EP2179268A2

EP2179268A2 - Method for the in-situ determination of the material composition of optically thin layers, arrangements for performance and applications of the method

Info

Publication number: EP2179268A2
Application number: EP08784316A
Authority: EP
Inventors: Raik Hesse; Hans-Werner Schock; Daniel Abou-Ras; Thomas Unold
Original assignee: Helmholtz Zentrum Berlin fuer Materialien und Energie GmbH
Current assignee: Helmholtz Zentrum Berlin fuer Materialien und Energie GmbH
Priority date: 2007-07-20
Filing date: 2008-07-09
Publication date: 2010-04-28
Also published as: DE102007034289B3; US20100291714A1; WO2009012748A3; JP2010533874A; US8338194B2; WO2009012748A2

Abstract

Known methods are based on the measurement of the total reflection and analyze reflections of optical radiation in the case of smooth surfaces. Therefore, they require a normalization by measurement of the reference light or other measurement of the refractive indices. As a result, they cannot be used universally; in particular they cannot be used for process control of evaporation processes. The control methods known in this field operate only qualitatively on the basis of specific characteristics in the production process. For genuine quantative regulation, however, it is essential that the components to be regulated (optical layer parameter) are also made measurable. The method according to the invention directly determines the material composition of optically thin layers and is based on an optical layer model that was derived from electromagnetic conduction theory with field resistances. The deposited layer is irradiated with preferably incoherent spectral light, preferably from a white light source, and the reflection intensities outside total reflection are detected by a spatially resolving optical detector, preferably CCD, and fed into the layer model. The characteristic functions of the layer model are fitted to the real process values and serve for numerically ascertaining the optical layer parameters from which the specific material composition can be derived.

Description

Method for the in-situ determination of the composition of optically thin layers, arrangements for carrying them out and applications of the method.

DESCRIPTION

The invention relates to a method for in-situ determination of the material composition of vapor deposited on a substrate optically thin layers having a interferometrically evaluable correlation between the optical layer parameters (complex refractive index n _kO mpiex with real part: refractive index n _Sc hic _ht and imaginary parts: absorption coefficient κ _Sc hicht and real layer thickness d _Sch ic _ht), as well as arrangements for carrying out the method and applications.

STATE OF THE ART

DE 699 17899 T2 describes the determination of the doping of a silicon wafer by means of reflection spectrometry. Reference and sample measurements are taken and calculated according to known physical calculation methods for the complex refractive index and the thickness of an optically thin layer. The description of the formulas of the optical reflection of radiation is shown in detail, whereby the modeling of the layer to be measured is traceable. However, the exact measurement is possible only ex-situ. The measurement is carried out in the infrared range, since the layers considered are in a thickness range which requires IR measurement in order to be able to evaluate an interferometric wave for determining the thickness. DE 197 23 729 A1 discloses a method for in-situ determination of optical layer constants in plasma and ion beam assisted etching and coating. A method for determining optical constants from the reflection of light caused by the self-radiation of the surface processing sources will be described. During surface treatment, the spectra of the reflections are recorded from interferences that occur. Four different wavelengths are individually detected via a wavelength filter. From the measured signals of the normalized radiation, the absolute (real) thickness of the layer is determined. However, this method can only be used in systems with a plasma or an ion beam for surface treatment, since the intensity of the plasma or the ion beam must be additionally recorded. The substrate itself is not measured. It is a normalized refractive index is measured, the real value is determined in retrospect. The formulas used herein are consistent with and are limited to Fresnel's laws for a three-layer system (air-nonabsorbent medium-substrate).

EP 1 435517 A1 describes the use of spectroscopic ellipsometers for the determination of thin multilayer systems. However, ellipsometry only works reliably with very smooth layers. In addition, an ellipsometric measurement is possible only ex-situ, since the reflection must be evaluated by a polarization measurement method. The evaluation is carried out by changing the polarization of the incident light and the resulting influence on the reflected radiation. The determination of the complex refractive index is effected by a defined wavelength change. The method works for multilayer systems with an optical model, but which is based on the numerical adaptation (fitting) of a frequency-dependent function of thickness and refractive index (similar to FFT, Fast Fourier Transformation). EP 1 467 177 A1 likewise discloses a thickness measuring method for multilayer systems. This method is based on an ex-situ measurement, which is followed by an evaluation due to a Fourier transformation. Light is radiated onto the sample and a frequency spectrum is created by the FFT. By evaluating the peaks of the FFT, the individual layers are determined in their thickness. A CCD (Charged Coupled Device) is used as a spatially resolving optical detector for different wavelengths.

From DE 10 2005 023 735 A1 an ex-situ method for the automatic performance of a surface investigation is known in which a theoretical curve is adapted to a measurement curve with an FFT method or gradient method. In this case, however, layer thicknesses above 10 microns are observed, which are no longer classified as optically thin layers. The reflection spectrum is compared with a calculated spectrum. In addition, only one FFT spectrum is evaluated and the occurring peaks of this spectrum are considered. The number of layers is determined by the FFT spectrum. Thus, the layer must already be finished, otherwise no spectrum of this kind can be recorded. Different approximation methods are derived from this spectrum.

DE10 2005 023 737 A1 shows how the total thickness or the refractive index of a thin layer can be determined from the total reflection. The determination of a layer thickness or a dispersion parameter from a reflection spectrum is described. This is the

Measurement compared with a model spectrum. However, only the currently changed layer is considered, whereby the model used is not explained further.

The aforementioned publications are in the field of analysis of

Reflections of optical radiation on smooth surfaces. All methods are based on the measurement of total reflection and thus require one Normalization either by measuring the reference light or by subsequent measurement of refractive indices by other methods. Used optical models are based on the use of Fresnel equations. No method is used to control an evaporation process of materials to obtain an optically thin layer.

