The present invention relates to growing crystals without encountering the problems normally associated with conventional Czochralski crystal growth of: strong melt convection that causes deteriorated crystal quality, constantly changing thermal conditions; segregation causing non-uniformity in dopant concentrations; a slow growth rate and an uncontrollable interface shape; slow turn around between crystal ingots; and significant loss of feedstock materials if a growth run fails. The invention eliminates the foregoing problems through the use of a shallow melt with a heat shield and an interface heater-heat sink to overcome microdefects, stress management and defect minimization, swift turnaround between crystal ingots, and the saving of feedstock materials through the use of a double barrier that offsets connecting holes for replenishing the melt.
Affecting crystal growth using the conventional Czochralski (CZ) process has been accomplished utilizing a pulling apparatus employing the CZ technique comprising: A gas type chamber; a crucible for storing a semiconductor melt that is positioned inside the chamber; a heater for heating the semiconductor melt; and a pulling mechanism for pulling a single crystal of the semiconductor. With this apparatus, a seed crystal of the single crystal of the semiconductor is immersed in the semiconductor melt inside the crucible, whereupon the seed crystal is gradually pulled upwards, thereby growing a large diameter single crystal of the semiconductor having the same orientation as the seed crystal.
As the Czochralski method evolved as the prevalent one for the operation of pulling a silicon single crystal, a method of the version adopted a double-wall crucible formed by disposing a cylindrical partition wall in an outer crucible to operate the crucible by supplying solid or fused silicon as a raw material, either batch wise or continuously, into the crucible through a gap between the inner surface of the outer crucible and the outer surface of the cylindrical partition wall and pulling a silicon single crystal from the molten mass of silicon in the cylindrical partition wall.
U.S. Pat. No. 5,871,581 discloses a single crystal pulling apparatus comprising:
A crucible for pulling silicon single crystal is disclosed in U.S. Pat. No. 5,720,809. The crucible is constructed by coaxially disposing a cylindrical partition wall inside an outer crucible for holding a molten mass of silicon as a raw material and operated by heating the outer crucible and meanwhile supplying the raw material silicon to the gap between the outer crucible and the cylindrical partition wall and introducing the consequently produced molten mass of silicon into the interior of the cylindrical partition wall via a passage below the surface of the molten mass interconnecting the outer crucible and the inside of the cylindrical partition wall and meanwhile pulling the single crystal bar from the molten mass of silicon inside the cylindrical partition wall, which double-wall crucible is characterized in that at least the cylindrical partition wall is formed of quartz glass having a hydroxyl group (OH group) content of not more than 30 ppm.
There is a need in the art of crystal growth utilizing the Czochralski method to avoid the problems of: (1) strong melt convection leading to deteriorated crystal quality (high oxygen content when a silica crucible is commonly used, and increased microdefects); 2) constantly changing thermal conditions even with synchronized crucible lift; (3) segregation caused non-uniformity in dopant concentrations; (4) a slow growth rate (productivity) and a generally incontrollable interface shape (for stress management and defect minimization); (5) slow turn around between crystal-ingots; and (6) a significant loss of feedstock materials if a growth run fails.
One object of the present invention is to provide a semicontinuous Czochralski crystal growth from a shallow melt apparatus that overcomes the increase of microdefects during crystal growth.
Another object of the present invention is to provide a semicontinuous Czochralski method of crystal growth from a shallow melt apparatus that lessens stress management and minimizes defects in the crystal growth.
A further object of the present invention is to provide a semicontinuous Czochralski method of crystal growth from a shallow melt apparatus that hastens turn around between crystal ingots.
A yet further object of the present invention is to provide a semicontinuous Czochralski method of crystal growth from a shallow melt apparatus that saves feed stock materials.
Another object yet still of the present invention is to provide a semicontinuous Czochralski method of crystal growth from a shallow melt apparatus that offsets connecting holes for replenishing the melt.
In general, the CZ crystal growth methods of the invention are accomplished by the use of:
In the semicontinuous Czochralski crystal growth from a shallow melt of the invention, the essential elements of the crucible comprises a crucible with a deepened periphery to prevent snapping of a shallow melt, and has a double barrier to isolate heat in the feeding compartment from that in the growth compartment, and further contains an interface thermal gradient control.
The invention will be better understood and the objects and features thereof apparent by reference to the drawing figures, compared to a conventional CZ crystal growth process utilizing a different crucible construction.
In this connection, reference is now made to
On the other hand, and by contrast, semicontinuous CZ crystal growth from a shallow melt of the invention, as depicted in the apparatus of
One aspect of the invention discovery is that a crucible with a deepened periphery prevents snapping of a shallow melt.
A shallow melt is needed to reduce turbulent melt convection. Turbulence is related to the Grashof number:
Where g is the gravity, β is the thermal expansion coefficient of silicon melt, L is the melt height, ΔT is the vertical temperature difference in the melt, ν is the dynamic viscosity of silicon melt.
