METHOD FOR RECORDING A STATE OF A CVD REACTOR UNDER PRODUCTION CONDITIONS

Abstract

During a process involving one or more process steps of a process phase, in which a substrate is located in the process chamber of a CVD reactor, a process temperature and a pressure are each set and a process gas flow is fed into the process chamber by way of control data delivered by a controller in accordance with a formula stored in the controller. Additionally, sensors are used to determine measurement data from which a current fingerprint is calculated and then compared with a historic fingerprint. The fingerprint includes only values or groups of values that are obtained from measured values that are recorded during one or more conditioning steps of a conditioning phase in which a conditioning temperature and a conditioning pressure are each set and a conditioning gas flow is fed into the process chamber in accordance with control data specified by the formula.

Claims

1. A method for operating a chemical vapor deposition (CVD) reactor (1) having a process chamber (3), the method comprising: during each of one or more process steps (R1, R2, R3) of a process phase (PR), in which a substrate (2) is located in the process chamber (3), and in accordance with first control data that is varied by a user, setting at least one process temperature (T) and one process pressure (P), and feeding a process gas flow (Q) into the process chamber (3); before or after the process phase (PR), and during each of one or more calibration or conditioning steps (C1.1, C1.2, C1.3) of a calibration or conditioning phase (PC, PC′), setting at least one calibration or conditioning temperature (T) and one calibration or conditioning pressure (P), and feeding a calibration or conditioning gas flow (Q) into the process chamber (3) in accordance with second control data that is not varied by the user; determining a current fingerprint from data measured by a plurality of sensors during the calibration or conditioning phase (PC, PC′); and comparing the current fingerprint with a historical fingerprint.

2. A device with a chemical vapor deposition (CVD) reactor (1), and a controller (10) storing a formula that causes the controller (10) to: during each of one or more process steps (R1, R2, R3) of a process phase (PR), in which a substrate (2) is located in a process chamber (3) of the CVD reactor, and in accordance with first control data that is varied by a user, set at least one process temperature (T) and one process pressure (P), and feed a process gas flow (Q) into the process chamber (3); before or after the process phase (PR), and during each of one or more calibration or conditioning steps (C1.1, C1.2, C1.3) of a calibration or conditioning phase (PC, PC′), set at least one calibration or conditioning temperature (T) and one calibration or conditioning pressure (P), and feed a calibration or conditioning gas flow (Q) into the process chamber (3) in accordance with second control data that is not varied by the user; determine a current fingerprint from data measured by a plurality of sensors during the calibration or conditioning phase (PC, PC′); and compare the current fingerprint with a historical fingerprint.

3. The method of claim 1, wherein the one or more calibration or conditioning steps have first calibration or conditioning steps (C1.1, C1.2, C1.3) and second calibration or conditioning steps (C2.1, C2.2), wherein the current fingerprint is determined from data measured by the plurality of sensors during the first calibration or conditioning steps (C1.1, C1.2, C1.3), and wherein third control data of the first calibration or conditioning steps (C1.1, C1.2, C1.3) are stored in an invariable manner in the controller (10).

4. The method of claim 1, wherein the current fingerprint comprise one or more of cooling water temperatures, temperatures of temperature control baths (14), flow rates, temperatures of pumps, gas lines or fluid lines, temperatures measured in an exhaust gas flow or control cabinet exhaust air (17), or gas concentration values.

5. The method of claim 1, wherein the one or more calibration or conditioning steps (C1.1, C1.2, C1.3) comprises a purge step, in which a purge gas is fed into the process chamber (3), wherein the purge gas contains a halogen, chlorine, a hydride, or ammonia.

6. The method of claim 1, wherein one or more of the current fingerprint or the historical fingerprint are obtained by a statistical evaluation of measured data, and wherein the statistical evaluation comprises computing one or more of statistical mean values, minimum values, maximum values, and standard deviations.

7. The method of claim 1, wherein the one or more calibration or conditioning steps (C1.1, C1.2, C1.3) comprises a temperature control step, in which a temperature control gas, or hydrogen, is fed into the process chamber (3) at a temperature in a range between 700 and 1200° C.

8. The method of claim 1, further comprising performing a maintenance event (W) prior to the calibration or conditioning phase (PC), during which ambient air is introduced into the process chamber (3), wherein the current fingerprint comprises a thermal fingerprint.

9. The method of claim 1, wherein the comparison of the current fingerprint with the historical fingerprint is carried out in accordance with a rule-based decision system.

10. The method of claim 1, wherein at least one value of the fingerprint is calculated from at least one of: (i) a series of measurements obtained successively over time, and (ii) a derivative with respect to time formed from the data measured by the sensors.

