JPH01264A - Method and apparatus for controlling the deposition of reactive layers on substrates - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The invention relates to the use of a magnetron cathode operated at high efficiency with a target consisting of at least one electrically conductive component of a layer material. controlling the inflow of reactant gas depending on this intensity, and distributing this reactant gas adjacent to the target under the setting of a target value for said intensity; The present invention relates to a method and apparatus for controlling the deposition of a reactive layer onto a substrate by controlling the deposition of a reactive layer on a substrate.

Conventional Technology In the case of this type of coating method, a conductive target, often a metal target, is usually sputtered in a reactive atmosphere in order to be able to supply a direct voltage to the magnetron cathode. Ru. In this case, a layer consisting of the reaction product of the target material is deposited on the substrate, the composition of the layer material being highly dependent on the atmosphere surrounding the magnetron cathode.

Coating processes of this type are distinguished by six different operating conditions for the magnetron cathode: ``metallic operation'', ``reactive operation'' and transitional operation. In the case of "metallic operation", the sputtering surface of the target has metallic, ie electrically conductive, properties. 1. In the case of "reactive operation", there is a corresponding reaction product on the sputtering surface of the target which is generally an electrical insulator and which clearly reduces the sputtering rate of the target material, i.e. the amount sputtered per unit time. A "transitional operation" exists between the two mentioned operating states and can be considered to be highly unstable, since this transitional operation also occurs in one direction. This is because they tend to turn in other directions as well.

This "transitional operation" is a typical characteristic of magnetron cathodes in which the locally confined sputtering process forms increasingly deeper erosion grooves within the sputtering surface. The wall area of the erosion groove can still be kept significantly free of reaction products by means of high sputtering rates, and the remaining target surface is coated with reaction products - hence the unstable operating behavior already described. happens.

In order to prevent or suppress unstable operating behavior as far as possible, the reactive gas is supplied at a position shielded by a shutter, away from the target surface, and an inert gas acts exclusively on sputtering within the target area. It is already known (West German Patent Application No. 3331707). If in the case of such a process the supply of the reaction gas is carried out depending on the strength of one or more spectral lines, the production process,
In particular, very long time constants will result for such control methods which are not acceptable for continuous production processes. Refreshingly, in this particular case, there is a back and forth between "metallic operation" and "reactive operation." This is because, in particular, the decomposition of the reaction products on the sputtering surface of the target lasts for a relatively very long time due to the extremely low sputtering rate in this case.

This back-and-forth between "metallic operation" and "reactive operation" can be represented by a hysteresis curve when a reaction in excess of stoichiometric proportions flows over the reactant gas stream. , was already known (European Patent Application Publication No. 0121019).Moreover, the constructed control method itself does not eliminate the problem of instability in the connection of the mass spectrometer. , because of this, the time constants of the individual parts of the measurement section and the control section do not have to be long.

It becomes a problem to obtain information that has a timely nature.

This layer property can be obtained together with narrow tolerances and high long-term stability. A series of proposals have already been made for controlling the process parameters of magnetron sputtering for certain layer properties using control circuits with sensors. However, all proposals have yielded only very limited results. This means that many layer properties cannot be measured within the coverage zone or can only be measured very inaccurately because the measurement signal is influenced by the plasma. Therefore, for this type of control circuit, measurements were always performed outside the target band. In this method, the time difference between the end of the coating process and the measurement is disadvantageous for achieving narrow tolerances. The time constant must be selected to be significantly larger than the time difference mentioned in connection with the stability of the control circuit, so that large substrate areas are covered with unmeasurable tolerances. This substrate range can be extended over a large number of substrates or over long sheet lengths during sheet coating, so these must not be excluded and discussed. No.

Even more essential than the drawbacks already mentioned when controlling using measured values of layer properties can be seen in the drawback that such control circuits do not guarantee process stability due to the deposition of reactive layers. .

Such process instabilities are caused, for example, by undesired changes in the gas composition and arc discharge at the target, which are also manifested in changes in the properties of the deposited layer. This kind of instability can be caused by “driving like a sparrow”.
The transition of the coating process during the coating of the substrate means complete waste in relation to the layer properties.