A method for deposition processes of chalcopyrite thin films on moving substrates is known, for example, from DE 102 56 909 B3. Here, as a light source, a laser of coherent light of a wavelength is directed to a moving substrate to control the process of deposition and formation of a chalcopyrite thin film. The control in this method is based on the scattering of laser light on rough surfaces. The process of vapor deposition of Cu (In, Ga) Se ₂ layers is divided into three stages. Stage I involves the evaporation of indium and gallium (co-evaporation or sequential evaporation), stage II the co-evaporation of copper and stage III the co-evaporation of indium and gallium. During the entire process selenium is evaporated. The substrate is described comprehensively in material (glass, titanium or plastics) and properties. Different concepts of substrate movement are shown (run, rotation, roll-to-roll). Concepts for moving the laser used are given. With the known method, it is possible to carry out the process in a controlled manner. The process control in the known method is based on laser light scattering (LLS) and uses individual characteristic points. The scattered signals of the laser light are logged during the individual stages. In particular, the second stage leakage signals are used to estimate stoichiometric ratios in the deposited layer. However, this control does not offer the possibility of carrying out a process in a reproducible manner since this requires a control with a numerical value assignment of the optical layer parameters. There is only qualitative monitoring of the process. In the case of qualitative The production parameters "temperature of the evaporation sources" and "speed of the substrate" are changed accordingly.

In the abstract I of R. Scheer et al., "Cu (lni _-x Gax) Se ₂ growth studies by in situ spectroscopic light scattering" (Applied Physics Letters 82 (2003), p.2091-2093) the method of LLS extended to a spectral light scattering (SLS) to detect the dependence between the roughness of the deposited layer and the scatter signal. Coherent laser light works only on rough surfaces. With smooth surfaces, as they are usually present through the substrate at the beginning of the process, there are no evaluable measuring signals. For the SLS, instead of a laser, a white light source and an SSD spectrometer are used as detectors. Also a process control based on the SLS is from the publication Il of K. Sakurai et al. known in-situ diagnostic methods for thin-film fabrication: utilization of heat radiation and light scattering (Progress in Photovoltaics: Research and Applications 12 (2004), pp. 219-234), of which the present invention as closest state of the However, process control is not discussed, and the numerical values of the optical layer parameters are unknown, so that no quantitative control of the growth of thin layers and thus control of the actual values of the controlled variables (optical layer parameters) to the setpoints by adjusting the manipulated variables (process parameters ) can be made.

One way of quantitatively controlling the growth of thin layers is exemplified in US 5,450,250. In this known method, interferometric measurements for determining the thickness of an optically thin layer are carried out with the aid of a laser and a CCD camera. A laser illuminates the surface of a silicon substrate. By means of a CCD camera w.erden the reflections of the incident laser beam recorded and evaluated. The method is based on interferometry (cf. DE 10 2005 050 795 A1 and DE 10 2006 016 132 A1). In this case, incident and reflected beams are superimposed in such a way that, in extreme cases, they can overlap destructively or constructively. As a result, absolute thicknesses of thin layers can be determined. A determination is only possible if the optical properties of the measured material are sufficiently known. Above all, the size of the complex refractive index must be known. Furthermore, it is not considered that the material properties mentioned often undergo changes during the deposition process. Thus, in the said method, only a fundamental possibility for quantitative process control and a possible regulation is given.

For a process control in the sense of a real control loop, it is essential to be able to measure also the components to be controlled, as described, for example, in US Pat. No. 7,033,070 B2. To monitor a crystal pulling process (float zone process) for the recovery of silicon single crystals, the temperature of the silicon melt is monitored by means of a CCD camera. In this process, the grown crystal is locally heated by a halogen lamp. The camera is aimed at the area of the crystal which is heated by the halogen lamp. The CCD camera measures the luminosity of the melt by filtering the reflected and scattered light of the light source and thereby estimates the temperature. The CCD camera is equipped with filters that make infrared light visible. Thus, the known method numerically monitors melting of silicon by the intrinsic luminosity of the melt. However, this process does not provide information about the composition or thickness of the material. Rather, the camera monitors to what extent the crystal to be drawn has melted. TASK

The TASK for the present invention is to be developed, the above-mentioned, generic method for in situ determination of the composition of material deposited from the vapor phase on a substrate optically thin layers with an interferometrically evaluable correlation between the optical layer parameters so that the material composition of the deposited layer can be determined with knowledge of the optical layer parameters. However, the process should be cost-effective and immune to interference in real time feasible. A preferred arrangement for carrying out the method should have corresponding components.

Furthermore, the reproducible production of highly efficient semiconductor layers for photovoltaics must be ensured within a continuous co-evaporation process. The continuity of the evaporation process leads to the need for an in-situ process, which allows a regulation at any time of the evaporation process. However, in order to be able to control a process, it is necessary for the control mechanism to be free from errors or disturbances and to be able to measure the controlled variables. Likewise, it is necessary to keep the influence of unavoidable sources of interference negligible. In applications of the method according to the invention, therefore, a process control in the sense of a control loop that quantitatively accesses the specific optical layer parameters should also be feasible. For this, it is essential to be able to measure the components to be controlled (layer parameters) as well.

The SOLUTION for these tasks can be found in the ancillary procedural, product and application claims. advantageous

Further developments are shown in the respective subclaims, which are explained in more detail below in connection with the invention. The method according to the invention is suitable for the in-situ determination of the material composition of optically thin layers which are deposited from the vapor phase on a substrate. This is in optically thin layers are those layers in which the complex refractive index nk _o mpi _e x (Real part: refractive index n _SC hicht; imaginary: absorption coefficient κ _Sc hi _ch t) and the real coating thickness dsohicw) in an interferometrically evaluated correlation are dependent on each other. The limit of interferometry is reached when the layer is too thick or too high absorbing. In general, optical layers exhibit a thickness up to a maximum of the ^"l Ofachen the irradiated wavelength or a thickness of less than 1000 nm.

For the in-situ determination of the material composition of such optically thin layers during their production, the method according to the invention comprises at least the following method steps:

During the deposition of the optical thin film, the substrate on the deposition side is irradiated with incoherent light in the visible optical range. In this case, light is directed at least three different wavelengths λi, X ₂ , λ ₃ onto the deposited layer. Three different wavelengths are required because three unknown optical layer parameters (refractive index nsc _Wc ht, absorption coefficient κs _Ch i _cht and layer thickness d) are also to be determined. Incoherent light can be used on both rough and smooth surfaces. The use of coherent light is limited to rough surfaces.

The diffuse or direct light scattering emanating from the deposited layer is detected optically spatially resolved in the three different wavelengths λ-i, λ _2, λ ₃ . The time-dependent course of the reflection intensity R is measured in each case. The light is radiated in such a way that no total reflection results. Total absorption also does not occur with optically thin layers.

The numerical values of the detected reflectance intensity R for the different wavelengths λ-i, X _2, λ ₃ are time-fed parallel to the measurement in an optical layer model, which thus takes place synchronously to the deposition process. This optical layer model is based on the general theory of conduct, known from electrical communications engineering. In the method according to the invention, each deposited layer is interpreted as an electromagnetic conductor with a variable field wave resistance Z _G , a propagation constant γ and a real layer thickness equivalent to the conductor length dschicw.

