rus
Issue #4/2017
S.Konakov
Localized microreactor deposition of thin films and nanostructures as new approach to investigation of chemical vapor deposition
Localized microreactor deposition of thin films and nanostructures as new approach to investigation of chemical vapor deposition
The paper analyzes the current state of the chemical vapor deposition (CVD). The basic problem of the technology is described and the fundamental reason for its existence is analyzed. The paper demonstrates a constant deficit of the experimental data on the results of CVD, as well as inadequate approaches to experimental research. All of this leads to the poor solution of the optimization problem of thin film deposition with desired physical and chemical parameters. We are firstly proposing to use a microreactor CVD. The definition of microreactor CVD is formulated in this paper, and the basic properties of the technology are described.
Теги: chemical vapor deposition microreactor thin films xимическое осаждение из газовой фазы μr cvd микрореактор мр хогф тонкие пленки
The Chemical Vapor Deposition (CVD) is one of the basic technologies of modern micro- and nanoelectronics [1]. It is used for creation of metal, semiconductor and insulating layers with thickness up to hundreds of microns on substrates [2]. CVD and its modifications are a universal technology for thin films deposition including ones with nanostructure. The current stage of development of the CVD method has the following features:
• wide range of deposited materials. Hundreds of substances can be obtained by the CVD method [2];
• discontinuity of scientific knowledge. There are a large number of papers on the study of the compounds used in the manufacture of high-tech products (SiO2, Si3N4, TiO2, C, SiC, GaAs, Ni, etc.). Other substances are poorly investigated, and only about a dozen of papers are devoted on the deposition of some materials, such as ZrB2 [3];
• сomplexity of the physical-chemical processes occurring during CVD and insufficient knowledge on it. The process of CVD includes many phenomena at different spatial and temporal scales, so their detailed study is very time consuming.
These features lead to the fact that the development of new technological processes of CVD faces considerable practical difficulties. Most of them can be reduced to the optimization of technology for obtaining coatings with desired physical-chemical properties. At the present stage of development of the CVD, the optimization problem is most actual for scientific and engineering practice.
The main problem of CVD can be formulated as a lack of a theoretically justified model of the process, which would have sufficient accuracy in the description of the experimental data and would be suitable for solving optimization problems. Currently, there is no generic approaches that would allow to optimize the process of CVD on the properties of the deposited coatings or other parameters.
OBJECTIVE REASONS
OF SCIENTIFIC PROBLEM
To investigate the causes of the problem we will formulate the stages of creation of the task for optimization of CVD. In practice, in the first stage, the requirements to the functional properties of the coating are formulated, and not a single parameter but a set of properties is considered, so in general case we have the task of multiple parameter optimization. In the second stage, the requirements to the film on the micro level, for example, the constraints on the composition, impurities, defects etc. can be formulated. The third stage is the choice of the basic technological schemes of the CVD, that is, the choice of precursors, method of initiating a chemical reaction, process equipment. This determines the number of important process parameters and the range of their changes. Next, the task of optimization to find such combination of process parameters of CVD process, which would ensure deposition of a film with a set of the required characteristics differing from those specified within a tolerance, is formulated.
To simplify the task by using a discrete set of possible values of arguments, that is, considering some array of points in the entire region of change of technological parameters.
We can simplify the task by using a discrete set of possible values of arguments, that is, considering some array of points in the entire region of change of process parameters. In this case, each point corresponds to conditions of the particular experiment of CVD. Then the direct solution of the optimization task is possible, which consists in conducting experiments at all points of the considered range of parameters. and then according to the measured properties of the precipitated film are selected values of that satisfy the problem conditions. After this, based on the measured properties of the deposited film the values that satisfy the task conditions are selected. It is possible to estimate the number of required experiments, if to consider the substrate temperature, pressure, gas flow rate, chemical composition of the gas mixture (for example, of two independent components) as the main process parameters. If each of these five parameters can take only ten different values, the total number of the experiments required will be 105. In the traditional approach to the implementation of CVD and at the rational expenditure of resources, it is impossible to conduct such number of experiments. Hence, it can be concluded that the practical solution of the problem of optimization of the CVD process is carried out with a significant shortage of experimental information. This is a key feature of the current state of researches, which directly affects the optimization potential.
NEW APPROACH TO CONDUCTING EXPERIMENTS ON RESEARCH
OF CVD PROCESS
A radical change in this situation requires a new approach that would allow more than 105 experiments whit use of a rational amount of resources. This approach should satisfy the following conditions:
• low cost per experiment. This requires minimization of the amount of process materials (precursor gases, substrate material);
• maximum automation of the processes of deposition and investigation of properties of obtained materials. Automation can reduce costs per operation.