The described transition occurs in the case of negative electrical sputtering parameters at any rate upon removal of the critical reactant gas flow.

The height of this flow rate is one measure of the fundamental nature of the reactive process.

If the reaction d inflow force is smaller than the transient violet, the magnetron cathode will generally have the required layer properties within the operating range, e.g. efficiency and reactant gas flow rate. This is achieved by keeping process parameters such as constant, as well as controlling time.

In contrast, within the "metallic range" it is generally not possible to separate layers from the reaction products having the required stoichiometry.

Using special methods for active decoupling of the reactant gases on the target or at the substrate, it is possible and possible to produce such layers at correspondingly low rates (e.g. the method already described). (See West German Publication No. 5651707).

For said method to achieve the required layer properties,
Are the reaction gas flow rates and/or electrical operating parameters kept unchanged? 1fflJ was controlled.

However, if the process is controlled by the measured layer properties, the already mentioned time delays not only limit the layer's optical value by the time defects are measured, but even more importantly the time delay of the target coating. Hysteresis occurs. for example,
The necessary changes in the target coverage that occur when the efficiency changes can result in a time delay, which is one consequence of positioning the measurement location relative to the layer parameters. greater than the interval. This time lag can be between a few seconds and 2.3 minutes, thus resulting in unmeasurable changes in I-temporality.

For economic reasons, layers can generally only be obtained in 711 "transitional operation" with high reaction rates and high precipitation rates of the compounds, which transitional operation is not suitable for unstable or semi-deterministic conditions of the stumbling process. This is a normal operating condition, which has already been further pointed out above. Coating with reaction products during the erosion of the target is still directly inhibited, and the reaction products adhere to the substrate with high reaction rate and high speed.

Work in the case of "transitional operation" requires feminizing the discharge with additional equipment.

According to US Pat. No. 4,166,784, it is already known to carry out a control for detecting the strength of a coat characteristic of a target material in a plasma.

The stabilization of the working point of the reactive discharge is carried out by a control circuit which, depending on the signal of said strength, determines, in the case of a constant reactant gas flow rate, the electrical parameters of the current supply or With electrical operating parameters of , the reaction gas flow rate is controlled as a measured value.

According to the description in German Patent Application Publication No. 239810, the output of a light beam is extracted from a plasma discharge,
and spectrum 4s to implement m control of the approximate gas amount.
It is known to relate a group of spectra to signal formation. Historically, Germany's current patent application announcement No. 23
9811 Akira#l! According to the line description, it is known in what way the reference signal can be found to fix the position of the working point.

However, control circuits for ensuring process stability at a given working point of the discharge under the use of the ejection signal in the reactive gap electrode sputtering process generally have unacceptable tolerances and/or long lengths of the layer properties. It cannot be prevented from migrating over time. To this end, a series of factors exists: When a new substrate surface is introduced into the inspection area, the surface layer is left open by allowing the plasma to act on the substrate. Thereby, in the event of a discharge, the gas composition changes and, as a result, the emission signal also changes in a controllable manner.

- Depopulation processes of device parts after target exchange also cause undesired changes in the emission signal.

The uncontrollability and effects of this emission signal are critical for the above-mentioned conditions over long periods of time in the range of 4 to 4.

Furthermore, all geometrical changes give rise to signs of target erosion, which especially occurs during the coating process, and a disturbing change in the emission signal.

OBJECT OF THE INVENTION It is therefore an object of the invention to describe a method of the above-mentioned type which makes it possible to achieve layer properties with deliberately narrow tolerances in the case of reactive coating methods. It is.

Means for Solving the Problems According to the present invention, a) the intensity of the spectral lines in the plasma is measured between the plasma and the target substrate, and b) the reaction The control is carried out during layer formation with as short a time constant as possible, in which case the target value is adjusted smoothly, depending on at least one property of the finished layer, so that the properties of a given layer remain as unchanged as possible. It is resolved by

In this case, it is essential to measure at least one spectral line in the plasma between the target and the substrate. In this case, optical measurements, ie measurements with a spectrophotometer directed directly at the plasma, are necessary. This means that when using a magnetron cathode, the optical sensor is oriented in the volume element in close proximity to the target surface.

