Czochralski Puller Research Articles

Adjustment of oxygen transport phenomena for Czochralski silicon crystal growth

During silicon crystal growth, oxygen, a well-known major impurity, affects the final silicon wafer's mechanical and electrical properties. This study focused on regulation of discharge of different concentrations of oxygen from the quartz crucible into the silicon melt while considering the crucible angular speed and the friction at the melt–crucible interface. The three-dimensional transient governing equations for heat transfer, fluid flow, and impurity transportation in the Czochralski (CZ) puller were solved numerically. The oxygen solvation equation representing the crucible to silicon melt was modified to evaluate the accuracy of oxygen concentration calculations during the CZ process. Experimental measurements using the Fourier-transform infrared (FTIR) technique were used to confirm the simulation results. The results demonstrate that the crucible angular speed affects the oxygen concentration near the crucible wall and therefore in the silicon ingot. The proposed modifications for evaluating oxygen concentration offer a more comprehensive understanding of the oxygen dynamics during the CZ crystal growth.

A cost-effective approach for transitioning to larger wafer production by small Czochralski pullers

As the semiconductor industry continues to evolve, the demand for larger silicon wafers has grown significantly. However, transitioning from traditional 4 and 6-inch wafers to larger sizes poses numerous challenges, especially for smaller semiconductor companies with limited financial resources. This article explores a cost- effective approach to address this transition by modifying a small Czochralski (CZ) puller to grow a larger silicon wafer. The study employed a simulation-based method using 3D transient finite element analysis to evaluate the effects of modifying the CZ puller's B-chamber and thermal shield and applying a cooling jacket. The results show that these modifications can enhance the temperature gradient during crystal growth and improve stability when growing an 8-inch silicon ingot. Practical experiments also highlight the complexities involved in achieving a successful transition to 8-inch wafers, including the need for precise adjustments in gap distances and argon gas volume.

Czochralski (CZ) process modification with cooling tube in the response to market Global silicon shortage

Recently, the silicon wafer producers, affected by Covid-19 and USA-China competition, looks for new production processes to increase the production. On the other hand, the common parts of CZ puller such as heater, crucible and thermal shield are optimized over time and now the common CZ process is reached to limitation for further improvement. Here, we propose a modified CZ method by adding a cooling tube into the growth zone. The new proposed Cz method is applied to the 8″ crystal growth process. A fully 3D transition model including energy equation, Navier–Stokes equation, surface-to-surface radiation heat transfer, moving mesh and thermal stress equations is implemented. The simulation is performed for both original and new CZ method. It was proved that the new CZ method increases the pulling speed up to 25 %. To ensure about the crystal quality, the thermal stress is compared between original and new proposed CZ method. Although it was found that the thermal stress increases about twice but still the maximum von Mises stress never exceeds the critical value 25 MPa. Additionally, the power consumption is also found to enhance maximum 2 kW under new conditions. To evaluate the model the interface and heater power for the original CZ puller is compared with industrial CZ process and it shows acceptable accuracy.

Physically-based, lumped-parameter models for the prediction of oxygen concentration during Czochralski growth of silicon crystals

Lumped-parameter models are derived from boundary layer and other physical arguments to describe oxygen concentration levels during the Czochralski (CZ) growth of silicon. These models are assessed against predictions from a detailed, high-fidelity 2D-3D numerical simulation of the entire CZ puller, whose solutions are realistic but require intense computational effort. Comparisons of predictions show that the lumped-parameter model captures the correct trends of melt oxygen levels influenced by melt height, crucible rotation, and crystal rotation. A simple fitting of coefficients provides reasonably good quantitative predictions by the lumped-parameter model, and its near-instantaneous computations make it an interesting candidate for real-time growth optimization and control. Possible model improvements and extensions are discussed.

Design and Development of Nano-Resolution Wireless Czochralski System for High Quality Crystal Growth Applications

Remotely operated Czochralski puller, Automatic Crystal Puller System (ACPS) with a translation nano-resolution of 20 nm, is designed and developed for high quality crystal growth applications. The system comprises the Programmable System-on-chip hardware with integrated firmware and user-friendly graphical user interface. The controller is attached to high precision specially designed mechanical crystal pulling system. ACPS provides controlled bi-directional rotation speed of 0.01–59.99 rpm and translation motion of 0.001–99.99 mm/h. The wireless operation of ACPS from remote desktop devices is implemented using ZigBee Radio-frequency modules, which facilitates controlling the system for user without entering into the furnace room. The performance of the system was evaluated by growing KCl single crystals and analysing the quality of the grown crystals using various techniques. This paper describes the design and development of wireless nano-resolution Czochralski puller and evaluation of performance for real time crystal growth applications.The developed system can be used to grow many technologically important crystals with optimized growth procedures.