The optical layer model is numerically evaluated for the different wavelengths λ-i, λ ₂ , λ ₃ . It will become apparent by numerical adaptation of the layer model or this function describing the time course of the detected reflection intensities, the real values for the optical layer parameters (refractive index n _SCh i _CHT, absorption coefficient Kschicht, real layer thickness d _Sc hich _t) of the deposited layer. In this case, the wavelength-independent real layer thickness dschic _h t is determined to the reference value, so that the inventive method is self-referencing. A reference formation from the outside with additional quantities to be detected, as known from the prior art, is eliminated.

The actual material composition of the detected deposited layer is then determined from the numerically determined values for the optical layer parameters. This is done by comparison with standard values of values for optical layer parameters of known material compositions.

In an advantageous embodiment of the method according to the invention, it is provided that when changing the material composition between different layers, the determined values of the optical layer parameters of the previous layer are stored and used as reference values for a next layer. Therefore, the method according to the invention can also be used for any stack of optically thin layers, because it always analyzes the uppermost layer with respect to the underlying layers as a reference. Furthermore, it may be advantageous to additionally measure the reflection intensity which is caused by incident coherent light of a single wavelength (LLS). In this way, the influence of the surface roughness of the deposited layer can be taken into account. This requires, however, that a rough surface is present. Smooth surfaces can not be analyzed with coherent light.

Further details of the individual investigations and embodiments of the method according to the invention and also the preferred arrangement for carrying out the method, in particular an optical imaging device, such as a CCD or CMOS camera for the colors red, blue and green, as an optical detector and a White light source, can be found in the special execution part.

A preferred application of the method according to the invention consists in an embodiment as an in-situ process control in the production of vapor phase deposited on a substrate optically thin layers. In this case, the method is incorporated into a control loop, with which the calculated actual values of the optical layer parameters of the optically thin

Layers are controlled as controlled variables to predetermined setpoint values by adaptation of the production parameters as manipulated variables in accordance with the determined actual values for the optical layer parameters.

In addition, the process according to the invention can still be used as an in-situ process control even if no optically thin layers, but absorbent layers are grown. In this case can the method according to the invention is used at least for the estimation of the stoichiometry. In this case, the reflection intensity R of the diffuse or direct light scattering emanating from the layer deposited on the substrate outside of the total reflection in at least two wavelengths λ-i, λ ₂ is interpreted accordingly. A significant change in the course indicates the point of stoichiometry achieved. Furthermore, in the method according to the invention, a simultaneous temperature monitoring of the evaporation sources via the spatially resolved optical detection by interpretation of the occurring color values of the temperature radiation of the evaporation sources done. Finally, an application can also be made to rough substrates or rough deposited layers, as there is no dependence here due to the use of incoherent light. If, however, the influence of roughness is to be taken into account, appropriately coherent (laser) light must be used. Further details of the preferred applications can also be found in the specific description part.

EMBODIMENTS

Embodiments of the invention in all three claimed categories are explained in more detail with reference to the schematic figures. Showing:

FIG. 1 A diagram for the transfer of conduction theory into optically thin layers, FIG.

FIG. 1 B schematic diagram for interferometry on optically thin layers, FIG.

FIGURE 2 Representation of the conversion of the time dependence of

Reflection Intensity in Thickness Dependence, FIGURES 3A-D Diagrams of fitted reflection in the deposition process, FIGURE 4 Diagram showing the measurement signals of all three

Color components of an image signal,

FIG. 5 process sketch for an entire PVD process,

FIGURE 6 stacks of deposited optically thin layers for the entire PVD process,

FIG. 7 SEM image of a stack of several layers, FIG.

FIG. 8 sketch for an arrangement for a complete PVD process,

FIG. 9 control loop for process control,

FIG. 10 Measurement signals of a CCD detector (stage I), FIG. 11 Diagram of the temporal intensity profile of the reflections

(Level I),

FIG. 12 Measurement signals of a CCD detector (stage II), FIG.

FIG. 13 Diagram for the temporal intensity profile of the reflections

(Stage II), FIGURE 14 Comparison of measurement signals invention / prior art and

FIG. 15 Comparison of measurement signals on rough layers between coherent and incoherent light sources.

I) THEORETICAL PRINCIPLES OF THE PROCESS OF THE INVENTION

The approach to the development of the method according to the invention is in the wave theory. This theory describes the propagation of waves in dielectric media and is known, for example, in the field of high-frequency technology. In the invention, the wave theory is transferred to optically thin layers. Each layer is interpreted as a single line with a parameter set of propagation constant, field wave resistance and thickness (corresponds to the conductor length) and clearly described (see FIG. 1A). In the left part of FIG. 1A, two optically thin layers are represented by two characteristic impedances ZF1, ZF2 and the thickness d of the just growing, upper layer. The field wave resistance Z _F is defined by the dielectric in which the shaft is located. A material can be clearly described by the permeability μ and the permittivity ε.

^Z> <(1)

With the aid of the two variables mentioned, the field wave resistance ZF can be defined according to equation (1). In this case, according to another notation for the field resistance ZF also applies:

Z _F - Z _F0 ^■ "- ■ / * ^" (1 a)

The value n is the (real) refractive index and K is the extinction coefficient (imaginary refractive index). ZF ₀ here is the characteristic impedance of the vacuum and a natural constant. Thus, with a single layer, two different wavelengths are required for the system solution, since two unknowns (n and K) occur.

If another unknown layer grows on the now known layer, the following formula applies, which covers the thickness of the layer. The origin of the equation is the telegraph equation (general form of the wave equation). The known layer (Z _F _i; at a first layer to be deposited, it is then to the substrate, wherein each subsequent layer to the underlying layers) and the growing (unknown) layer (ZF _2, "always the uppermost layer) are added to a common layer (Z _G ), which depends on the thickness, and the complex refractive index.Thus, three different wavelengths are needed in the detection to determine the three unknown quantities. FIG. 1B shows the interferometry on optical thin layers with the sheet resistances ZF-i, ZF ₂ (optically thin layers) and ZF ₃ (substrate, in the selected exemplary embodiment made of molybdenum) compared with air ("air") (Layer thickness 20 - 30 nm) consists of a compound of the evaporated selenium with the molybdenum of the substrate, the next layer consists of a compound of gallium and selenium.The layer thickness Δd (at known temperature of the evaporation sources), the refractive index n (real part ) of gallium from the third main group (Ga / III) and the absorption coefficient K (imaginary part) of Ga / III.