11. A method for determining a plurality of values for a plurality of parameters for controlling a temperature of a surface of a substrate (2) supported by a susceptor (4) of a chemical vapor deposition (CVD) reactor (1) to a prescribed substrate temperature, in order to thermally treat the substrate (2) at the prescribed substrate temperature in at least one process step (R1) of a process phase (PR), wherein the plurality of parameters includes a first parameter (TS, LS) that is a heat input parameter, which affects a first heat flux (H1, H2) supplied from a heat source (6) to the susceptor (4), and a second parameter (TS, QC) that is a heat release parameter, which affects a second heat flux (H3, H4) dissipated from the surface of the substrate (2) to a heat sink (22), the method comprising: in a calibration phase (PC′) preceding the process phase (PR), determining a plurality of tuples in a plurality of calibration steps (C1.1, C1.2, C1.3), wherein each of the tuples includes a first value of the first parameter (TS, LS), a second value of the second parameter (TS, QC), and an actual temperature (TW) of the substrate surface that is obtained when the CVD reactor (1) is controlled with the first and second values; and forming a function (F), associating the actual temperature (TW) with at least one of the first or second parameters, from the plurality of tuples by means of an interpolation, from which function a temperature value (TW1) is obtained, which is correlated with the actual temperature (TW) of the substrate surface that comes closest to the prescribed substrate temperature.

12. The method of claim 11, wherein the heat input parameter is a susceptor setpoint temperature, against which a first control loop (23) regulates a susceptor actual temperature (T.sub.S) by variation of a heating power (LS) fed into a heating device (6), or regulates the heating power (LS), and wherein the heat dissipation parameter is one of (i) a setpoint temperature of a cooling device (22), (ii) a mixing ratio of a temperature control gas consisting of two gases with different heat conduction properties, which is fed into a gap (20) between a cooling device (22) and a process chamber ceiling (18), or (iii) a process chamber ceiling setpoint temperature, against which a second control loop (24) regulates a process chamber ceiling actual temperature (T.sub.C).

13. The method of claim 11, wherein in order to carry out the interpolation, a one-dimensional or multidimensional function is formed.

14. The method of claim 1, wherein the one or more calibration or conditioning steps comprise one or more of: a plurality of calibration or conditioning steps that are performed one after another; and a plurality of calibration or conditioning steps that are sequentially carried out at temperatures increasing or decreasing in a stepwise manner, and/or with changing cooling parameters.

15. The method of claim 1, wherein the measured data comprise a substrate temperature (TW) and a process chamber ceiling temperature (TC), wherein during the calibration or conditioning phase (PC, PC′), a temperature of the susceptor (TS) and thermal conductivities of gap regions are successively modified so as to control a heat flux (H1, H2, H3, H4) from a heating device (6) to a cooling device (22).

16. The method of claim 1, wherein the calibration or conditioning steps (C1.1, C1.2, C1.3) are performed with process parameters that are held constant throughout the calibration or conditioning steps (C1.1, C1.2, C1.3), and the current fingerprint is determined from only data measured during the calibration or conditioning steps (C1.1, C1.2, C1.3).

17. (canceled)

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In what follows, the invention is explained in more detail by means of example embodiments. Here:

[0034] FIG. 1 shows a schematic of a CVD-reactor,

[0035] FIG. 2 shows a first example embodiment of a formula,

[0036] FIG. 3 shows a second example embodiment of a formula,

[0037] FIG. 4 shows a graphical illustration of a fingerprint relating to the first example embodiment in FIG. 2,

[0038] FIG. 5 shows an illustration as in FIG. 4 relating to the second example embodiment in FIG. 3,

[0039] FIG. 6 shows a schematic illustration of a cross-section through a process chamber to illustrate the heat flux H1, H2, H3, H4 from a heating device 6 to a cooling device 22, and temperatures TC, TS and TW measured within the process chamber,

[0040] FIG. 7 shows an illustration of a temperature curve T against time t in four successive conditioning steps C1.1, C1.2, C1.3 and C1.4, during which a susceptor temperature TS is increased in a stepwise manner from about 700° C. to 1200° C., and a substrate temperature TW and a ceiling temperature TC are measured,

[0041] FIG. 8 shows a first example of how tuples K1 to K16 of a fingerprint are derived from the measured values M1 to M16 and N1 to N16, obtained during the conditioning process as in FIG. 7,

[0042] FIG. 9 shows schematically how tuples K1 to K4 of a fingerprint are compared with tuples K1′ to K4′ of a historical fingerprint,

[0043] FIG. 10 shows an illustration of a two-dimensional function in terms of the susceptor temperature TS and the process chamber ceiling temperature TC, calculated from a multiplicity of value tuples by means of interpolation, wherein the function F represents the surface temperatures TW of the substrate, and

[0044] FIG. 11 shows an illustration of a standard calibration curve 25, and a calibration curve 25′ determined in the calibration phase in a coordinate system, which on the X-axis specifies the process chamber ceiling temperature TC, and on a Y-axis specifies the surface temperature TW of the substrate.