Thus, a new CVD process requires the use of automatic processing line in which the deposition of a thin film is carried out with minimal consumption of gas reactants in a small area of the substrate, as if "pointwise". At the same time, the area of the deposition should shift relative to the substrate with the formation of an array of such "points". For the analysis of this array, it is necessary to use processing equipment, which allows automatically scanning and mapping of the substrate. This process is standard for companies that produced microelectronics. Thus, the only missing component for implementation of the described technique is the method of the point-deposition.
MICROREACTOR CVD PROCESS
The question of local, point-deposition of materials have been solved by the development of specialized technology, called Microreactor Chemical Vapor Deposition (μR CVD). We introduce two definitions:
• microreactor Chemical Vapor Deposition is a modification of the CVD method, the main distinctive feature of which is the deposition in a localized region on the surface of the substrate with characteristic dimensions of 1 mm and less, which is achieved due to the special way of the jet flow of the gas into the area above the substrate;
• microreactor for μR CVD is a MEMS device, which thanks to a system of microchannels provides a special configuration of the output gas jet to localize the area of deposition on a substrate located at a distance from units to hundreds of microns from the microreactor.
The use of microreactor enables local, point-deposition of material on a substrate. Accordingly, it is the main functional unit of the processing line for conducting a large number of experiments on CVD. It also includes a system to move the microreactor relative to the substrate, vacuum and gas systems.
PRINCIPLE OF LOCALIZATION
OF DEPOSITION IN MICROREACTOR
A key feature of the μR CVD method is the localization of the area of deposition. The physical causes of this phenomenon can be illustrated on the example of the simple design of a microreactor with a single microchannel. This model uses a point source of gaseous phase that moves in a plane gap formed by the walls of the microreactor and the substrate, as shown in Fig.1.
During the reaction of film deposition the source components of the gas mixture are consumed, and their concentration will decrease with distance from the source. The corresponding distribution of the concentration of the reactants over the substrate determines the distribution of the deposition rate. Using the simulation of the reactor in the mode of ideal displacement, it is possible to obtain an analitical dependence of the distribution of the deposition rate on the distance to the source. Modeling with numerical methods of solving problems in the mechanics of viscous reactive gas also shows the presence of localization of the deposition rate. Papers [4, 5] contain the descriptions and results of calculations for some μR CVD processes. Fig.2 shows a typical distribution graph of the deposition rate in a microreactor.
For operational numerical estimates of deposition profile you can use its half-width, which is determined on the graph of dependence of the thickness of the deposited film or deposition rate on the radius (this is direct data obtained in the measurement on the profilometer). Analysis of physical-chemical processes in the microreactor allows to formulate the main factors that affect the distribution of the deposition rate:
• rate of surface chemical reactions;
• temperature of the substrate;
• pressure of the gas over the substrate;
• concentration of initial components in the gas mixture;
• rate of gas flow in the gap between the microreactor and the substrate.
Even a common review shows that, using the directly measured value of half-width of deposition profile we can estimate the rate of surface chemical reaction, its dependence on temperature, pressure, concentration of initial substances, and also the impact of diffusion on the deposition process. It is natural that at the same time also the main objective of μR CVD is achieved – carrying out a large number of experiments at rational expense of resources.
For the numerical estimate, let us assume that the standard 100 mm wafers are used as a substrate. We assume that μR CVD allows deposition in an area with a diameter of 100 µm. To eliminate the mutual influences, the neighboring points are placed at a distance of 1 mm from each other. Thus, one substrate can accommodate over 7000 areas of deposition corresponding to the experiments with different conditions. So only 15 wafers is necessary to implement 105 experiments, which is a rational amount even for laboratory studies.
Thus, the proposed approach to experimental study of the CVD process with the use of microreactors allows to effectively conduct experiments to obtain a large set of statistical information. μR CVD is today the only method which is able to solve the problem of lack of experimental information.
CONCLUSION
The present paper describes the main problem of the CVD technology at the present stage of its development – the lack of a theoretically justified model of the process, which would provide sufficient accuracy of the description of the experimental data and would be suitable for solving optimization problems. The reason for this problem is the lack of experimental information. This is caused by the large number of technological parameters that should be included in the model of the process, as well as by the fact that current approaches to the design of experiments on CVD not allow to carry out systematic experimental work covering the entire field of admissible values of technological parameters at the use of a rational amount of resources.
