KR20040080724A - Sputter process simulation method using monte carlo method - Google Patents

Abstract

PURPOSE: A sputter process simulation method using a Monte-Carlo method is provided to predict topography of a substrate by considering a scattering effect between target atoms and calculating emission distribution of the target atoms. CONSTITUTION: A simulator sets up a processor and performs a reset process(201). The processor performs a calculation process. Each processor calculates information of a target from an input variable(203). Each processor calculates stopping force before an ion implantation process is performed(204). The ion implantation process is performed(205). Each processor calculates tracks of all atoms(206). Each processor calculates a collision phenomenon by using a chain collision algorithm(207). The implanted ion energy is compared with the predetermined energy of the target(208). Ions are implanted when the implanted ion energy is less than the predetermined energy of the target(209). Output results of each processor are formed after the simulation process is completed(210). The output results are transmitted to a central processor(211). The central processor processes the output results and transmits the processed results to a postprocess system(212).

Description

SPUTTER PROCESS SIMULATION METHOD USING MONTE CARLO METHOD}

The present invention simulates the sputter deposition process, one of the most widely known methods for metal thin film deposition in semiconductor integrated circuit fabrication, and numerically calculates the three-dimensional behavior of sputter target atoms using an algorithm for efficient computation time. And a three-dimensional emission distribution calculation of the sputter rate at the target and the target atom.

The sputter deposition method is one of the most essential methods for thin film deposition because it can deposit various materials easily and efficiently. Because of the advantages of the sputter process, major semiconductor processes are adopting the sputter process and are spurring the development of improved sputter reactor systems. However, to reduce development costs and risks, it is necessary to simulate the distribution of sputter deposition atoms prior to putting them on the line.

The trajectory of the particles emitted from the target and the scattering of the particles within the target provide a fundamental reason for predicting accurate sputter deposition rates. In order to calculate the emission distribution of target atoms, many research institutes have attempted to use analytical functions. D. S. Bang described the emission distribution of target atoms by using an analysis function defined by the cosine function at the IEEE IEDM International Conference in 1995.

Another method of calculating the emission distribution of sputter target atoms is the molecular dynamics method presented by Y. Yamada at the IEEE IEDM International Conference in 1995, JF Ziegler and Y. Yamamura, respectively, in the 1984 Applied Physics Paper and 1995 Journal. In the article of Vacuum Science Technology, the Monte Carlo method was used to report the sputter yield and emission distribution of target atoms.

However, the approach using the analytical function may be faster in terms of simulation speed, but has a disadvantage in that the applicability and accuracy of the results are much lower than those of the physical phenomena-based Monte Carlo method and molecular dynamics method. Hereinafter, a problem with a conventional analysis function will be described in detail with reference to FIG. 1A of the accompanying drawings.

That is, in the equation, the target erosion (Er) and the target emission (Em) are based on experimental data and are a combination of the cosine function that functions as a function of the distance (R) and the angle of incidence between the substrate and the target. Implement the distribution of atoms released from the target. However, since these analytical functions depend on experimental data, it is difficult to implement sputter rate and emission distribution for various materials, and it is difficult to realize sputter emission distribution according to the incident angle of ions with only a simple cosine function.

Molecular dynamics method calculates all binding energy between particles with various components related to the force between atoms, so that more accurate and reliable result can be obtained. However, the molecular dynamics method takes into account the binding energy of the implanted ions and all target atoms in the collision calculation between atoms, so that the speed of the simulation decreases significantly as the number of target atoms increases, resulting in the use of positive computer memory. It is far behind in other ways.

In contrast, the Monte Carlo method assumes binary collisions between atoms, and considers scattering between atoms using random numbers, which is superior to other methods in terms of computer efficiency and accuracy of results, and is most widely used. However, the existing Monte Carlo method sputter simulation is also pointed out that the speed is still a big problem, it is necessary to develop a method that can solve these problems.

Accordingly, a first object of the present invention is to develop a numerical analyzer capable of predicting the topographical shape of a substrate by considering the scattering effect between target atoms using the Monte Carlo method and calculating the emission distribution of atoms from the target.

The second object of the present invention, in addition to the first object, is to apply a new function and parallel using a CRAY T3E supercomputer to reduce the computer execution time problem, which is a big problem in the conventional numerical analysis method based on physical phenomena. By applying a computing method, the user is provided with a result in a short time.

In addition to the first object, a third object of the present invention is to fabricate a simulator that can be directly applied to various processes including sputter deposition, a sputter etching process, a bias sputtering process, and an ion beam sputtering system.

1A is an emission distribution diagram of sputter target atoms shown using a conventional analytical method.

2A is a flowchart of a sputter simulator by a parallel processing method.

Figure 2b is an algorithm showing a chain collision process between the incident ions and target atoms.

2C is a diagram showing a depth limit function for shortening a simulation execution time.

<Explanation of symbols for the main parts of the drawings>

301: primary knock on atom (PKA) causing the collision

302: secondary knock on atom (SKA)

303: one-dimensional array for storing various information of atoms

401: Implantation Ion

402: target atom

403 simulation depth limit

404: energy in the reduced coordinate system of implanted ions

405: screening length of the target atom

406: atomic density of the target atom

In order to achieve the above object, the present invention is to apply a parallel computation to increase the computation time, to apply a function that limits the simulation depth of the target, to calculate the collision scattering angle between atoms using the Monte Carlo method And a step of calculating a chain scattering effect of the target internal atoms, a step of calculating a distribution of atoms released from the target.