However a shallow melt tends to ball up. The melt/gas and melt/crucible interface have energy contents proportional to their areas. Therefore, the melt shrinks to minimize energy. The energy per unit area can also be regarded as a surface/interface tension, or force per unit length.
The total free energy of the melt may be defined as follows:
E=ΣσiAi+∫ρgdz+∫PdV(V=constant, P(z)=ρg(zo−z)+Po+γ(1/r1+1/r2)cos θ)
where σi is the surface energies (melt/crucible interface and free surface) per unit area, Ai is the area of the surfaces, ρ is the melt density, P is the internal pressure in the melt, z is the vertical coordinate, zo is vertical reference position, Po is a reference pressure, γ is the surface tension of the melt, r1 and r2 are the two nominal radii of the melt surface curvature, and θ is the angle between the tangent of the melt edge and the crucible wall on the melt side.
If the interface energy is larger than melt surface energy (i.e., non-wetting crucible, θ>90°, the melt will tend to ball up until the surface/interface area and radius reductions are balanced balanced by gravity potential gains, which will lead to separation of the shallow melt from the side walls of the crucible. This causes an unstable melt, and the melt motion NM is depicted in the direction of the arrows in
If both melt surface energy and interface energy are very high compared to the gravity potential energy, the melt may retreat from one side and travel to the other side of the crucible, causing undesired asymmetry in the melt shape and temperature distribution. Both situatuations are called melt snapping. The asymmetric melt is depicted in
By the use of a crucible design that has a deepened periphery, snapping of melt is prevented even though a shallow melt is still formed in the crystal growth region, as illustrated by the melt stabilized in
From
A double barrier is used to isolate heat in the feeding compartment from that in the growth compartment, and as a result of this configuration, the over heated feed compartment functions to melt solid feeding materials and minimizes thermal impact to the growing crystal.
The heat transfer is comprised of:
Conduction∝−−kΔ−T
Radiation∝−−σ(ε1T14−εT24) and
Melt flow∝−Cpρ(T1−T2)V/L
Where k is the thermal conductivity, T is temperature, σ is the Stefan-Boltzmann constant, ε is the emisivity, Cp is the heat capacity of silicon melt, ρ is the silicon melt density, V is the amount of silicon melt being transferred, and L is the length of melt travel.
The double barrier (with offsetting connecting holes) crucible serves to: Reduce ∇T;
Increase L to minimize convective transfer and to eliminate the short path of thermally conductive melt; and
Reduce radiative heat transfer with opaque barriers.
In a primary embodiment of the invention, a round double barrier crucible is utilized for semiconductor ingot growth. A top view of a round double barrier crucible is depicted in
In another embodiment of the invention, a round double barrier crucible is utilized for semiconductor ingot growth. A top view of a round double barrier crucible is depicted in
In yet another embodiment of the invention, a rectangular double-barrier crucible is utilized for sheet or ribbon-shaped crystal growth. The rectangular double-barrier crucible is shown in
Finally, the double barrier crucible of the invention utilizes an interface thermal gradient control in that the shallow melt in the growth compartment operates as a thermal gradient control and allows for a much better thermal gradient control near the interface. By using an interface heater or a heat sink under the growth compartment as shown in
In a simplified one dimensional case, the crystal growth rate v is determined by the vertical thermal gradients in the crystal and in the melt near the interface:
Ks(dT/dz)s−K1(dT/dz)1=LatentHeat*v/Area
Therefore, an increase in the crystal growth rate may be obtained with the combination of a heat shield employed on top of the feeding compartment and an interface heat sink or heater under the growth compartment. The shallow melt allows close proximity between this interface heat sink/heat source and the solid/liquid interface, thus affording easier control of heat transport.
The United States Government has rights in this invention pursuant to Contract No. DE-AC36-99GO-10337 between the United States Department of Energy and the Midwest Research Institute.
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/US01/46211 | 11/1/2001 | WO | 00 | 4/29/2004 |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO03/038161 | 5/8/2003 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
4203951 | Goriletsky et al. | May 1980 | A |
5034200 | Yamashita et al. | Jul 1991 | A |
5126114 | Kamio et al. | Jun 1992 | A |
5139750 | Shima et al. | Aug 1992 | A |
5143704 | Nakaham et al. | Sep 1992 | A |
5312600 | Kamio et al. | May 1994 | A |
5314667 | Lim et al. | May 1994 | A |
5720809 | Iino et al. | Feb 1998 | A |
5871581 | Atami et al. | Feb 1999 | A |
5891245 | Atami et al. | Apr 1999 | A |
Number | Date | Country | |
---|---|---|---|
20040200408 A1 | Oct 2004 | US |