However, the γ k The calculations of γ can be performed simultaneously. These measurement zones can be on the same wafer or on several wafers. A further advantage is that this allows recalculation more frequently than once per revolution. When the layer growth is sufficiently uniform, γ k is expected to be constant for all measurement zones and measurement points, so γ from various measurement zones k The measured value can be taken as the valid value.

[0078] As explained, over the course of the process, γ k Using the method presented here, γ k can be modified continuously. However, γ k The sudden change of γ k The continuous variation of can be smoothed using an appropriate filtering method (eg, a low-pass filter, etc.).

[0079] In the method of the third exemplary embodiment, instead of simple averaging, successive approximations are used to estimate γ kDepending on the predicted change, the approximation can be linear, polynomial, exponential, sinusoidal, or a combination of these functions. Depending on the signal format, an appropriate function can also be selected automatically.

[0080] This procedure can also be used in the methods of the first and second embodiments. In this case, γ k is a time-dependent function γ k Replaced by (t). i. In case of a temperature step or pause in the growth during the process, the time function γ k This must be taken into account in (t), e.g., by removing unused values ​​or shifting the time axis so that there are no gaps in the data.

[0081] According to the method of the first exemplary embodiment, the data U′ E (U R ) is ideally equal to 1. If the deviation is greater than a specified value, γ k We can declare certain values ​​of γ to be invalid. k The value of will continue to be used.

[0082] If sufficient data is not available, γ k can be specified, which can be obtained, for example, from empirical values ​​or previous process results. This value then becomes the basis for rejecting data, serves as the basis for a filter, or serves as a starting point for an approximation.

[0083] In case of extreme temperature changes (e.g. between two layers), the data U E (U R ) cannot be used. This can be done automatically by excluding data points, or manually by recipe control and / or detection of temperature setpoint changes. Data before and after the temperature change can be used to calculate γ k It is possible to perform continuous calculations of

[0084] All three methods were used simultaneously to measure γk It is also possible to calculate the effective γ k can be selected using the methods described above. If there are multiple valid values, the value can be defined by averaging or prioritizing in a pre-specified manner.

[0085] Gamma i and T i (and the resulting C(T i Since both γ ) can vary over time, the second exemplary embodiment also uses an iterative function to adjust γ i and T i Then, from this time course, γ k can be determined. In the iterative procedure, γ k or gamma k For each calculation of (t), k or gamma k Considering (t), T i may be recalculated. b. C(T i ) can be used to perform optimization calculations according to the method of the first or third exemplary embodiment. Then, C(T i ) is taken into account in a linear approximation.

[0086] Time interval t k The length of can be automatically determined to accurately map one or more vibration periods.

[0087] The error in tilt caused by temperature change is U E Or U R This allows us to determine whether each t i The calculations in the intervals can be corrected.

[0088] The temperature change is T i and T i+1 If the temperature change is constant, U E and U RUnless calculations are performed at the inversion point of T i and T i+1 shifts in the same direction. Then, this is used repeatedly to i and T i+1 This allows the calculation of

[0089] The above description describes the inventions which are the subject of this application as a whole, which independently advance the prior art through at least the following combinations of features, any two, more or all of which may be combined:

[0090] k multiple measurement intervals t i During the measurement period in, at least two pairs of measurements {U E,n ,U R,n} are measured, and each of the measurement intervals t i In the temperature dependence coefficient C i (T i ) is determined, and the coefficient C i (T i ) is used to calculate the correction value γ.

[0091] The coefficient C i (T i ) is a product of a first calibration parameter A and an exponential function of a quotient of a second calibration parameter B and a temperature T.

[0092] The coefficient C i (T i ) is the reflectance value U R Emissivity values ​​U plotted against E is obtained by regression or by optimization methods.

[0093] The emissivity values ​​U recorded during at least several of said measurement intervals E For each coefficient C i (T i ) to normalize the emissivity value U' E is generated, and a reflectance value U R Normalized emissivity values ​​U' plotted againstE The correction value γ is calculated by regression or by optimization methods.