Experimental investigation on the buoyancy-induced flow in a model of the Czochralski crystal growth process

Within this paper we present a model experiment focusing on investigations of the flow field in a Czochralski puller. Low melting point liquid metals as GaInSn are an important tool to investigate the flow structure for such industrial processes. The topology of the prevailing thermally-driven convection might be very complex and is mainly determined by the aspect ratio of the liquid volume and the strength of the convection described by the characteristic dimensionless Grashof number. The measurements of the fluid flow have been conducted by means of the ultrasound Doppler velocimetry (UDV) with and without the influence of external magnetic fields. Two kinds of sensor configurations were used to investigate the flow. Firstly, measurements of the radial velocity component by means of single UDV transducers were carried out shortly below the melt surface across the entire diameter of the cylindrical liquid column at various azimuthal angles. Secondly, a vertically arranged UDV array was applied at the side of the cylinder to obtain more detailed information about the radial velocities in the covered meridional plane. The results reveal the complex flow structure of natural convection in a Czochralski crucible which gains in complexity with applied external magnetic fields.

Ultrasonic flow measurements in a model of a Czochralski puller

An experimental study of the buoyancy-induced flow in a model of a Czochralski crystal growth system was conducted. Ultrasonic velocimetry was used to measure fluid velocities. To approach the thermal boundary conditions in an industrial growth facility, a double-walled glass crucible, which is flown through by a heating fluid, was chosen to hold the fluid. Similarity of the heat transfer conditions was achieved by selecting a liquid metal as the fluid under investigation, which was the ternary alloy GaInSn having a Prandtl number of 0.021. Because of the double-walled crucible, measurements through the container wall are difficult if ever possible. Since the availability of relatively short ultrasonic transducers it is practicable to have the sensor immersed into the fluid. Measurements of the radial velocity component shortly below the melt surface across the entire diameter of the crucible at various azimuthal angles reveal the complex flow structure of natural convection in a Czochralski crucible. As it is not to be expected to grow high quality mono-crystalline crystals from such a non-axisymmetric flow, rotating magnetic fields (RMF) are often proposed to render the flow more axisymmetric. This paper also addresses the question what happens to the buoyancy-driven flow when such an RMF is applied.

Experimental investigation on the electromagnetically controlled buoyancy-induced flow in a model of a Czochralski puller

A great deal of work with respect to both physical and numerical modelling of Czochralski crystal growth has been conducted in the generic Rayleigh-Benard system. In order to come closer to the conditions in a real Czochralski puller, specific effects such as the influence of a rounded crucible bottom, deviations of the thermal boundary conditions from the generic case, crucible and/or crystal rotation, and the influence of magnetic fields are often studied separately. Within this paper we present a model experiment focusing on investigations of the impact of magnetic fields on the flow in a Czochralski puller. To achieve similar thermal boundary conditions as in an industrial growth facility, a double-walled rounded bottom glass crucible was chosen to hold the fluid. Similarity of the heat transfer conditions was guaranteed by selecting the ternary alloy GaInSn as the model fluid. Measurements of the fluid flow have been conducted by means of the ultrasound Doppler velocimetry. The results reveal the complex flow structure of natural convection in a Czochralski crucible. Because the growth of high quality mono-crystalline crystals is impeded by such a non-axisymmetric flow, rotating magnetic fields (RMF) are often proposed to render the flow more axisymmetric. To study the effect of an RMF on the natural convection in a Czochralski system the experimental apparatus was mounted inside the home-made MULTIpurpose MAGnetic field facility (MULTIMAG). In the present contribution the three-dimensional convective patterns as well as the resulting temperature fluctuations will be discussed both for the pure buoyant case and for the application of an RMF.

Model experiments for the Czochralski crystal growth technique

A lot of the physical and the numerical modeling of Czochralski crystal growth is done on the generic Rayleigh-Benard system. To better approximate the conditions in a Czochralski puller, the influences of a rounded crucible bottom, deviations of the thermal boundary conditions from the generic case, crucible and/or crystal rotation, and the influence of magnetic fields are often studied separately. The present contribution reviews some of these topics while concentrating on studies of the flow and related temperature fluctuations in systems where a rotating magnetic field (RMF) was applied. The three-dimensional convective patterns and the resulting temperature fluctuations will be discussed both for the mere buoyant case and for the application of an RMF. It is shown that a system between a Rayleigh-Benard and a more realistic configuration, which is still cylindrical but whose surface is partially covered by a crystal model, behaves much the same as a Rayleigh-Benard system. An RMF can be used to damp the temperature fluctuations. Secondly, a more Czochralski-like system is examined. It turns out that the RMF does not provide the desired damping of the temperature fluctutions in the parameter range considered.