The range in which the measurements are carried out parallel to the target surface lies in particular at a distance of about 10 to 30 m from the target surface.

The precision of such measurements is clearly also distinguishable from measurements with mass spectrometers, in which the volumetric elements of the gas atmosphere are filtered with suction from the coverage area and analyzed at a distance therefrom.

This measurement method has the disadvantage that it is associated with significant time delays and, moreover, the gas composition changes over the course of the conveyance.

It can be recognized that the fact that the reaction controls the amount of inflow and inflow with the shortest possible time constant is related to the amount of reaction gas inflow near the target. It is thus possible to adapt the reaction gas flow rate to slightly changed operating conditions in a very short time.

In this case, the measurement of at least one property of the finished layer is carried out numerically, if necessary, at a clearly longer time.

This is because the substrate with the currently finished layer must be transported from the coating zone to the measurement zone.

That is, in this case, adaptation to the target value is performed at a slow speed. However, experience has shown that this method is quite sufficient to solve the problem.

In other words, in principle, it is necessary to balance the two control circuits; one of them has an extremely short time constant, and the other has an extremely large time constant. has. - The influence of the measurement signal on the plasma side and the influence of the measurement signal on the other side substrate side are different 1
By necessity, of the whole? The stability of IIJ behavior can be clearly improved. That is, for example, it is possible to evaluate the component of the measurement signal from plasma at 60-90% and the component of the measurement signal from substrate measurement at 10-40%.

The method of the present invention, which achieves the required processing stability on one side and the required constancy of layer properties on the other side using fundamentally different equipment, very effectively solves the posed problem. There was found.

When keeping the time constant of the cough control circuit as short as possible,
It has been found that it is possible to make incremental improvements in plasma emission control for process stabilization. However, reducing the time constant of this control circuit can only be achieved if the temporal inertia of supplying the reactant gas flow is overcome.

A short diffusion path of the reactant gas from the inlet into the dense plasma region in the magnetic case over the erosion contour is ensured by the inlet and the inlet as close as possible to the target. This feed inlet, in which the target area and the substrate area are very closely coupled, can lead to a drastic reduction in the basic stability of the discharge if it is not controlled by a Sly-P target value. Probably. The close coupling between the inlet of the reactant gas and the target surface is in conflict with solutions that typically achieve sufficient decoupling between the target and the substrate area in conjunction with a high fundamental stability. circulate.

The short time constant and thus the close return of the measuring signal from the plasma allows stable operation at arbitrarily selected working points within the "transitional range" in connection with the control of the setpoint value in a sly-P manner. Make it familiar. This means, in practice, that a degree of freedom in processing technology is gained, for example in order to achieve layers with adjustable layer properties defined with large variation ranges. That is, for example, the refractive index of the aluminum rat layer can vary within the range of 1.95 to 2.1. When producing indium-tin-oxide layers, the resistivity can be varied by a dimensional degree.

However, if the bond between the target region and the substrate region is further strengthened by controlling the plasma emission, layer properties that are not achievable with hitherto available devices can even be produced. That is, for example, an indium-tin-doped layer has a resistance of 1 x 10-'Ω/cIIL on an unheated substrate, compared to 8 x 10-'Ω/cIIL in the conventional method.
″″Achieved with a surface resistance of 4 Roux or less.

Nevertheless, according to the method of the invention it is possible to maintain a stable discharge, which is explained by the fact that in the case of close alignment the fraction of the reactant gases converted into the desired compounds is relatively increased. This can be recognized in particular. This is because the excitation of the reactive gas also increases in the vicinity of the substrate. On the other hand, the close bonding and the simultaneously reduced fundamental stability result in an enhanced influence of the emitted signal and thus on the layer properties during the desorption process followed by a change in the gas composition.