[0094] The correction value γ is the normalized emissivity value U′ E The method is characterized in that the correction curve is generated from the slope of the correction curve passing through the

[0095] The correction value γ is the coefficient C i (T i ) the temperature value T i For each of the measurement pairs {U E,n ,U R,n} the temperature value T' calculated from i The method is characterized in that the signal is obtained by modifying the signal.

[0096] Each of the measurement intervals t i Correction value γ i are determined, and these correction values ​​γ i wherein an average value is generated from

[0097] During deposition of one of the layers 23 to 31 or the multilayer structure 21 consisting of a plurality of layers 23 to 31, the emissivity U E and the reflectance value U R changes periodically with a periodic length, and during one measurement interval t i The duration of all measurement intervals t i wherein the total duration of said period is greater than one-quarter of said period length or greater than several times said period length.

[0098] The measurement interval t used to determine the measurement duration or correction value γ i is kept constant and / or the correction value γ is varied over time.

[0099] The correction value γ may be continuously optimized during deposition of one or more layers on the substrate, and / or the correction value γ may be adjusted based on newly determined measurements vs. {U E,n,U R,n} and / or a constant number or a number of measurement pairs that varies within a predetermined range are used in calculating the correction value γ.

[0100] The temperature value T C is an actual value and a temperature control device (5, 6') for regulating the temperature of said substrate (22) is adjusted to a setpoint value s using said actual value.

[0101] 12. A measuring apparatus, wherein the arithmetic unit is programmed so that the correction value γ is determined by a method according to any one of claims 1 to 11.

[0102] 12. An apparatus comprising a measuring device according to claim 11.

[0103] All disclosed features are essential to the invention (both by themselves and in combination with one another). The disclosure of the present application includes in its entirety the disclosure content of the relevant / added priority documents (copies of the earlier application), also with a view to incorporating the features of these documents into the claims of the present application. The dependent claims are characterized by an independent, inventive further development of the prior art, even without the features of the claims cited, in particular in order to file a divisional application on the basis of these claims. The invention specified in each claim may additionally have one or more features specified in the preceding description, in particular those given reference signs and / or specified in the explanation of the signs. The present invention also relates in particular to embodiments in which individual ones of the features mentioned in the preceding description are not implemented, insofar as they are obviously unnecessary for the respective purpose of use or can be replaced by other means having the same technical effect. [Explanation of symbols]

[0104] 1 Reactor housing 2 Gas inlet member 3 Gas supply lines 4 Susceptor 5 Heating device 6 PCB holder 6' Gas Cushion 7 Substrate 8 Process Chamber 9 Rotation Axis 10 Beam splitter 11 Reflectance value measuring instrument 12 Emissivity measuring device 13 Measurement points 14 Rotational drive unit 15 Arithmetic unit 21 Multilayer structure 22 Substrate 23 Nucleation layer 24 Transition layer 25 Transition layer 26 Transition layer 27 Buffer Layer 28 Buffer Layer 29 Boundary layer, two-dimensional electron gas 30 Barrier Layer 31 Top Floor Gamma correction value Gamma i Measurement interval correction value λ wavelength α Calibration parameter A Rotation Axis α Calibration parameter A Calibration parameters B. Calibration parameters U E Emissivity Value U' E Normalized emissivity value U R Reflectance Value U E,n Emissivity measurements U R,n Reflectance measurements T C Corrected temperature, actual temperature T M Measurement temperature T S Set temperature t i Time Interval i Measurement interval index n Index of measurement pair k is the number of measurement intervals t k Total measurement time s Setting value

Claims

1. A method for coating a substrate (22) with at least one layer (23-31), comprising the steps of: During the deposition of the layers (23-31), at least one optical measuring instrument (10, 11) measures a pair of measurements ({U Ｅ，ｎ , U Ｒ，ｎ }) in turn, and for each of the measurement pairs ({U Ｅ，ｎ , U Ｒ，ｎ }) is the emissivity value (U Ｅ ) and the reflectance value (U Ｒ ) and The actual value of the substrate temperature (T Ｃ ) is the pair of measurements ({U Ｅ，ｎ , U Ｒ，ｎ }) and a predetermined correction value (γ), Multiple (k) measurement intervals (t ｉ During the measurement period in {U Ｅ，ｎ , U Ｒ，ｎ }) are measured, and each of the measurement intervals (t ｉ ) where the temperature dependence coefficient (C ｉ (T ｉ ) is determined, and the coefficient (C ｉ (T ｉ ) is used to calculate the correction value (γ).