シリコン単結晶におけるボイド欠陥の界面エネルギーの第一原理計算

By means of first principles calculation we studied the interfacial energy of void defects which were formed during the crystal cooling process in a Czochralski(CZ) puller. In this study the interfacial structure of void defect was assumed to be either a vacuum/oxide film/Si(111) structure or a vacuum/Si(111) structure. The calculation result showed that the interfacial energy was 0.886 Jm−2 in the case with an oxide film. It was close to the value(0.830 Jm−2) which was derived from the viewpoint of void defects formation around 1373 K. And it was 1.551 Jm−2 in the case without an oxide film, which was twice as large than an interface with an oxide film. For the range of Si crystal growth conditions for semiconductor use, we calculated the size and density of void defects with the interfacial energy which was obtained from first principles calculation in the case with an oxide film. The results agreed with the experimental results.

Temperature fluctuations in a LiNbO3 melt during crystal growth

Variations in temperature induced by forced convection on the surface of a LiNbO3 melt during crystal growth have been studied. Temperature measurements on the melt surface of single crystals growing (∅ 50mm) at rotation rates of 15–40rpm on an RF-heated Czochralski puller has revealed that the melt surface continuously alternates between a steady and unsteady state of flow. This was attributed to the intermittently turbulent flow mode at intermediate rotation rates. The fluctuation period is thought to depend on the thickness of its boundary layer. The boundary layer varies in thickness due to the melt flow, which stops as the interface moves toward the crystal and resumes once the interface reverts to its former position. By contrast, at above 60rpm, the melt surface temperature drops without fluctuation, indicating that turbulent flow is dominant at faster rotation rates.

Thermal history simulation of Czochralski silicon crystals and its application to the study of defects formation during crystal growth

A finite element quasi-steady-state modeling of temperature distribution during the growth of a silicon crystal by the Czochralski (CZ) method is presented. Computations are performed for a comprehensive geometrical model of an actual CZ puller, accounting for detailed radiation heat exchange and introducing a reviewed set of temperature-dependent thermophysical properties. The crystal growth process is simulated step-wise, with automatic update of the mesh at each successive growth step. Temperature and pull rate data are then input to a mathematical routine for the calculation of the dwell time for user-defined temperature intervals and crystal position. Applications of this simulation tool for the interpretation of crystal quality characteristics such as oxygen precipitation and light scattering defect axial profiles are presented.

Oxygen distribution in silicon melt during a standard Czochralski process studied by sensor measurements and comparison to numerical simulation

The distribution of oxygen in silicon melts kept in a 14″ diameter silica crucible of a standard Czochralski puller was measured using an electrochemical oxygen-sensor. Three different situations were analyzed; without the presence of a crystal and with 4″ diameter crystals being pulled at crucible rotation rates of 2 and 5 rpm. The experimental results are discussed within the frame of existing modelling results in the literature and own simulations. It is found that boundary conditions and models which are presently proposed in the literature are obviously not adequate to provide quantitative correlation with experimental data.

Oxygen incorporation and precipitation behavior in heavily boron-doped Czochralski silicon crystals

To understand the effect of heavy boron-doping, p+-type (100) silicon crystals doped with boron in the order of 1 × 1019atoms cm−3 (0.006–0.01 Ω cm) and a p-type (100) silicon crystal doped with boron in the order of 1 × 1015atoms cm−3 (12–16 Ω cm) were grown in a Czochralski puller and studied. The oxygen contents in the p+-type crystals were determined by gas fusion analysis and shown to be on the average about 23% higher than those in a comparable p-type crystal. Etch pit features of oxide precipitates after a high-low-high annealing were generally different between the two types of silicon, and three distinct regimes of behavior about the critical boron and oxygen concentrations were identified. Those are: with resistivity ≳ 0.007 Ω cmor oxygen concentration ≲ 30 ppma (Old ASTM), the oxide precipitate density, in the order of 1010 cm−3, is basically unaffected by heavy boron-doping and dependent on oxygen concentration; with resistivity ≲ 0.007 Ω cmand oxygen concentration ≳ 37 ppma, however, large platelets, which are 10–15 μm in size and in the order of 107 cm−3 in density, are formed instead and completely gettered oxide precipitates. In a medium-oxygen range of about 28–30 ppma, both p- and p+-type (100) silicon wafers formed an ideal internal gettering structure, consisting of a well-defined precipitate-free zone (PFZ) to a depth of about 15 μm from the surface and dense bulk oxide precipitates in the order of 1010 cm−3 below the PFZ, after a high-low-high annealing.