DETAILED DESCRIPTION

[0045] FIG. 1 shows schematically a CVD-reactor, as it is part of the device in accordance with the invention, and on which the method in accordance with the invention can be practiced.

[0046] A process chamber 3 is located in an externally gas-tight housing of the CVD-reactor 1, which in particular is made of stainless steel. The process chamber is located above a susceptor 4 made of graphite, or coated graphite, which can be heated from below by means of an IR-heating device 6. The alternating electromagnetic fields generated by the heating device 6 produce eddy currents in the susceptor 4, which lead to the heating of the susceptor 4. On the upper face of the susceptor 4 lie one or a plurality of substrates 2, which are to be coated during a process carried out within the process chamber 3. Carrier gases or process gases can be fed into the process chamber 3 through a gas inlet 5. A vacuum pump 12 is provided, in front of which is arranged a throttle valve 11. A controller 10 can be used to adjust gas flows of the gas sources, designated 7, 8 and 9, which are fed into the process chamber 3 by means of the gas inlet 5. The controller 10 can also be used to control the heating device 6, the throttle valve 11, and the pump 12.

[0047] A fluid source 13, which contains, for example, an organometallic compound, is located in a temperature control bath 14, the temperature of which can be monitored. A process chamber ceiling 18 has a cooling device (not shown), and means for influencing its heat conduction properties. The whole device is located in a cabinet 15, which can essentially be sealed in a gas-tight manner from the environment. The cabinet 15 has incoming air 16 and exhaust air 17. The gas composition in the exhaust air 17 and the exhaust air temperature can be measured.

[0048] A cooling flow can pass through the cavities of the spiral-form heating device 6, the temperature of which is monitored by sensors.

[0049] Formulae that differ from one another can be stored in the controller 10; these can be used to deposit various layers on a substrate 2.

[0050] In each case, a process comprises a multiplicity of steps, wherein the steps can be divided into phases that differ from one another. In a conditioning phase PC, which can, for example, be carried out before the actual coating phase, the process chamber 3, that is to say, the reactor system, is brought up to a setpoint state. The conditioning steps C1.1, C1.2 and C1.3 are carried out with conditioning parameters that differ from one another. During the different conditioning steps C1.1, C1.2, C1.3, for example, the temperatures T within the process chamber 3, or the total pressures P within the process chamber 3, or the gas flows Q (quantity and/or quality) of the conditioning gases, can differ. However, in any process carried out with the CVD-reactor 1, the control parameters SP specified for the control of the temperature, the total pressure, or the gas flows, are fixed and cannot be modified by the user.

[0051] In the example embodiment shown in FIG. 2, the conditioning phase has three conditioning steps, each of which must be carried out with invariable control parameters.

[0052] In the example embodiment shown in FIG. 3, the conditioning phase PC has, in addition to the first conditioning steps C1.1, C1.2 and C1.3, which must be carried out with fixedly prescribed control parameters, second conditioning steps C2.1, C2.2, which can be carried out with variable control parameters. In the conditioning steps C2.1, C2.2, the user can modify the control parameters.

[0053] The conditioning phase PC is followed by a process phase PR, which has various successive process steps R1, R2 and R3, which serve to treat substrates arranged in the process chamber 3, for example to coat the substrates 2 with one or a plurality of layers. The control parameters of the process steps R1, R2 and R3 can be varied by the user.

[0054] In accordance with the invention, measured values are collected during all steps of the conditioning phase PC and the process phase PR; these are determined with measuring sensors. The measured data are stored in a memory of the controller 10, or in a storage system.

[0055] In accordance with the invention, fingerprints are formed only from the measured values that are obtained during the invariable first conditioning steps C1.1, C1.2, C1.3. For this purpose, the measured values are combined with one another in a suitable manner such that at least one value forming the fingerprint, or a group of values forming the fingerprint, are generated. These values can be evaluated statistically. For example, a minimum value, a maximum value, and a standard deviation, can be formed. In particular, this is done by using measured values from a multiplicity of processes carried out in the past, which were considered to be sound.

[0056] In general terms, a fingerprint can be an individual value. Preferably, however, the fingerprint consists of a multiplicity of values, which have been obtained from a multiplicity of measured values. The measured values can, for example, be coolant temperatures, exhaust air temperatures, temperatures of the temperature control bath 14, pump temperatures, gas flow rates or pressures. From these measured values, series of measurements can be formed during a conditioning step. Statistical values can be generated from these measurement series, for example a minimum, maximum, mean, and standard deviation. The statistical values can be part of the fingerprint.

