Study of Repeatability, Relative Accuracy and Lifetime of Thermocouple Instrumented Calibration Wafers for RTP
- PDF / 579,801 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 98 Downloads / 229 Views
C
Siwafer
bond area (cement)
TC
/Siwaer/
bond area (cement)
/
Fig. 1.1: In the new 1530 structure, each lead makes a separate 180 degree rotation around the edge and the leads are welded at the opposite side in an underdercut area, close to the Si.
F 1.2: In the 1501 structure, the thermocouple is mounted in the center of a reentrant cavity, filled with alumina based cement (the bond area). 17
Mat. Res. Soc. Symp. Proc. Vol. 470 01997 Materials Research Society
The fundamental sources of error are: (1) the error on the thermoelectric coefficient of the thermocouple wire and (2) a temperature difference between the thermocouple junction and the undisturbed silicon. The first error comes from normal variations of TC wire within the specification and from drift due to oxidation or other aging effects. However, within a manufacturing lot of TC wire, a high reproducibility from TC to TC is typical (< 0.1 "C). Also, a precision calibration of each TC wire lot is available. The second error source is typical in RTP systems, since there is no temperature equilibrium in the oven. In particular, the TC wire outside the bond area is cooler than the Si wafer. This causes a loss of heat along the TC wires and a junction temperature slightly below the Si temperature. From analysis it was found that the longer embedding of the TC wires in the cement in the 1530 structure, causes a large reduction of the heat loss from the TC junction to the ambient. This explains the better results found with 1530 structures. EXPERIMENTAL CONDITIONS A commercial oven for 6" wafers (AGA Heatpulse 610) was modified with precision lamp voltage control and controlled cooling water and cooling air (see [1] for more details). The wafers (see table I) were p- 6" Si-wafers, in which 6 TC's were mounted near the center of the wafer on a circle with a radius of 8 nmn. These 6 TC's are named TC1 to TC6. On wafer #9, 3 TC's of
type 1501 were combined with 3 TC's of type 1530 in alternating order on the 8 mm radius circle. On the other wafers, 6 TC's of type 1530 were mounted on the 8 mm radius circle. The thermocouple signals were measured and logged with the Thermal MAP 2 system from SensArray. The ambient was N 2 (with a purity of 6 nines from a liquid nitrogen container) flowing at 4 slpm. Before the first heating cycle, the chamber was purged with 10 slpm N2 for 2 minutes. Wafers #9 and #12 were processed with nitrogen from cylinders of a lower purity. This caused oxidation of the K-type wire. This wire oxidation on wafers #9 and #12 caused the lower stability upon loading/unloading. The wafers were heated under open loop, lamp power control. The heating processes (P1050 and P1150) yielded peak temperatures of nominally 1050 "C and 1150 "C. The lamps were on during 30 seconds with a power profile that yielded a steady state time of 20 seconds. Between two consecutive heating cycles, there was a cool down period of 330 seconds (cycles were repeated every 6 minutes). The maximum temperature at the end of each heating cycle was measured and is rep
Data Loading...