In field wave resistance theory, both layers are combined into a total resistance ZQ. The following applies:

Z _F] + Z _F? ■ tanhfr • d, _m ,) Z _G = Z _F0 - - ^Fl - ^F2 ) ^L " ^ι " [(2)

Z _F2 + Z _Fλ - tanh [γ- d _0l J

With the gain of the total resistance Z _{G of} the two layers, the reflection intensity R of the incident light radiation can be calculated by the layer system. It applies with Z _F o as the field wave resistance of the vacuum:

The value Y (propagation constant) describes the propagation of a wave and is described only by the imaginary refractive index and the wavelength used:

2 iτπz Z / _{F mn} 2 IτTcC i / .. _\ \ 2 u7tz. _x

7 = - - ~ 1 ± - ⁼ - ^" [\ κ ^K + ⁺ _j Jn ^} V) == - - ^} ho, n _P le _X (4)

^J Z ^J F _E - _T 1 A A In an evaporation process, the calculation unit performs the calculations of equations (2) and (3) in-situ to determine the thickness and the complex refractive index. For this purpose, the measurement signals of the optical detector in the three different wavelengths are numerically adapted ("fitted" by using varied values for complex refractive index and layer thickness) with the optical layer model and compared with results of formulas 2 and 3.

II) EXPIRATION OF THE CALCULATIONS OF THE COMPLEX BREAKING INDEX AND THE THICKNESS

A LUBRICATED LAYER ON A KNOWN SUBSTRATE.

Preliminary remark: The use of three wavelengths λ-i, X ₂ , λ ₃ is necessary for the method according to the invention. The reason for this lies in the missing reference size of the incident light. The light usually enters the evaporation chamber via a disk, which is regularly involuntarily co-evaporated. Thus, the knowledge of the intensity of the radiation source is of no use, since a vaporized disc through which the light passes through greatly influences the characteristic of the light. In the measurement of the reflection, therefore, a different reference had to be found in the invention. One of the three unknown quantities (real and imaginary refractive index and the thickness) should serve as a reference. However, the complex refractive index is wavelength-dependent and therefore can not be used. However, the thickness is not wavelength dependent and can therefore be used as a reference since it must be identical for all measured wavelengths. The method according to the invention can therefore be termed "self-referencing".

The starting point of the calculation for the method according to the invention is the measurement of the reflection intensity in three different wavelengths Xu X ₂ , λ ₃ . For this purpose, it is particularly advantageous to use a CCD ("Charged Coupled Device") as an optical detector. <br/><br/> The reflection intensity R for three wavelengths λ-i, X _2, λ ₃ (red, green and blue) is present through the CCD and evaluated individually The conversion from time to thickness is done at the beginning of the deposition process on the basis of the occurring interference maxima.The calculated thickness is the optical thickness d _opt , ie it remains the previously unknown Refractive index in thickness initially included.

Here m corresponds to the interference order, φ is the angle of incidence, λ the wavelength of the incident light. With the aid of the equation, the conversion from the time-dependent representation into a thickness-dependent one can be traced.

a ^a opt - ^a opt2 (6)

Given the ratios, the optical layer thickness d _{opt is} calculated as half the wavelength XR of the time course of the reflection intensity

d "", = λ "opt (6a)

FIG. 2 shows a diagram for converting the time dependence of the reflection intensity R (ordinate: RGB signal, abscissa: time / time in any unit) into the thickness dependence. The parameter d _{opt is} calculated from the quotient of the real layer thickness d _SCh ic _ht and the complex refractive index n _k0 mpleχ: d opt = ^ ^uc hL (7) n complex

The reflection intensity can be defined by means of the characteristic impedance according to equation (3). The characteristic impedance of the vacuum Z _F0 is known and thus can be taken into account by changing the equation 3, the total resistance Z _{G of} the vapor-deposited layer. The resistor Z _G describes the vapor-deposited layer and the underlying substrate. In order to separate the two layers from each other, equation (2) is still applied considering equation (4).

A fit compares the equation (2) with the measured reflection intensities. In this case, a determination of the measured values is carried out individually for all three measured wavelengths by the fit of equation (2).

THE CALCULATION STEPS AT A GLANCE:

I) For all three wavelengths, the real thickness d is calculated numerically, with the thickness of the deposited layer for all three

Wavelengths must have the same value.

II) For each of the wavelengths, using the calculated thickness of step 1, equation (2) is solved. Here, the reflection measurement of each wavelength is fitted with Equation (5) and Equation (2).

The fit is designed such that the known substrate and the known thickness are used in equation (2). The value Y can be described by equation (4). Thus, there remains a numerical equation for the characteristic impedance of the deposited layer. This equation is fitted and the smallest

Errors are determined by the real thickness and refractive index. III) For the entire process, repeat steps 1 and 2 until the limit of interferometry is reached (layer too thick or too high absorbing).

FIG. 3A shows the applied reflection (solid line) to the measured reflection intensity values (+) of Ga ₂ Se ₃ for a deposition process (ordinate: RGB signal, abscissa: time / time in arbitrary unit). FIG. 3B shows by way of example how the method according to the invention operates. During the measurement the Fit (the optical layer model) runs along and calculates the thickness and the complex refractive index during the process. While the thickness in Figure 3B is visibly juxtaposed as the axis of time, the measurement of the complex refractive index is not visibly included. In optically thin systems, it is only possible to determine the thickness and the complex refractive index together since both are correlated with each other. Thus, more information is needed to determine the refractive index. It is known from interferometry that three wavelengths are sufficient (thickness and complex refractive index). With white light, there are infinitely many wavelengths available for the measurement. FIG. 3C shows the temporal

Intensity curve of the red (circles) and the blue (squares) reflections with a simultaneous evaporation of Ga and Se on a molybdenum foil as a substrate, while Cu and In are closed. The influence of the initially forming MoSe is indicated by an arrow. As in FIG. 3B, it can be well deduced that the optical layer thickness corresponds to half the distance between the two minima. At the end of this first evaporation phase in step I, the composition of the deposited layer is well known by knowledge of complex refractive index and real layer thickness and is used as reference for the subsequent deposition of In and Se (with closed Ga and Cu) deposited layer. The FIGURE 3D shows the corresponding temporal intensity course of the red (circles) and the blue (squares) reflections in stage I. Shown is also the occurring absorption by the deposited layer by a corresponding simulation of the curve envelope. At the end of the second evaporation phase in stage I, the calculated optical layer parameters result in a proportion Ga / III of 0.5028 as a proportion Ga to the proportion In + Ga (Ga / Ga + In). For the subsequent processes, therefore, the composition of this layer is known and can be taken into account accordingly.