[0057] A fingerprint of a current process, determined according to the same rules, can be compared with a historical fingerprint, which has been formed from the evaluation of a multiplicity of historical processes. Here, for example, a check can be made as to whether the current fingerprint, that is to say, the values embodying the fingerprint, lie in a permissible value window.

[0058] The measured values used to form the current or the historical fingerprint can take the form of measured values that are determined inside the process chamber 3, for example temperatures that are determined on the walls, ceilings, or other regions of the process chamber 3. However, they can also take the form of measured values that are determined outside the process chamber 3, for example in an exhaust gas flow. Here, the exhaust gas temperature, or gas concentrations in the exhaust gas, can be measured. Furthermore, it can be envisaged that pump temperatures, valve positions, or actual gas flows, are used as measured values to form the fingerprint. The temperature of the coolant flowing through the heating coil 6 can also, for example, be used as a measured value.

[0059] Some of the gas sources 7, 8, 9 can be arranged in temperature control baths. The temperatures of these temperature control baths can be used as measured values to form the fingerprint.

[0060] The device can have a control cabinet, for example a cabinet in which electrical components are arranged, or in which a gas mixing system, or a loading or unloading device for the CVD-reactor, is arranged. A sensor can be provided to measure a characteristic control cabinet temperature in the latter. This temperature can also be used in the determination of the fingerprint.

[0061] The system can include cooling water circuits, with which, for example, a process chamber ceiling, or the reactor housing 1, is cooled. The cooling water temperature can also be used in the formation of the fingerprint.

[0062] It is essential that the parameters of the conditioning steps that are used for the historical fingerprint are identical to the parameters of the conditioning steps, from whose measured values the current fingerprint is determined.

[0063] With the above-described method, the current fingerprint can be evaluated in accordance with prescribed statistical rules. In particular, a check can be made as to whether information needs to be provided to the user so that the latter can plan maintenance and servicing measures. The comparison of the current fingerprint with the historical fingerprint is thus carried out according to a rule-based decision system. Depending on the rule, only data recorded on the last occasion, or also data from conditioning phases far in the past, are taken into account in the evaluation.

[0064] The rules used in the formation of the fingerprints, and also in the comparison of current fingerprints with historical fingerprints can be: [0065] Univariate and multivariate value range and limit value checks: [0066] e.g. mean value x outside the interval [y, z], [0067] standard deviation a>b, [0068] mean value a inside the interval [b, c], and mean value x outside the interval [y, z], [0069] Checking of the change from the previous conditioning process: [0070] e.g. mean value [n]<mean value [n−1]*0.9 [0071] Value range check based on a sliding window (with variable window widths) over the previous conditioning processes: [0072] e.g. standard deviation [n] outside the interval (standard deviation [n−1 . . . n−10]−0.5; standard deviation [n−1 . . . n−10]+0.5) [0073] Value range and limit value checks of an extrapolated value on the basis of historical data.

[0074] In the case of historical statistical data recorded before and after a maintenance or servicing measure, a difference can be found that can result in a rule violation and a potential false alarm.

[0075] To prevent this, a rule can be defined in such a way that, when considering historical data (e.g. a moving average), the only data taken into account is that which was carried out after the point in time of the last maintenance event.

[0076] The system executing the rules receives the information as to when a maintenance event has been carried out from a higher-level production control system.

[0077] As stated above, the typical production cycle has alternating process phases and conditioning phases. In a first variant of the invention, a multiplicity of values are obtained in conditioning steps of that conditioning phase, which, for example, has been recorded by means of a permanent measurement during the conditioning step. For example, temperatures, flows, or pressures, can be measured over a longer period of time during the at least one first conditioning step. To form a characteristic fingerprint from these measured values, temperature mean values, standard deviations, minima and maxima, are calculated from these measurements. These statistical data then form a fingerprint, so that the fingerprint can have a multiplicity of statistical data from various measured values. By comparing these fingerprints with historical fingerprints, the current state of the coating plant can be characterized. Here it can be envisaged that the historical data only comprises data that has been obtained, for example, from the last ten conditioning phases in the past.

[0078] It can be envisaged that the process chamber must be opened for the replacement of replacement parts, or also for other reasons, and in any event for maintenance purposes. When the process chamber is opened in this way, ambient air can enter the process chamber, so that moisture contained in the air can adsorb onto walls of the process chamber. In order to condition the process chamber after a maintenance event, the process chamber is heated to high temperatures under near-vacuum conditions, or during a feed of hydrogen into the process chamber, which is pumped out again by means of a pumping device, wherein such temperatures lie in the range of 700° C. and 800° C. The heating can take place in a plurality of steps. The heating takes place according to process parameters that are prescribed in the controller 10, and, in particular, cannot be modified by the system operator.