Furthermore, by closely coupling the partial process in the target on the - side and the partial process on the substrate on the other side by the control method according to the present invention, the influence of the erosion state of the target (in this case, this change causes the sputtering rate to be reduced). displacement and excitation of the reactant gases) are relatively large relative to the layer properties. However, the deviations caused by the two effects of the layer properties have large time constants and, with the inventive combination of the control method, can be effectively prevented by slidingly adjusting the setpoint value. can.

Other preferred embodiments of the method according to the invention are evident from one of the claims 2 to 6.

It is thus particularly advantageous for the control method according to the invention to be carried out only after the conditioning of the target narrow face has taken place.

This means that a new target or f! The purpose is to purify the surface of the target, such as that exposed to air.

In this case, after conditioning the target surface, the reactant gas inflow is programmed in time and is monotonically predetermined until the desired value of at least one layer property is achieved with an approximately constant flow of reactant gas. By increasing the adjustment time, it is advantageous to adjust the process parameters to the required layer properties near the working point.

Furthermore, the actual value for the at least one intensity signal, which is obtained in this case after an excursion of the conditioning time and the regulating time of the reactant gas flow, is compared with a desired value of the layer properties and from this comparison value a first control is determined. This is particularly advantageous when forming target values for circuits.

The intensity of the luminescence, measured on the conditioned 1-carbohydrate, characterizes the defined starting state. The working point in the case of reactive operation, such as achieving the required layer properties, is first characterized by reference values as strength values determined on the basis of preliminary tests.

A similar problem is also solved by a device according to the invention, which comprises: a) a light sensor of a spectrophotometer device directed into the space between the magnetron anode and the substrate; the optical path is oriented parallel to and in close proximity to the target surface; b) a gas distribution device is located adjacent the target surface and coupled to the reactant gas source via a control valve; C) The outlet of the spectrophotometer device is coupled via an amplifier and a control device with a control element for the control valve, in this case a light sensor spectrophotometer device, amplifier, control device, 11tI
the I control element and the control valve form a first control circuit with a short time constant; d) a second sensor is positioned immediately in front of the substrate to be coated for at least one characteristic of the substrate to be coated; In this case the outlet of the second sensor is connected to a second amplifier, the outlet of this second amplifier being connected to a setpoint value transmitter for the control device, in this case the second sensor,
The second amplifier, the setpoint value transmitter, the regulator, the control element and the control valve downstream of this control circuit form a second control circuit with a longer time constant than in the case of the 171tlJ control circuit; The first control circuit and the second control circuit are characterized in that they are connected together in the regulator.

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the subject matter of the invention will now be explained in detail with reference to the drawings.

In the drawing, the #fr plane of the vacuum chamber 1 is shown, through which the substrate 2 is guided along a travel path, which is indicated by dotted lines. In the bottom plate 3 of this vacuum chamber, while maintaining an overall insulation gap shorter than the darkroom gap that should be considered under operating conditions, there is a magnetron holder 4 consisting of the following members: having a front plate 6; The cathode body 5 is inserted into the base plate 3 below the intermediate connection of the insulation material layer 1 . Above the front plate 6 with the cooling channels there is a target 8 made of electrically conductive material, which target is made of one component of the layer material. On the underside of the front panel 6 there is a magnetic system 4, which consists of a ferromagnetic 4 iron and a permanent ferrite placed on the yoke, intercalated with each other, in this case the inner Both present in and both present on the outside have their respective opposite positions, which is represented by the illustrated arrows. At the immediate edge of the target 8 there is a gas distribution device 9, which does not cover the target 8 straight and flat in plan view. The gas distribution device 9 includes a gas distribution device 1 that is attached to the periphery of the target at least in a substantial part of the periphery and has perforations.
It consists of one or two conduits through which the reactant gas can exit in the direction of the target. The gas distribution device 9 is connected via a peripheral valve 10 and a common control valve 11 to a source 12 for the reaction gas or the mixture of reaction gas and working gas (argon). 400-1000
A further voltage source for supplying a negative DC voltage of volts to the magnetron cathode 4 is not shown for the sake of simplicity.