The routine shown is performed for all three wavelengths. For the phase of vaporization of gallium and selenium (stage I, sequential vaporization of gallium (in gallium zero, indium is evaporated, permanent evaporation of selenium) shows FIGURE 4 (ordinate: RGB signal, abscissa: time / time in any unit) simultaneously all three evaluated wavelengths.The triangles show the evaluation of the blue wavelength, the squares of the green wavelengths and the circles of the red wavelength.

III) APPLICATION OF THE CALCULATIONS TO A CONCRETE EXEMPLARY EMBODIMENT.

The embodiment of the method according to the invention relates to the sequential co-vapor deposition of metals and non-metallic elements of the fifth to seventh main group, wherein there is at least one sequence of evaporation. A possible timing of the evaporation for the elements A (eg In), B (eg Ga), C (eg Cu) and D (eg Se) can be seen in FIG. The process described here is a sequential co-evaporation of the type Ga + Se, followed by In + Se, to obtain a layer stack of Ga ₂ Se ₃ and ln ₂ Sβ ₃ . The stack of optically thin layers that can be generated by this evaporation process in stage I is shown schematically in FIG. A high-resolution SEM image of a complex stack made up of a plurality of superimposed, optically thin layers produced by the process control according to the invention is shown in FIG. 7. The method according to the invention can always use the respectively uppermost deposited layer with respect to the underlying layers be analyzed as a reference layer in terms of their material composition. Therefore, the method is also applicable to stacks with any number of individual layers. In step II of the above process, an absorbing Cu-containing layer is produced, to which interferometry is not applicable. Here, however, the method according to the invention can be used to estimate the stoichiometry. In stage III, then again analyzable optically thin layers are deposited.

FIG. 8 shows the basic structure of an arrangement for carrying out the method according to the invention. In a PVD chamber PVD, the elements A, B, C, D are evaporated sequentially and deposit on the deposition side of the substrate S rotating in the embodiment as composite layers. Depending on their composition, the depositing layers reflect incident light as diffuse or direct reflections of incoherent radiation (no total reflection). During the entire deposition process, the diffuse or direct reflections with at least one spatially resolving optical detector D! added. Another detector D2 may be provided, for example, orthogonal to the first detector D1 laterally from the substrate S. The optical radiation interacts with the vapor-deposited optically thin layer in a characteristic manner, so that from the reflection with the detectors D1, D2, the quality of the vapor-deposited layer can be assessed directly with respect to its composition and any deficiencies in the control loop can be used to correct the process. In the selected exemplary embodiment, the evaporation sources A, B, C, D, X, Y are simultaneously used as incoherent light sources ILQ in order to irradiate light of different wavelengths onto the deposition side of the substrate S. Alternatively or additionally, white light sources ILQ (without evaporator function) with an infinite wavelength spectrum can also be used. In addition, a coherent light source KLQ (here a laser) may be provided in order to be able to take into account the influence of the surface roughness of the depositing layers. Likewise, further evaporation sources X, Y, for example, for the evaporation of non-metallic elements, be provided.

The control of Koverdampfens requires no spatial separation of the individual evaporation sequences, but it does not exclude them. Essential for the regulation of the vapor deposition process is the reflection measurement and evaluation of the incident on the moving substrate optical radiation. For this purpose, the radiation must be spatially separated from each other in order to evaluate the individual radiations in the system can. For this purpose, the spatial resolution optical detector D, for example a CCD camera equipped with optical instruments, is mounted inside or outside the PVD chamber PVD. When mounted outside the PVD chamber PVD a corresponding window F is provided in the chamber wall, which is also occupied during the evaporation process with a Abscheideschicht so that an external referencing of the control system is not possible. This is also not required in the method according to the invention by its self-referencing capability. The CCD cameras convert optical radiation received by means of an array of optical sensors into color values in the wavelength range from 400 nm to 700 nm. These then correspond to the incident wavelength and have been calculated with the spectral sensitivity of the individual sensors of the CCD chip. A CCD camera consists of an array of optical sensors and can thereby locally separate the incident signals. The use of CCD cameras does not limit the type of optical radiation. Rather, it can be used for monitoring any optical radiation in the visible spectrum. Furthermore, the number and location of the CCD cameras are arbitrary and can be freely selected due to the quality of the cameras used. The general control of moving or unmoving substrates in connection with a spatially resolved optical detection is also solved particularly elegantly by the use of a CCD camera, which is aligned with its optics in the direction of the substrate. Preferably, the method according to the invention can be used as a process control for an evaporation process of the type described above. By the numerical determination of the optical layer parameters, a quantitative control of the process by influencing the process parameters is possible. Thus, this regulation differs significantly from the known regulations in this field, with which only a qualitative control (according to predetermined characteristic points) is possible without precise numerical knowledge of the optical layer parameters. The function of the associated control circuit is indicated schematically in FIG. 9 for the general case (cf. FIG. 8). The detectors D1 and D2 serve to measure the reflection intensities and transmit the measurement signal to the control unit CU. The control unit CU uses the measured signals using the type of light sources used for control (ILQ, KLQ). Here, the operation of the light sources is freely selectable. These may be chopped or choppy light signals. From the measured signals correlated to the light sources used, the control unit controls the movement of the substrate S, as well as the evaporation sources A ₁ B ₁ C ₁ D ₁ X ₁ Y (X, Y may be non-metallic evaporation sources, for example). After the eventually required change of the material sources (change of the evaporation rate) or of the substrate (change of the traversing speed), the control unit CU again uses the incoming measuring signals and thereby controls the evaporation process continuously and in-situ. FIG. 10 shows, by way of example, the measurement signal of an optical CCD detector with visible reflections of the light sources and the temperatures of the evaporation sources for stage I of the exemplarily selected 3-stage vapor deposition process of Cu (In, Ga) Se ₂ thin films. For this measurement signal recording, the CCD camera is located in the bottom of the PVD chamber and is oriented with its optics perpendicular in the direction of the substrate S (see FIG. 8).