The proposed method of μR CVD differs by implementation of deposition on a small area of the substrate with minimal consumption of precursors, which allows to automatically carry out many experiments (more than 105). Direct evaluation of the results of deposition with the use of standard analytical equipment allows to obtain information about the physical-chemical regularities of the investigated CVD process. Experimental data can be used to solve current problems and for generalization when creating a theoretically justified model of CVD process. The practical implementation of μR CVD will contribute to the development of CVD technology and to the broadening of the scope of its practical application. ■
• wide range of deposited materials. Hundreds of substances can be obtained by the CVD method [2];
• discontinuity of scientific knowledge. There are a large number of papers on the study of the compounds used in the manufacture of high-tech products (SiO2, Si3N4, TiO2, C, SiC, GaAs, Ni, etc.). Other substances are poorly investigated, and only about a dozen of papers are devoted on the deposition of some materials, such as ZrB2 [3];
• сomplexity of the physical-chemical processes occurring during CVD and insufficient knowledge on it. The process of CVD includes many phenomena at different spatial and temporal scales, so their detailed study is very time consuming.
These features lead to the fact that the development of new technological processes of CVD faces considerable practical difficulties. Most of them can be reduced to the optimization of technology for obtaining coatings with desired physical-chemical properties. At the present stage of development of the CVD, the optimization problem is most actual for scientific and engineering practice.
The main problem of CVD can be formulated as a lack of a theoretically justified model of the process, which would have sufficient accuracy in the description of the experimental data and would be suitable for solving optimization problems. Currently, there is no generic approaches that would allow to optimize the process of CVD on the properties of the deposited coatings or other parameters.
OBJECTIVE REASONS
OF SCIENTIFIC PROBLEM
To investigate the causes of the problem we will formulate the stages of creation of the task for optimization of CVD. In practice, in the first stage, the requirements to the functional properties of the coating are formulated, and not a single parameter but a set of properties is considered, so in general case we have the task of multiple parameter optimization. In the second stage, the requirements to the film on the micro level, for example, the constraints on the composition, impurities, defects etc. can be formulated. The third stage is the choice of the basic technological schemes of the CVD, that is, the choice of precursors, method of initiating a chemical reaction, process equipment. This determines the number of important process parameters and the range of their changes. Next, the task of optimization to find such combination of process parameters of CVD process, which would ensure deposition of a film with a set of the required characteristics differing from those specified within a tolerance, is formulated.
To simplify the task by using a discrete set of possible values of arguments, that is, considering some array of points in the entire region of change of technological parameters.
We can simplify the task by using a discrete set of possible values of arguments, that is, considering some array of points in the entire region of change of process parameters. In this case, each point corresponds to conditions of the particular experiment of CVD. Then the direct solution of the optimization task is possible, which consists in conducting experiments at all points of the considered range of parameters. and then according to the measured properties of the precipitated film are selected values of that satisfy the problem conditions. After this, based on the measured properties of the deposited film the values that satisfy the task conditions are selected. It is possible to estimate the number of required experiments, if to consider the substrate temperature, pressure, gas flow rate, chemical composition of the gas mixture (for example, of two independent components) as the main process parameters. If each of these five parameters can take only ten different values, the total number of the experiments required will be 105. In the traditional approach to the implementation of CVD and at the rational expenditure of resources, it is impossible to conduct such number of experiments. Hence, it can be concluded that the practical solution of the problem of optimization of the CVD process is carried out with a significant shortage of experimental information. This is a key feature of the current state of researches, which directly affects the optimization potential.
NEW APPROACH TO CONDUCTING EXPERIMENTS ON RESEARCH
OF CVD PROCESS
A radical change in this situation requires a new approach that would allow more than 105 experiments whit use of a rational amount of resources. This approach should satisfy the following conditions:
• low cost per experiment. This requires minimization of the amount of process materials (precursor gases, substrate material);
• maximum automation of the processes of deposition and investigation of properties of obtained materials. Automation can reduce costs per operation.
Thus, a new CVD process requires the use of automatic processing line in which the deposition of a thin film is carried out with minimal consumption of gas reactants in a small area of the substrate, as if "pointwise". At the same time, the area of the deposition should shift relative to the substrate with the formation of an array of such "points". For the analysis of this array, it is necessary to use processing equipment, which allows automatically scanning and mapping of the substrate. This process is standard for companies that produced microelectronics. Thus, the only missing component for implementation of the described technique is the method of the point-deposition.