[0079] Instead of a conditioning phase, however, a calibration phase can also be carried out before the process phase. In particular, however, it is also envisaged that both a calibration phase and a conditioning phase will be carried out before a process phase. In a calibration phase, a plurality of value tuples are determined in one or a plurality of calibration steps, by variation of the susceptor temperature TS and the process chamber ceiling temperature TC, wherein each value tuple has a value of the susceptor temperature TS, a value of the process chamber ceiling temperature TC, and a measured value of the surface temperature TW of the substrate 2.

[0080] FIG. 6 shows a schematic cross-section through a CVD-reactor, in which a heating device 6 generates a heat flux H1 into the susceptor 4. A substrate holder 19 is located in a pocket of the susceptor, which holder is supported by a gas cushion, which is generated by a purge gas flow QS. By this means a gap 21 forms between the floor of the pocket of the susceptor 4 and the lower face of the substrate holder 19, through which gap 21 a second heat flux H2 flows. The heat flows through the substrate holder 19 and through the substrate 2 resting on the substrate holder 19. H3 denotes a heat flux, which flows from the surface of the substrate 2 towards the process chamber ceiling 18. The process chamber ceiling 18 is spaced apart from a cooling device 22 by a gap 20. A temperature control gas is located in this gap 20. The temperature control gas is formed by the purge gas QC, which is fed into the gap 20, wherein the purge gas QC is a mixture of gases with mutually differing thermal conductivities, for example H.sub.2 and N.sub.2. The heat flux H4 through the gap 20 can be influenced by variation of the composition of the purge gas QC.

[0081] Above the process chamber ceiling 18 is located the cooling device 22, which is cooled to a setpoint temperature by means of a cooling fluid. The coolant temperature can be measured, and also used in the formation of the fingerprint.

[0082] By balancing the heat conduction properties of all components of the heat fluxes H1, H2, H3, H4, the temperature of the surface of the substrate 2, namely the substrate temperature TW, and the temperature of the process chamber ceiling, namely the process chamber ceiling temperature TC, can be influenced.

[0083] Reference number 23 denotes a first temperature regulator, which regulates the susceptor temperature TS measured on the lower face of the susceptor 4 against a setpoint value T.sub.So. This is done by influencing the heating power LS fed into the heating device 6. The susceptor temperature TS, or the heating power LS, forms a heat input parameter, with which a heat flux H1, H2 from the heating device 6 to the substrate 2 is influenced.

[0084] Reference number 24 denotes a second control loop, with which the process chamber ceiling temperature TC is regulated against a setpoint value T.sub.Co. This can be done by influencing the coolant temperature of the cooling device 22. However, this can also be done by variation of the mixing ratio of the temperature control gas QC that is fed into the gap 20. The temperature control gas consists of a mixture of a gas having a high thermal conductivity, for example H.sub.2, and a gas having a low thermal conductivity, for example N.sub.2. The process chamber ceiling temperature TC, or the mixing ratio of the temperature control gas QC, or the cooling performance of the cooling device 22, form a heat dissipation parameter.

[0085] FIG. 7 is a temperature/time diagram. FIG. 7 shows four first calibration or conditioning steps C1.1, C1.2, C1.3 and C1.4, successively carried out. During these calibration or conditioning steps C1.1, C1.2, C1.3 and C1.4, the susceptor temperature TS, which is measured on the lower face of the susceptor 4, is increased in a stepwise manner from about 750° C. to about 1200° C. The substrate temperature TW and the process chamber ceiling temperature TC are measured in parallel.

[0086] It can be seen that the measured values N1, N2, N3, N4, NS, N6, N7, N8, N9, N10, N11, N12, N13, N14, N15, N16 of the substrate temperature TW are lower than the susceptor temperature TS, which is used to regulate the process temperature.

[0087] The lower curve in FIG. 7 shows the profile of the process chamber ceiling temperature TC, wherein the mixing ratio of the purge gas QC fed into the gap 20 is varied successively in time in each of the four calibration or conditioning steps. The measurement M1 is a temperature measurement of the process chamber ceiling temperature, in which the H.sub.2 content in the H.sub.2/N.sub.2 mixture of the purge gas QC is 95 percent. The measurement M2 was obtained at a mixing ratio of 35 percent, the measurement M3 at a mixing ratio of 65 percent, and the measurement M4 at a mixing ratio of 5 percent H.sub.2. A total of 16 temperature values TC were obtained, and 16 temperature values TW were obtained during the four calibration or conditioning steps. In each of the calibration or conditioning steps, the four above-designated purge gas compositions were sequentially injected into the gap 20 of the process chamber ceiling 18. Each of the measurements M1 to M16, and N1 to N16, was taken over a period of a number of seconds, for example 20 seconds. During the measurement, mean values and other statistical data were obtained. The measured values N1 to N16, M1 to M16, and any additionally calculated statistical data, represent a thermal fingerprint of the CVD-reactor.