Specifically, in FIG. 10,

In addition to the use of a CCD camera for spatially resolved measurement of different light sources, the possibility of monitoring the function of the evaporation sources by estimating their temperature via the control unit CU is simultaneously given by the use of a CCD camera. In the gray depiction in FIG. 10, the colors of the original measurement protocol are reproduced only by their gray value. The original measurement protocol shows a red reflection of the light sources Ac and B ₁ a green reflection of the light sources A _B and the evaporation sources B and X as well as external noise and a blue reflection at the evaporator sources A and C (with red islands). While the evaporation sources A and B have approximately the same temperature and thus have a similar temperature radiation, distribution and color, the evaporation source C is operated at a significantly higher temperature. The higher temperature is represented by the CCD camera as blue color. If used incorrectly, the source of evaporation might not turn blue, but would not be visible. The evaporation source C Due to the high temperature, it has intense dots measured by colors other than blue within the temperature distribution. The CCD camera is adapted to the human eye, displaying areas of highest intensity as green, as the human eye has the highest sensitivity to green.

FIG. 10 shows the reflections of the individual light sources (also evaporation sources). For use as a measurement signal for a control loop, the measurement signals must first be evaluated by the control unit CU for semitransparent, optically thin layers. The CCD camera separates the reflections locally from each other and thus enables a separate evaluation within the control unit. The control unit then converts the measured, separate reflection intensities of the individual light sources into numerical intensities and applies them in a time-dependent manner.

FIG. 11 shows the time profile of the locally separated reflection intensities processed by the control unit at the substrate surface (red-green-blue intensity "RGB intensity" in any unit (au) over the time "time"). The wavelengths of light sources A and B are indicated by their wavelength. In the exemplary evaporation system used, a laser module was used as light source B. The light source A is due to the temperature radiation of the source A (here: gallium melt). From the physical laws that are well known for the determination of an optically thin layer, it is possible with two different wavelengths to determine the thickness of the resulting layer as well as the complex refractive index (two components) of the deposited layer (see above). Furthermore, a difference between the behavior of the short-wave temperature radiation of the source A and the long-wave radiation of the light source B in FIG. 11 can be seen. The intensity of this radiation decreases visibly. As a result, it can be determined by the control unit that the deposited substance absorbs the blue radiation. biert. In contrast, the red radiation is not absorbed, made visible by the fact that the interference signal decreases only insignificantly. The control unit has data on the expected layer on the substrate and, with the aid of these data, which are in the form of step diagrams and values for the wavelength-dependent complex refractive index, determines the optical layer parameters and, depending thereon, the layer composition of the substance formed. In the example shown in FIG. 11, the control unit provides the information that the deposited layer is Ga ₂ Se ₃ . This substance has a refractive index of 2.4 and absorbs only at a wavelength below 580nm.

The method according to the invention is limited only by the thickness of the deposited layer, since from a thickness of about ten times the wavelength of the incident radiation the interferometry provides no longer valid results. However, this limitation is not achieved by restricting the invention to optically thin layers and is therefore negligible. For optically thin layers, the method according to the invention in an embodiment according to FIG. 8 provides for the interferometric calculations a sufficient control possibility and can thus be used for the purpose of a process control. Nevertheless, with the method according to the invention, at least one stoichiometric estimate can be carried out due to the special optical evaluation for absorbing layers. Such an estimate is also known for example from DE 102 56 909 B3.

Using the example of vapor deposition of a CulnGaSe ₂ layer, the procedure should be explained. In step I, a semitransparent layer of indium, gallium and selenium was deposited on the substrate. FIG. 12 shows the measurement signal of the CCD camera for the evaporation of copper in stage II on the semitransparent layer just described. It becomes an exemplary measurement signal of the detector with visible reflection of Light sources, and the temperature of the evaporation sources for stage Il shown.

Specifically, in FIG. 12,

FIGURE 12 shows the recorded area of the CCD camera used to control the process. In comparison to FIG. 10, the interference signal of an external radiation has been cut out here, which can be achieved by the spatial resolution of the measurement signal. During this exemplary process, the rotation causes a partial overlap of the two light sources A and B. However, due to the processing of continuous measurement signals, the control unit is able to separate both reflections and to evaluate them individually. The said reflection of the light source B corresponds to the already described exemplary light source for the temperature estimation of the evaporation sources for proof of function.

The aim of this exemplary process is the estimation of the stoichiometry in the molar amount of the metals indium and gallium and the metal copper. For a stoichiometric Cu (In, Ga) Se ₂ thin film, the amounts of indium and gallium should correspond to the amount of copper on the substrate. The control unit is due to the target composition for this Substance calibrated and can thus carry out the estimation of the stoichiometry. In this particular example, the CCD camera measures the process until the processed measurement signals in the control unit detect the occurrence of a copper selenide (Cu _2-x Se) coating on the surface of the deposited layer. For a CCD camera, this effect is evidenced by the continuous increase in the intensity of the reflection of the light source Ac with the simultaneous occurrence of a decrease in the reflected light source B in the exemplary system.

FIG. 13 shows the time profile of the CCD camera measurement (red-green-blue intensity "RGB intensity" in arbitrary units (a.u.) over time "time") evaluated by the control unit. The course of the temperature reflection of the copper source (upper curve, color green, "Intensity from Lightsource A") is shown together with the reflection of the light source B (here: laser 635 nm wavelength, lower curve, color red, Intensity from Lightsource B ") , The behavior described is indicated by the oval ("approximation of stoichiometry") in FIGURE 13. The occurrence of this behavior satisfies the requirement for an estimation of the stoichiometry within the deposited layer.

To control the method according to the invention for measuring optical reflections, the evaluations of the control unit are compared with those of the control system according to DE 102 56 909 B3. In this known control system, a photodiode followed by a lock-in amplifier is used. The photodiode receives a modulated reflected signal from the substrate surface, integrates the entire signal, and extracts the modulated signal for analysis. Thus, there is no local resolution of the various signals. Nevertheless, as already mentioned, a qualitative control with the known method according to DE 102 56 909 B3 is possible on the basis of characteristic individual points. FIG. 14 shows a comparison between the two methods in the time course of the entire control process according to the invention (red-green-blue intensity "RGB intensity". in any unit (au) over the time "time") with additional indication of the course according to the method according to DE 102 56 909 B3. The lower, light gray curve in step 1 shows the time course of the reflection intensity of the color green (from the RGB According to the invention, the average, thin black curve shows in comparison the signal profile from the method according to DE 102 56 909 B3 a very similar result can be seen in the measurements (compare upper two curves), the lower curve for the method according to the invention shows the general insensitivity of the laser signal to a relatively smooth substrate / layer during stage I. However, the measurement according to DE 102 56 909 B3 is highly prone to error as the process has to be rough-layered One of the problems is that laser signals, ie coherent beams Due to the roughness of the surface is influenced more than incoherent radiation.