MICROREACTOR CVD PROCESS
The question of local, point-deposition of materials have been solved by the development of specialized technology, called Microreactor Chemical Vapor Deposition (μR CVD). We introduce two definitions:
• microreactor Chemical Vapor Deposition is a modification of the CVD method, the main distinctive feature of which is the deposition in a localized region on the surface of the substrate with characteristic dimensions of 1 mm and less, which is achieved due to the special way of the jet flow of the gas into the area above the substrate;
• microreactor for μR CVD is a MEMS device, which thanks to a system of microchannels provides a special configuration of the output gas jet to localize the area of deposition on a substrate located at a distance from units to hundreds of microns from the microreactor.
The use of microreactor enables local, point-deposition of material on a substrate. Accordingly, it is the main functional unit of the processing line for conducting a large number of experiments on CVD. It also includes a system to move the microreactor relative to the substrate, vacuum and gas systems.
PRINCIPLE OF LOCALIZATION
OF DEPOSITION IN MICROREACTOR
A key feature of the μR CVD method is the localization of the area of deposition. The physical causes of this phenomenon can be illustrated on the example of the simple design of a microreactor with a single microchannel. This model uses a point source of gaseous phase that moves in a plane gap formed by the walls of the microreactor and the substrate, as shown in Fig.1.
During the reaction of film deposition the source components of the gas mixture are consumed, and their concentration will decrease with distance from the source. The corresponding distribution of the concentration of the reactants over the substrate determines the distribution of the deposition rate. Using the simulation of the reactor in the mode of ideal displacement, it is possible to obtain an analitical dependence of the distribution of the deposition rate on the distance to the source. Modeling with numerical methods of solving problems in the mechanics of viscous reactive gas also shows the presence of localization of the deposition rate. Papers [4, 5] contain the descriptions and results of calculations for some μR CVD processes. Fig.2 shows a typical distribution graph of the deposition rate in a microreactor.
For operational numerical estimates of deposition profile you can use its half-width, which is determined on the graph of dependence of the thickness of the deposited film or deposition rate on the radius (this is direct data obtained in the measurement on the profilometer). Analysis of physical-chemical processes in the microreactor allows to formulate the main factors that affect the distribution of the deposition rate:
• rate of surface chemical reactions;
• temperature of the substrate;
• pressure of the gas over the substrate;
• concentration of initial components in the gas mixture;
• rate of gas flow in the gap between the microreactor and the substrate.
Even a common review shows that, using the directly measured value of half-width of deposition profile we can estimate the rate of surface chemical reaction, its dependence on temperature, pressure, concentration of initial substances, and also the impact of diffusion on the deposition process. It is natural that at the same time also the main objective of μR CVD is achieved – carrying out a large number of experiments at rational expense of resources.
For the numerical estimate, let us assume that the standard 100 mm wafers are used as a substrate. We assume that μR CVD allows deposition in an area with a diameter of 100 µm. To eliminate the mutual influences, the neighboring points are placed at a distance of 1 mm from each other. Thus, one substrate can accommodate over 7000 areas of deposition corresponding to the experiments with different conditions. So only 15 wafers is necessary to implement 105 experiments, which is a rational amount even for laboratory studies.
Thus, the proposed approach to experimental study of the CVD process with the use of microreactors allows to effectively conduct experiments to obtain a large set of statistical information. μR CVD is today the only method which is able to solve the problem of lack of experimental information.
CONCLUSION
The present paper describes the main problem of the CVD technology at the present stage of its development – the lack of a theoretically justified model of the process, which would provide sufficient accuracy of the description of the experimental data and would be suitable for solving optimization problems. The reason for this problem is the lack of experimental information. This is caused by the large number of technological parameters that should be included in the model of the process, as well as by the fact that current approaches to the design of experiments on CVD not allow to carry out systematic experimental work covering the entire field of admissible values of technological parameters at the use of a rational amount of resources.
The proposed method of μR CVD differs by implementation of deposition on a small area of the substrate with minimal consumption of precursors, which allows to automatically carry out many experiments (more than 105). Direct evaluation of the results of deposition with the use of standard analytical equipment allows to obtain information about the physical-chemical regularities of the investigated CVD process. Experimental data can be used to solve current problems and for generalization when creating a theoretically justified model of CVD process. The practical implementation of μR CVD will contribute to the development of CVD technology and to the broadening of the scope of its practical application. ■
Readers feedback