[0088] In addition, the timewise gradients, that is to say the first derivatives with respect to time, can also be calculated from the individual measurements obtained during the measurement sequence in question; these are statistically evaluated in the above-described manner. Statistical data can also be calculated from the said derived values; these are included in the thermal fingerprint. The thermal fingerprint thus obtained can be compared with one or a plurality of historical fingerprints, or an averaged historical fingerprint. FIG. 8 shows tuples of measured values K1, K2, K3, K4, up to K16, representing the measurements (M1,N1), (M2,N2) . . . (M16,N16). The process chamber ceiling temperature TC of the respective measurement M1, M2 . . . is shown on the X-axis. The measured substrate temperature TW is plotted on the Y-axis. It is thus the measured values N1, N2 . . . that here determine the vertical positions of the points K1, K2 . . . . The solid line is a regression curve, and can be compared with a dotted line that represents the historical fingerprint.

[0089] FIG. 9 shows a similar presentation to FIG. 8, but here, instead of a linear regression curve, a quadratic regression curve is drawn through the points K1, K2, K3 and K4. The points K1, K2, K3, K4 on the curve can be compared with the historical measurement points K1′, K2′, K3′, K4′, wherein a separation distance D.sub.1 is determined for the susceptor temperature TW, and a separation distance D.sub.2 is determined for the process chamber ceiling temperature TC. These separation distances D.sub.1, D.sub.2 can be compared with maximum values. If the separation distances D.sub.1, D.sub.2 exceed the specified maximum values, this is an indication that something has been altered in the state of the CVD-reactor. However, if the separation distances D.sub.1, D.sub.2 are within prescribed value windows, this is interpreted to mean that the CVD-reactor is located in a setpoint state.

[0090] The method shown in FIG. 7 is also used to determine the thermal characteristics of a CVD-reactor 1, and in particular to calibrate a characteristic curve. In a calibration phase, which is carried out before a process phase, in which a substrate 2 is thermally treated, in particular coated, in one or a plurality of process steps, parameters are determined, with which a prescribed substrate temperature TW can be set. The steps represented in FIG. 7 by C1.1, C1.2, C1.3 and C1.4 form calibration steps, in which the mixing ratios of the temperature control gas QC are modified at a respectively fixed value of a susceptor temperature TS. The mixing ratio and a cooling performance of the cooling device 22 form further parameters. This is repeated for a multiplicity of different susceptor temperatures TS. In each case, the measured values determined in this manner form tuples, which include the following elements: the set, and in particular regulated, susceptor temperature TS, the mixing ratio of the temperature control gas, or the regulated temperature TC of the process chamber ceiling and, as required, in addition the cooling performance of the cooling device 22 as well as, further heat input parameters or heat dissipation parameters, as required, influencing the heat flux into the substrate 2 and the heat dissipation from the substrate.

[0091] For simplicity, FIG. 10 shows the influence of only two parameters, namely the susceptor temperature TS, and the process chamber ceiling temperature TC, on the substrate temperature TW. The substrate temperature TW is there shown as a two-dimensional function F over the two arguments TC and TS. The function F is calculated by means of a surface (two-dimensional) interpolation over the grid points formed by the measured value tuples. From the “hilly surface” obtained in this way, the point can be determined that corresponds to the value TW1 that comes closest to a prescribed substrate temperature, that is to say, corresponds to a prescribed substrate temperature. The said value TW1 corresponds to a value TC1 of a process chamber ceiling temperature TC, and to a value TS1 of a susceptor temperature TS.

[0092] In a process step of a process phase carried out after the calibration phase, in which the substrate temperature TW is to reach the said value TW1, the susceptor 4 is regulated to the susceptor temperature TS1, and the process chamber ceiling 18 is regulated to the said temperature TC1.

[0093] FIG. 11 shows a standard characteristic curve 25, as it is prescribed by a formula. The point 26 is the point at which, under standard conditions, a process chamber ceiling temperature TC1, at a prescribed susceptor temperature TS, leads to a substrate temperature TW1. The standard characteristic curve 25 is one curve of a set of curves, wherein each curve of the set of curves corresponds to a different susceptor temperature TS.

[0094] The reference number 25′ denotes a corrected characteristic curve. This corrected characteristic curve 25′ has been determined by means of the above-described calibration method. The corrected characteristic curve 25′ is also one curve of a set of curves. The set of curves has a multiplicity of curves, which in each case have been recorded at one of the different susceptor temperatures TS, that is to say, in each case in a calibration step C1.1, C1.2, C1.3 or C1.4. From the measured values recorded in the individual calibration steps C1.1, C1.2, C1.3 and C1.4, the curve of the function F, which represents the corrected characteristic curve 25′, has been determined by means of an interpolation. The inverse function can be formed from this function F, so that the current process chamber ceiling temperature TC2 can be directly determined for a prescribed substrate temperature TW1 from the point 26′ on function F.