A quantitative control of the layer deposition is not possible with the known method according to DE 102 56 909 B3, since the measured interferences are superimposed on the roughness of the deposited layer. Since the availability of incoherent light means that this dependence does not occur in the method according to the invention, the corresponding control system with the CCD camera offers the possibility of characterizing the deposited layer by a numerical determination of the thickness and determination of the refractive index. For stage II, a clear occurrence of the stoichiometry can be observed with both systems, which, however, is represented differently. For the stage III, not further shown in FIG. 14, a quantitative analysis of the layer composition could again be produced with the method according to the invention. However, this is usually not necessary, since in Stage III it is only a matter of leaving Stage II in any case in the area of a copper-rich deposit. FIGURE 15 shows a comparison of coherent and incoherent light sources on rough substrates. A simultaneous measurement is shown between the method according to the invention and the method according to DE 102 56 909 B3 (LLS). In the diagram, the process has begun and the first gallium and the first subsequent indium layer are deposited. The measurement signal of the LLS (curve c) shows a very weak signal, while the signals of the method according to the invention are clearly visible. In the first indium phase, the influence of the roughness on the laser signal is again indicated by arrows. An apparent phase shift of the reflection signals of the LLS does not allow an accurate measurement, whereas the method according to the invention with white light sources shows no phase shift.

IV) OVERVIEW OF THE POSSIBLE VARIATIONS OF THE PROCESS AND THE PROCESSING DEVICE AND THE APPLICATIONS OF THE INVENTION

I. The named sources of evaporation A ₁ B ₁ C ₁ D ₁ X ₁ Y are sources of evaporation for any elements - A ₁ B ₁ C ₁ D: any metals, X ₁ Y: any non-metallic elements of the V. Vll, -Hauptgruppe.

II. The named light sources A, B are light sources of any nature a.) Coherent sources: lasers of any nature (e.g., gas discharge lasers,

Laser diodes and other sources of coherent, optical radiation) b.) Monochromatic, non-coherent light (eg light emitting diodes of any optical wavelength, division of continuous optical spectra by optical methods, eg monochromators, filters) c.) Non-monochromatic, non-coherent light (eg: white light sources, such as halogen lamps D.) Other light sources (eg: optical radiation of metal melts, the melts produced in the sources A, B, C, D and optical radiation of melts of the described sources X ₁ Y). III. The positioning of the designated light sources A, B can take place a.) Within the process device at any angle α b.) Outside the process device at any angle α using heated or unheated optically transparent devices (eg windows)

IV. The coupling of the radiation of the designated light sources A ₁ B can take place by a.) Direct illumination of the substrate during the deposition process to be controlled with II. Named light sources without optical instruments b.) Direct illumination of the substrate during the controlled vapor deposition with after II. Named light sources with any optical instruments connected upstream (eg linear and circular polarizers, convex and concave lenses for imaging, interference filters or other filters for wavelength adjustment of the light sources) c.) Direct illumination of the substrate during the vapor deposition process to be controlled with light sources named after II by guiding the optical radiation by means of light guides (eg: optical mono- and multimode fibers, mirror systems for deflecting the optical radiation and adjusting the position and the angle of incidence on the substrate)

V. The named detectors are spatially resolving, optical sensors of any nature in the formation of a.) Integrating, spatially resolving detectors without optical instruments (eg: photodiode arrays or arrays, line and line CCDs, which are suitable for the measurement of optical radiation, arrangements of several locally separated photodiodes) b.) integrating, spatially resolving detectors with optical instruments (eg named elements according to Va) with imaging convex or concave lenses, interference filters or other filters for adjusting the sensitivity to specific wavelengths, decomposition of optical signals with monochromatic prisms or C) non-integrating, spatially resolving detectors without optical Instruments (eg CCD cameras of various resolution and design, photographic plates, photosensitive papers (films), camera systems for imaging) d.) Non-integrating, spatially resolving detectors with optical instruments according to Vc) (eg cameras with optical imaging)

Facilities - concave and convex lens systems, lenses for digital and analog cameras for image acquisition)

VI. The named substrate is of any nature (e.g., metal foils, sheets of glass, plastic sheeting; the substrates may be electroconductive materials such as molybdenum, and suitable materials to aid in the development of the substrate

Layer formation, such as sodium fluoride, evidenced)

VII. The movement of said substrate is of any nature (e.g., rotation, linear motion in the vertical and horizontal directions and any combination of the two aforementioned movements, substrate standstill is included).

VIII. The layer to be deposited during the evaporation process is of any nature and can be applied to a.) Obtaining an absorber layer for use in photovoltaic applications (eg: in thin film solar cells of the structure ABCXY2, such as and the structure A ₂ (BC) _x Di _-x (XyY-yy) ₄ , such as

Cu ₂ (ZnSn) χGa _1-x (SySe _1-y ) 4 with arbitrary values for x, y between 0 and 1) b.) Obtaining precursors of binary and ternary compounds to obtain those in ViI. a.) described absorber layers of the type ABXY (eg: InS, GaS, InSe, GaSe as an example of different compositional possibilities) c.) Extraction of any binary, ternary and quaternary layers whose purpose is also usable outside of photovoltaic applications. d.) Obtaining absorbent semiconductor layers from the precursor layers described in VIII. b.) by reactive processes during the

Evaporation of X or Y (eg, selenization and sulfurization of precursor layers) e.) production of other layer systems not mentioned in VIII. a) - d) which can be controlled by the method according to the invention,

IX. The system for the use of the control system is of any nature and can serve for a.) Control of a sequential one-stage or multi-stage system which evaporates per sequence at least one metal and one non-metal according to I. to obtain optically thin layers according to VIII. Vapor deposition (physical vapor deposition - PVD), b.) Control of a thermal process, which performs the method of Vlll.d.),