Between the magnetron cathode 4 and the substrate 2, a shielding device 13 is arranged adjacent to the substrate, which can also perform the function of an anode and can be present at the same potential as the vacuum chamber. or can exist with a positive potential present on it. The reaction zone 14 consists essentially of the target 8, the substrate 2 and the shielding device 13?
ffIJ, in which case a reactive process takes place for coating the substrate 2 with a compound consisting of a target material and a reactive gas.

This reaction zone is directed by a light sensor 15 which narrows down the range of the reaction zone 14 by means of a light path running parallel to the target 8 . This optical path exists within a range of 10 to 60 meters above the target 8. A glass fiber cable 16 leads from the optical sensor 15 to a spectrophotometer device 1γ, which can consist of an opto-electronic transformer with an interference filter or a tunable monochromator, for example. It can consist of Thereby, one or more spectra 1 can be separated from the optical signal of the sensor 15 and detected depending on their intensity. As a result, at the output 11a of the spectrophotometer device, a negative signal is characteristic, the negative value of which is proportional to the intensity of the spectral line.

The spectrophotometer arrangement is followed by an amplifier 18 , which is further followed by a control device 19 , which acts on its side on a control element 20 of control valve 11 . This arrangement forms a first control circuit with a very short time constant of less than 150 mB. Thereby, the composition of the gas atmosphere or plasma in the reaction zone 14 can be controlled in the shortest possible time.

In the middle of the substrate 2, behind the outlet of the reaction zone 14, a second sensor 21 is present, which can be an optical sensor and/or an electrical sensor depending on the properties to be measured. can. The output signal of this sensor is fed via a line 22 to another amplifier 23, the output of which is again connected to a microprocessor 24. From this microprocessor, a further line leads to a setpoint value transmitter 25, which is preadjusted by means of a regulating device 26 (X-position meter). Since the sensor 21 detects a certain change in the layer properties of the substrate 2, it is for this reason that the output of the setpoint value transmitter 25 is characterized by a so-called sliding S value, which setpoint value is determined by the conductor 27.
It turns out that the signal is supplied to the A4 shaker 19 via the A4 shaker 19. As a result, the signals in the -1i unit 19 are interconnected in terms of control technology, which signals correspond to the output signals of the sensors 15 and 21. The properties of the finished layer correspond to the speed of movement of the substrate 2. Sensor 210 can only be detected with a time delay, so that the measurement signal corresponding to the layer properties is
This only occurs with a large time delay depending on the time constant of the first control circuit. Generally, the delay will be at least 0.5 seconds. The element 4 of the sensor 21 up to the ml 4 valve 11 via the setpoint output - 25 is a second control circuit with a large time constant, in this case the $1 control circuit from the regulator 19 and the second control circuit. The elements are the same.

Furthermore, a third sensor 28 for detecting the nature of the substrate entering the coating station is arranged immediately before the coating station formed from a magnetron cathode 4 with a reaction zone 14 in the middle of the substrate 2; The sensor is
It is connected to the microprocessor 24 by a third amplifier 29 . In the microprocessor 24 a comparison is made between the signal of the sensor 21 and the signal of the sensor 28, so that only changes in the properties of the substrate in the coating station with the reaction zone 14 are detected. The method is particularly characterized in that there is still one coating station with reaction zone 14.
This is important if one coating station is adjacent, so that the substrate with one or more previous coatings advances into the reaction zone '14.

[Brief explanation of the drawing]

The drawing is a schematic representation of an embodiment of an apparatus according to the invention for controlling the deposition of a reactive layer on a substrate. 2... Substrate, 4... Magnetron anode, 9... Gas distribution device, 11... Control valve, 12... Reaction gas source, 14... Reaction zone, 15... Optical sensor, 17.
... Spectrophotometer device, 11a... Output side, 18...
Amplifier, 19...Adjuster, 20...Control element, 21
...Second sensor, 23...Second amplifier, 24...
・Microprocessor, 25...Target value transmitter, 28
...5th sensor, 29...3rd amplifier.