[0095] It is considered advantageous, amongst other items, that value tuples are determined during the calibration phase by the selective variation of parameters that influence the heat flux into the substrate and the heat dissipation from the substrate. The parameters are, for example, the susceptor temperature TS, and/or the process chamber ceiling temperature TC. The value tuples contain the respectively measured substrate temperature as a further element. By means of interpolation, a multi-dimensional function, or a set of one-dimensional functions, is determined from these value tuples. By means of an interval nesting, a Taylor expansion, or other suitable mathematical, in particular numerical, methods, on the basis of these functions, or sets of functions, the parameter set is determined that corresponds to a substrate temperature TW that comes closest to, or corresponds to, a substrate temperature desired in a process step.

[0096] It is considered particularly advantageous if the calibration steps, carried out to determine the calibration function F, are part of a conditioning phase.

[0097] The above statements serve to explain the inventions covered by the application as a whole, which inventions also autonomously advance the prior art at least by means of the following combinations of features in each case, wherein two, a plurality, or all, of these combinations of features can also be combined, namely:

[0098] A method, which is characterized in that the fingerprint also comprises, or only comprises, such values, or groups of values, which are obtained from measured values, which are recorded during one or a plurality of calibration or conditioning steps C1.1, C1.2, C1.3 of a calibration or conditioning phase PC, PC′, in which in each case at least a calibration or conditioning temperature T and a calibration or conditioning pressure P are set, and a calibration or conditioning gas flow Q is fed into the process chamber 3, in accordance with control data supplied by the formula.

[0099] A device, which is characterized in that the fingerprint also comprises, or only comprises, such values, or groups of values, which are obtained from measured values, which are recorded during one or a plurality of calibration or conditioning steps C1.1, C1.2, C1.3 of a calibration or conditioning phase PC, PC′, in which in each case at least a calibration or conditioning temperature T and a calibration or conditioning pressure P are set, and a calibration or conditioning gas flow Q is fed into the process chamber 3, in accordance with control data supplied by the formula.

[0100] A method or a device, which is characterized in that the calibration or conditioning steps have first calibration or conditioning steps C1.1, C1.2, C1.3, and second calibration or conditioning steps C2.1, C2.2, wherein the fingerprint comprises, only such values, or a group of values, which are obtained from measured values recorded during the first calibration or conditioning steps C1.1, C1.2, C1.3, wherein the control data of the first calibration or conditioning steps C1.1, C1.2, C1.3 are stored in an invariable manner in the controller 10.

[0101] A method or a device, which is characterized in that the measured values are values of physical quantities, which are measured inside the process chamber 3, or outside the process chamber 3, and in particular are cooling water temperatures, temperature control bath temperatures of temperature control baths 14, flow rates, temperatures of pumps, gas lines or fluid lines, temperatures or gas concentration values measured in an exhaust gas flow, or control cabinet exhaust air 17, or the like.

[0102] A method or a device, which is characterized in that at least one of the one or plurality of first calibration or conditioning steps C1, C2, C3, C4 is a cleaning step, in which a cleaning gas, which contains, in particular, a halogen, for example chlorine, or a hydride, for example ammonia, is fed into the process chamber 3.

[0103] A method or a device, which is characterized in that the fingerprint, in particular the historical fingerprint, is obtained by a statistical evaluation of measured data, in particular for older processes, wherein statistical mean values, minimum values, maximum values, and/or standard deviations, are determined.

[0104] A method or a device, which is characterized in that at least one of the one or plurality of first conditioning steps C1.1, C1.2, C1.3 is a temperature control step, in which a temperature control gas, for example hydrogen, is fed into the process chamber at an elevated temperature, in particular in a range between 700 and 1200° C.

[0105] A method or a device, which is characterized in that a conditioning phase PC is carried out before or after each process phase PR, during which a fingerprint is obtained.

[0106] A method or a device, which is characterized in that a conditioning step PC, with one or a plurality of first conditioning steps C1.1, C1.2, C1.3, is carried out after a preceding maintenance event W, and/or in that a calibration or conditioning step PC, with one or a plurality of first calibration or conditioning steps C1.1, C1.2, C1.3, is carried out after a preceding maintenance event W, during which ambient air has entered the process chamber 3, wherein the fingerprint obtained from the one or a plurality of measured values is a “thermal fingerprint”.

[0107] A method or a device, which is characterized in that the comparison of the current fingerprint with the historical fingerprint is carried out in accordance with a rule-based decision system.

[0108] A method or a device, which is characterized in that at least one value of the fingerprint is calculated from a series of measurements obtained successively over time, wherein in particular it is envisaged that a derivative with respect to time is formed from the measured values.