X. The control unit of the control system according to the invention makes use of the digital and analog read-in of imaging and integrating methods with instruments to V. Processing of the read measuring signals by means of digital technologies (eg: computer) and regulating signals from the control unit are using digital technologies the systems to be controlled according to I, VII transferred. A.) Evaluation of the measuring signals with detectors according to V., b.) Control of the temperature of the sources according to I., of the substrate according to VI. and the movement of the substrate according to VII. (eg: temperature change or coverage of the sources according to I) and c.) Correlation of external measurements of all relevant systems with the control system for checking and calibration (eg: power and temperature measurements of the sources according to I., Velocity measurement of the substrate according to VI and VII.) XI. The use of the control unit is of any nature and suitable for a.) In-situ control of continuous processes according to IX for obtaining optically thin layers according to VIII and b.) In-situ control of sequential processes according to IX for obtaining optically thin layers VIII. FORMULA AND REFERENCE LIST

A ₁ B ₁ C ₁ D ₁ X ₁ Y Elements / Evaporation Sources / Light Sources

A _Al AB, A ₀ Evaporation sources as light sources CCD Charged Coupled Device CU control unit D spatial resolution optical detector dopt optical thickness d layer real layer thickness F window

Y propagation constant

ILQ incoherent light source

KLQ coherent light source

K layer extinction coefficient (imaginary part complete refractive index) λ wavelength

LLS Laser Light Scattering m Interference order

^ complex complex refractive index) ^ layer refractive index (real part complex refractive index) PVD PVD chamber φ angle of incidence

R reflection intensity

RGB Red-Green-Blue Signal S Substrate SLS Spectroscopic Light Scattering ZF characteristic impedance of an optically thin layer ZFO characteristic impedance of the vacuum ZG total characteristic impedance

Claims

1. A method for the in-situ determination of the material composition of deposited from the vapor phase on a substrate optically thin layers with an interferometrically evaluable correlation between the optical layer parameters (complex refractive index nkompiex with real part: refractive index ns _ch i _cht and imaginary part: absorption coefficient Kschicht and real layer thickness dsc _h ic _ht ), at least with the following method steps: irradiation of the substrate on the deposition side with incoherent light of at least three different wavelengths λi, λ ₂ , λ ₃ in the visible optical range during the deposition process,

^Spatially resolved optical detection of the reflection intensity R of the diffuse or direct light scattering emanating from the deposited layer outside the total reflection in various three

Wavelengths λi, λ≥, λ3 _>

• interpreting time-parallel layer for detecting numerical values feeding the detektie- rten reflectance intensity R in an optical layer model based on the general line theory, in which each layer deposited γ as an electromagnetic head with a variable box impedance Z _F of a propagation constants and the conductor length equivalent real layer thickness d _Sch becomes,

• determining the values of the optical layer parameters of the deposited layer of the optical layer model for different wavelengths .lambda..sub.i, λ _2, .lambda..sub.s by numerical adjustment to the timing of the detected reflection intensities, the wavelength-independent real layer thickness d _Soh i _CHT serves as a reference value, and

• quantitative determination of the material composition of the deposited layer from the determined values of the optical Layer parameters by comparison with standard values for optical layer parameters of known material compositions.

2. Method according to claim 1, characterized by a determination of an optical layer thickness d _{opt of} a deposited layer as half the distance of the interference maxima of the reflection intensity R of the wavelengths λ-i, λ ₂ and K ¹ ₃ , wherein the product of the optical layer thickness d _op t with the complex refractive index n _k ompiex the real layer thickness d _sc hicht results.

3. The method according to claim 1 or 2,

CHARACTERIZED BY a relationship between the detected reflection intensity R and the total field wave resistance Z ₀ according to:

with Z _F0 field wave resistance of the vacuum.

4. The method according to any one of claims 1 to 3,

CHARACTERIZED BY a total field resistance Z _G of two superposed layers according to:

d _opt )

with Z _Fι field wave resistance of the first layer Z ^J F "1 field wave resistance of the second layer γ Propagation constant = - • - ^ - = - ^• (/ r + y «) = - - n _kωlψkx

A Δ _FΪ AA

5. The method according to any one of claims 1 to 4,

CHARACTERIZED BY storing the determined values of the optical layer parameters of a layer and using the stored values as reference values for a next layer during a layer change.

6. The method according to any one of claims 1 to 5, characterized by the additional evaluation of a reflection intensity, which is caused by coherent light of a wavelength, to account for the surface roughness of the deposited layer.

7. Arrangement for carrying out the method according to one of claims 1 to 6 with at least one light source, an optical detector and an evaluation unit,

CHARACTERIZED BY at least one optical imaging device as an optical detector (D) for the colors red, blue and green.

8. Arrangement according to claim 7,

CHARACTERIZED BY a CCD or CMOS camera as an optical imaging device.

9. Arrangement according to one of claims 1 to 6 or 7 or 8,

CHARACTERIZED BY attaching at least the optical detector (D) in the bottom of a vapor deposition chamber (PVD) for deposition of optically thin layers (ZFI, ZF ₂ ) from the vapor phase on a substrate (S).

10. Arrangement according to one of claims 1 to 6 or 7 to 9,

Marked by evaporation for the deposition of optically thin layers (Z _F i, Z _F2 ) from the vapor phase on a substrate (S) used (A, B, C, D, X, Y) and / or a white light source as a light source (ILQ , KLQ).

11. Arrangement according to one of claims 1 to 6 or 7 to 10,

CHARACTERIZED BY a shuttle between the light source (ILQ, KLQ) and the deposited layer (ZF- _I , ZF ₂ ).

12. Application of the method according to one of claims 1 to 6 as an in situ process control in the production of vapor deposited on a substrate (S) deposited optically thin layers (Z _F -ι, Z _F 2) in conjunction düng with a control loop for controlling the calculated actual values of the optical layer parameters (nkompiex, d _SC hic _h t) of the optically thin layers (Zn, Z _F2 ) as control variables to predetermined setpoints by adjusting the production parameters as manipulated variables.

13. Application according to claim 12 in processes with an additional production of optical thick layers with an estimation of the stoichiometry by interpreting the reflection intensity R of the diffused light scattering emanating from the layer deposited on the substrate (S) outside the total reflection in at least two wavelengths λi, X ₂ .

14. Application according to claim 12 or 13 with a simultaneous temperature monitoring of the evaporation sources (A, B, C, D, X, Y) via the at least one spatially resolved optical detector (D).

15. Application according to one of claims 12 to 14 for rough substrates (S) or rough deposited layers (Z _F i, Z ^ z) -