2. The coefficient (C ｉ (T ｉ 2. The method according to claim 1, characterized in that the calibration parameter (A) is the product of the first calibration parameter (A) and the exponential value of the quotient of the second calibration parameter (B) and the temperature (T).

3. The coefficient (C ｉ (T ｉ )) is the reflectance value (U Ｒ ) plotted against the emissivity values ​​(U Ｅ 3. The method according to claim 1, wherein the parameter is obtained by regression or by optimization methods.

4. The emissivity values ​​(U Ｅ ) to each of the coefficients (C ｉ (Ti)) to obtain the normalized emissivity value (U' Ｅ ) is generated; and The reflectance value (U Ｒ ) plotted against the normalized emissivity values ​​(U′ Ｅ 4. The method according to claim 1, wherein the correction value (γ) is calculated by regression or by optimization methods.

5. The correction value (γ) is the normalized emissivity value (U′ Ｅ 5. The method of claim 4, wherein the correction curve is generated from the slope of a curve passing through the first and second inputs.

6. The coefficient (C ｉ (T ｉ )) the temperature value (T ｉ ) for each of the measurement pairs ({U Ｅ，ｎ , U Ｒ，ｎ }) ｉ 4. The method according to claim 1, wherein the correction value (γ) is calculated by modifying

7. Each of the measurement intervals (t ｉ ) for each correction value (γ ｉ ) are determined, and their correction values ​​(γ ｉ 4. The method according to claim 1, wherein the average value is generated from

8. The radiation value (U Ｅ ) and the reflectance value (U Ｒ ) changes periodically with a periodic length during deposition of the layer (23-31) or the multilayer structure (21) consisting of a plurality of layers (23-31), and during one measurement interval (t ｉ ) is shorter than a quarter or a tenth of the period length, and ｉ 8. The method according to claim 1, wherein the total duration of the period is greater than a quarter of the period length or a multiple of the period length.

9. The measurement duration or measurement interval (t ｉ 9. The method according to claim 1, characterized in that the number of (γ) is kept constant and / or the correction value (γ) is varied over time.

10. The correction value (γ) may be continuously optimized during deposition of one or more layers on the substrate, and / or the correction value (γ) may be adjusted based on the newly determined pair of measurements ({U Ｅ，ｎ , U Ｒ，ｎ 10. The method according to claim 1, wherein the correction value (γ) is updated every 100 ms.}) and / or a constant number or a number of measurement pairs varying within a predefined range is used to calculate the correction value (γ).

11. The temperature value (T Ｃ 11. The method according to claim 1, wherein the setpoint (s) is an actual value and a temperature control device (5, 6') for regulating the temperature of the substrate (22) is adjusted using the actual value relative to a setpoint (s).

12. The emissivity value (U Ｅ ) and reflectance value (U Ｒ ) and the measurement pair ({U Ｅ，ｎ , U Ｒ，ｎ }) in succession, and using a correction value (γ) to obtain an actual temperature value (T Ｃ In a measuring device comprising: Measuring device, characterized in that the computing device (15) is programmed such that the correction value (γ) is determined by a method according to any one of claims 1 to 11.

13. 1. An apparatus for depositing a layer on a substrate, comprising: a gas inlet member (2) disposed within the reactor housing (1) for passing a process gas therethrough and supplying the process gas to the process chamber (8); a susceptor (4) that can be heated by a heating device (5) and has a surface facing the process chamber (8), on which a substrate (22) can be placed; The emissivity value (U Ｅ ) and reflectance value (U Ｒ one or more optical measuring devices (10, 11) for measuring the The emissivity value (U Ｅ ) and reflectance value (U Ｒ and a control device for regulating the temperature of the substrate (22) using the A calculation unit (15) determines a correction value (γ) and calculates an actual temperature value (T Ｃ ) in said apparatus, 12. Apparatus comprising a measuring device according to claim 11.

14. A method or apparatus characterised by one or more of the characterising features according to any of the preceding claims.

Images