Hereinafter, a preferred embodiment of a method for manufacturing a sputter process simulator according to the present invention will be described in detail with reference to FIGS. 2A to 2C.

Figure 2a shows a parallel processing algorithm using the Monte Carlo method. Upon receiving the process condition from the user initially, the simulator sets up and initializes a processor to be used for parallel processing (201). Processors to which each operation is assigned perform calculations independently of the other processors. Each processor calculates 203 material information of the target from a given input variable 202 and performs an atomic stop force calculation 204 before implantation of ions.

When ions are implanted (205), the traces of all atoms generated at the time of collision are calculated (206) and based on the chain collision algorithm (207), the collision phenomenon that a collided atom can generate is calculated. The energy of the implanted ions is compared 208 with a predetermined energy inside the target at every collision and if less than this energy is defined as stationary and implanted 209 with other ions. When the simulation of all user-defined ions is completed, an output result of each processor is formed 210, and the result is transmitted to the central processor (211). The central processor properly processes the received data and then transmits the data to the post-processing system 212 to show the final result.

FIG. 2b illustrates an algorithm for considering the effect of chain collision between atoms during the collision process of FIG. 2a. The chain collision algorithm of the sputter simulator is continuously performed until the knock-on atom that has collided with the incident ions stops in the target or is emitted to the outside and can no longer perform calculations. That is, the collided particles collide with other atoms and this series of processes is carried out continuously. In this case, an atom that collides with energy is defined as a primary knock-on atom (PKA), and an atom that is energized by a collision of PKA is a secondary knock-on atom (SKA). To be defined.

Atoms first impacted by ions are labeled i = 1, and after collision with new atoms, PKA is labeled i + 1 and SKA is labeled i. If the PKA is sputtered or less than the lattice energy of the target atom and no longer performs a chain collision, the program will go back to label i and perform the previous process again. At this time, SKA becomes PKA to perform calculation in the same manner as PKA. After this, it finally reaches i = 0 (to the ion itself). The ions then collide with the next atom to perform the above process again. Each atom moves with information about its depth, current energy, location, and direction of movement on the target surface.

In the sputter simulation, the implanted ions collide with stationary target atoms to transfer energy. At this time, the stationary target atom receives energy and repeats a process of colliding with another target atom, and continues until all particles stop inside or are released to the target. In this series of processes, the simulation execution time increases in proportion to the overall number of collisions, and the larger the energy of implanted ions, the longer the simulation execution time.

However, in the sputtering process, scattering in the target is not important because only the distribution of particles emitted to the outside is considered, and only the collision process on the target surface should be considered by introducing an appropriate function for efficiency of execution time.

Figure 2c is a diagram showing the collision depth limit function between the particles applied in the present invention, implemented as a function of the energy 404 of the implanted ion in the reduced coordinate system, the screening length 405 of the target atom, the atomic density of the target 406 do. As a result of the application of the function, the execution time of the simulation was reduced up to 5 times or more.

The foregoing has outlined rather broadly the features and technical advantages of the present invention to better understand the claims of the invention which will be described later. Additional features and advantages that make up the claims of the present invention will be described below. It should be appreciated by those skilled in the art that the conception and specific embodiments of the disclosed subject matter can be used immediately as a basis for the fabrication or modification of other simulators for carrying out similar purposes to the present invention.

That is, in the second preferred embodiment of the present invention, the sputter rate and emission distribution of the target atoms provided by the simulator may be applied to various processes such as a sputter etching process, a bias sputtering system, and an ion beam sputtering system to predict a result. Could be used.

In addition, as a third preferred embodiment of the present invention, it is possible to predict the topography of the substrate during the sputtering process by providing an input to the surface evolution simulator based on the results of the simulator.

The inventive concepts and embodiments disclosed in the present invention may be used by those skilled in the art as a basis for making or modifying other simulators for carrying out the same purposes of the present invention. In addition, such modified or modified equivalent numerical solver by those skilled in the art may be various changes, substitutions and changes without departing from the spirit or scope of the invention described in the claims.

As described above, the fabrication of the semiconductor process simulator according to the present invention is to solve and supplement the problems of the conventional sputter simulator, and the present invention calculates the scattering process between atoms using the Monte Carlo method based on a physical mechanism. In addition, by applying parallel computing for efficient calculation time, the user can easily obtain the sputter process result without depending on the experiment. In addition, the sputter rate and emission distribution of the target atom provided by the simulator has flexibility that can be applied to various processes such as a sputter etching process, a bias sputtering system, and an ion beam sputtering system.

Claims (2)

Applying parallel computation to increase computation time; Applying a function that limits the simulation depth of the target; Calculating the collision scattering angle between atoms using the Monte Carlo method; Calculating chain scattering effects of target internal atoms; Calculating the distribution of atoms released from the target Method for producing a semiconductor sputter process simulator, characterized in that. The method for manufacturing a semiconductor process simulator according to claim 1, which can be applied to a sputter etching process and a bias sputtering process using a chain collision algorithm within a target.