Claims (1)

Claims: 1. The intensity of at least one spectral line of the target material in the plasma of the sputtering process in the case of a magnetron cathode operated with constant efficiency with a target consisting of at least one electrically conductive component of the layer material. While evaluating the strength, the amount of reactant gas inflow is controlled depending on this strength.
and controlling the deposition of a reactive layer on a substrate by distributing said reactive gas adjacent to a target under the setting of a target value for said intensity, comprising the steps of: a) controlling said spectral lines in a plasma; an intensity measurement is carried out between target and substrate; b) control of the reactant gas inflow is carried out with as short a time constant as possible during layer formation, the target value being dependent on at least one property of the finished layer; A method for controlling the deposition of a reactive layer on a substrate, characterized in that the properties of a given layer are adjusted smoothly so that they remain as unchanged as possible. 2. Amplify the intensity signal in the first control circuit, process it electrically, and control the amount of reactant gas inflow with a time constant of 150 ms or less, particularly 50 ms or less, to the first control circuit. The stated setpoint value is formed in a second control circuit, and the control circuit has a first value obtained from at least one layer characteristic.
2. The method of claim 1, further comprising providing a measurement signal. 3. The method of claim 2, wherein the measurement of at least one layer property is carried out outside the coating zone and is fed to a second control circuit having a time constant of 0.5 seconds or more. 4. Supplying another measurement signal characterizing the substrate properties before the coating process to the second control circuit, comparing the first measurement signal with the second measurement signal and comparing the comparison result to the target value of the first control circuit. 3. The method according to claim 2, wherein the correction value is used as a correction value. 5. After conditioning the target surface, time-programming the reaction gas flow rate and monotonically predetermined adjustment until the target value of at least one layer property is achieved in an approximately defined flow of the reaction gas. 2. The method of claim 1, wherein the process parameters are adjusted to the required layer properties in the vicinity of the working point by increasing the time. 6. Compare the actual measured value for the at least one intensity signal obtained after the expiration of the conditioning time and the regulating time for the flow rate of the reaction gas with the setpoint value for the layer property and from the comparison value setpoint value for the first control circuit. 6. The method according to claim 5, wherein: 7. Evaluating the intensity of at least one spectral line of the target material in the plasma of the sputtering process in the case of a magnetron cathode operated with constant efficiency with a target consisting of at least one conductive component of the layer material; The reaction gas inflow rate is controlled depending on this strength,
and controlling the deposition of reactive groups on a substrate by distributing this reactive gas adjacent to a target under the setting of a target value for said intensity, comprising: a) a spectrophotometer device (17); An optical sensor (15) is connected to the reaction zone (2) between the magnetron cathode (4) and the substrate (2).
14) in which the optical path of the sensor (15) is oriented parallel to and in close proximity to the target surface; b) the gas distribution device (9) is located adjacent to the target surface; and a reactant gas source (12) via a control valve (11).
), and c) the output (17a) of the spectrophotometer device (16) is connected to the control valve (11) via an amplifier (18) and a regulator (19).
), in this case a light sensor (15), a spectrophotometer device (17), an amplifier (1
8), the regulator (19), the control element (20) and the control valve (11) form a first control circuit with a short time constant; d) a second control circuit immediately in front of the substrate (2) to be coated; sensor(
21) is arranged for at least one characteristic of the substrate to be coated and is connected to a second amplifier (23), which furthermore provides a setpoint value transmitter (25) for the regulator (19). in this case the second sensor (21)
, the second amplifier (23), the setpoint value transmitter (25), the regulator (19), the control element (20) and the control valve downstream of this control element have a longer time period than in the case of the first control circuit. reactive groups on the substrate forming a second control circuit with a constant constant, the first control circuit and the second control circuit being connected together in a regulator (19); A device that controls the adhesion of 8. Immediately in front of the coating station formed from the magnetron cathode (4) a third sensor (28) is arranged on the way to the substrate (2) for detecting characteristics of the substrate traveling into the coating station; and 8. Device according to claim 7, connected to the microprocessor (24) via a third amplifier (29).