[0109] A method, which is characterized in that in a calibration phase, which in time precedes the process phase, in a plurality of calibration steps C1.1, C1.2, C1.3 a multiplicity of value tuples is determined, in each case having a value of the first parameter TS, LS, a value of the second parameter TS, QC, and an actual temperature TW of the substrate surface occurring at these values, wherein from the multiplicity of value tuples a function F representing the actual temperature TW by way of at least one of the parameters is formed by means of an interpolation, from which function a value TW1 of the at least one parameter TC, QC; TS, LS is obtained, which is correlated with an actual temperature TW of the substrate surface that comes closest to a prescribed substrate temperature.

[0110] A method, which is characterized in that the heat input parameter is a susceptor setpoint temperature T.sub.So, against which a first control loop 23 regulates a susceptor actual temperature Ts by variation of a heating power LS supplied to a heating device 6, or the heating power LS, and/or that the heat dissipation parameter is the setpoint temperature of a cooling device 22, the mixing ratio of a temperature control gas consisting of two gases with different thermal conductivity properties, which is fed into a gap 20 between a cooling device 22 and a process chamber ceiling 18, or a process chamber ceiling setpoint temperature T.sub.So, against which a second control loop 24 regulates a process chamber ceiling actual temperature Tc.

[0111] A method, which is characterized in that, in order to carry out the interpolation, a one-dimensional or multi-dimensional function is formed, whose grid points form the value tuples.

[0112] A method, which is characterized in that a plurality of first calibration or conditioning steps directly follow one another, wherein in particular it is envisaged that the plurality of calibration or conditioning steps are carried out at stepwise temperatures increasing or decreasing in a stepwise manner, and/or with changing cooling parameters.

[0113] A method, which is characterized in that the measured values comprise a substrate temperature TW and a process chamber ceiling temperature TC, wherein in the calibration or conditioning phase PC, temperatures TS and thermal conductivities are successively modified in a heat flux H1, H2, H3, H4 from a heating device 6 to a cooling device 22.

[0114] All disclosed features are essential to the invention (individually, but also in combination with each other). The disclosure of the application hereby also includes the full disclosure content of the associated/attached priority documents (copy of the previous application), also for the purpose of including features of these documents in the claims of the present application. The subsidiary claims, even without the features of a claim referred to, characterise with their features independent inventive developments of the prior art, in particular in order to make divisional applications on the basis of these claims. The invention specified in each claim can additionally have one or a plurality of the features specified in the above description, in particular those provided with reference numerals, and/or in the list of reference numerals. The invention also relates to forms of design, in which individual features cited in the above description are not realized, in particular to the extent that they can recognisably be dispensed with for the respective intended use, or can be replaced by other means having the same technical effect.

LIST OF REFERENCE SYMBOLS

[0115] 1 CVD-reactor [0116] 2 Substrate [0117] 3 Process chamber [0118] 4 Susceptor [0119] 5 Gas inlet [0120] 6 Heating device [0121] 7 Gas source [0122] 8 Gas source [0123] 9 Gas source [0124] 10 Controller [0125] 11 Throttle valve [0126] 12 Vacuum pump [0127] 13 Fluid source [0128] 14 Temperature control bath [0129] 15 Cabinet [0130] 16 Incoming air [0131] 17 Exhaust air [0132] 18 Process chamber ceiling [0133] 19 Substrate holder [0134] 20 Gap [0135] 21 Gap [0136] 22 Cooling device [0137] 23 First control loop [0138] 24 Second control loop [0139] 25 Standard characteristic curve [0140] 25′ Corrected characteristic curve [0141] 26 Point [0142] 26′ Point [0143] D.sub.1 Separation distance [0144] D.sub.2 Separation distance [0145] t Time [0146] C1.1 First calibration/conditioning step [0147] C1.2 First calibration/conditioning step [0148] C1.3 First calibration/conditioning step [0149] C2.1 Second calibration/conditioning step [0150] C2.2 Second calibration/conditioning step [0151] F Function [0152] H1 Heat flux [0153] H2 Heat flux [0154] H3 Heat flux [0155] H4 Heat flux [0156] K1 . . . K16 Measured tuple or point [0157] K1′ . . . K16′ Historical measured tuple or point [0158] LS Heating power [0159] M1 . . . M16 Measured value [0160] N1 . . . N16 Measured value [0161] p Pressure [0162] PC Conditioning phase [0163] PC′ Conditioning phase [0164] PR Process phase [0165] Q Process gas flow [0166] QC Purge gas [0167] QS Purge gas [0168] R1 Process step [0169] R2 Process step [0170] R3 Process step [0171] TS Susceptor temperature [0172] TC Ceiling temperature [0173] TW Substrate temperature