Development of an Electrochemical Oxygen Sensor for Czochralski Silicon Melts

An electrochemical oxygen sensor based on stabilized zirconia was developed for measurements of the oxygen concentration in molten silicon. The mixture of Mn and was found to be an appropriate system for the generation of the reference oxygen partial pressure which should be close to that of oxygen in the silicon melt. The free‐energy of solution of oxygen in molten silicon was evaluated by calibration experiments. A prototype sensor was tested to measure the oxygen content in the silicon melt in a commercial Czochralski puller. The difference in the radial oxygen concentration between center and periphery increased with increasing crucible rotation. This result is consistent with the experience in silicon crystal growth as well as with recent numerical and experimental results on the impact of crucible rotation on the flow patterns in silicon melts in an industrial‐size Czochralski crucible.

Study of characteristics of the crystal temperature in a Czochralski puller through experiment and simulation

The characteristics of the temperature of the 2 inch Si crystal in a CZ puller are studied through experiments and simulations. The mechanism determining the temperature is briefly described. The principal results are: (1) the temperature was fundamentally determined by the distance from the surface of the melt, (2) at the same distance, the temperature at the top portion of the crystal was higher than that at other portions, and (3) for a pull-rate of less than 1 mm/min, the temperature was almost the same as that with zero pull-rate.

Vacuum Bakeout Effect on Ambient Gas in a High Pressure LEC Puller

H2 and CO generation from graphite furnace components in a high-pressure liquid-encapsulated Czochralski (LEC) puller was studied in detail from the standpoint of reproducible contamination to the LEC charge (B2O3 encapsulant/GaAs melt). It has been found that H2 and CO contamination from the graphite furnace components varies widely according to the vacuum level as well as to the high-temperature duration time of the vacuum bakeout process. It has been concluded that controlling the vacuum bakeout condition is the key to achieving reproducible generation of H2 and CO during LEC GaAs single-crystal growth.

Computer modelling of heat transfer in Czochralski silicon crystal growth

A finite element quasi-steady state modelling of heat transfer during the growth of a silicon single crystal by the Czochralski (CZ) method is presented. Computations are performed for a comprehensive geometrical model of an actual CZ puller, accounting for detailed radiation heat exchange and introducing a reviewed set of temperature dependent thermophysical properties. The implemented problem solving procedure goes through the recursive scaling of the power yield in order to guarantee the melt/crystal/gas tri-junction temperature matches the melting one. An effective, time saving, numerical scheme is employed for the coupled convergence of both melt/crystal interface location and temperature distribution. Many different growth conditions are simulated for analyzing the sensitivity of the heat flow pattern versus changes of pull rate, gas convection and material properties. The results stress the strong influence of growth rate on the phase change interface deflection, while the input of a correct melt emissivity value is shown to be imperative for working out reliable predictions of the crystal thermal field. The computed temperatures are compared with the corresponding thermographic data collected by means of an infrared imaging system, allowing a satisfactory validation of the adopted mathematical model.

Turbulent heat transfer through the melt in silicon Czochralski growth

The effects of crucible rotation rate on the melt temperature are measured in a production scale Si Czochralski puller in order to examine a k-ϵ model of turbulence as a heat transfer model. The experimental results show that (1) the time-averaged temperature of melt surface decreases by the order of 10 K with increasing crucible rotation rate from 2 to 10 rpm and (2) the melt temperature fluctuates with the magnitude of 10–20 K peak-to-peak. The analytical results based on the k-ϵ model reproduce the nature of experimental results. The use of the turbulence model is reasonable for thermal modeling of Si Czochralski growth.

Melt motion in a Czochralski puller with a weak transverse magnetic field

The melt motion in a Czochralski silicon crystal puller with a uniform transverse magnetic field is intrinsically three-dimensional. When the magnetic field is relatively weak, the motion consists of an axisymmetric base solution, plus two perturbations which behave as sin 2θ and cos 2θ, neglecting terms which are proportional to the fourth power of the magnetic field strength. A 700 G magnetic field stabilizes a melt motion which is unsteady without a magnetic field. As the strength of the magnetic field increases, the magnitudes of the axial and radial velocities increase to maxima at some field strength above 1000 G. This contrasts with an axial magnetic field for which the axial and radial velocities always decrease as the field strength increases. We estimate that 1500 G is the maximum field strength for which our weak-field asymptotic solution is valid.