Heating A Wafer – How Difficult Can It Be?
Among the many processes involved in fabricating integrated circuits, it is deposition and dry etch which levy some of the most stringent requirements on the temperature uniformity of wafer heaters. Both chemical vapor deposition and planar plasma etch often stand out as the most demanding applications of “superuniformity" solutions to the wafer heating problem, reference Figure 1. Common to such processes is the need for a pedestal heater which supports and transfers thermal energy to a target substrate while maintaining surface temperature and flatness profiles within acceptable limits over the life of the equipment.
This article presents a case study examining the development of a rather unconventional pedestal heater to address an unusual confluence of thermal and structural requirements which proved to be highly coupled and conflicted. The problem presented here was the modification of a legacy pedestal heater design for a new application with an increased process heat load. The legacy heater configuration was characterized by a concentric, dual control zone layout of spiraled heating elements embedded in an aluminum casting. This product was supported on a bi-metallic shaft as depicted in the left half of Figure 2. In the new application, however, the heat transmission performance of the shaft became detrimental to temperature uniformity. Because the stainless steel end of the shaft was so effective at impeding heat conduction to the chamber floor, the excess process heat load manifested itself as an unacceptable hot spot at the center of the heater. The right half of Figure 2 depicts the attempted workaround. This design consisted of eliminating the stainless steel spool piece, extending the aluminum end of the shaft down to the chamber floor and actively cooling the chamber floor so the vacuum seal temperature would remain within safe bounds. Unfortunately, the net heat sinking effect of these changes caused an unacceptable cold spot.
The following considerations framed the issue set for reoptimizing the baseline, all-aluminum design: 1. Accommodation of multiple operating modes. The excess process heat load scenarios ranged from 0 to approximately 1000 watts, depending on which condition was in play. For the maximum heat load condition, maintaining temperature uniformity proved to be highly problematic with a center-mounted, all-aluminum shaft. To improve uniformity, the radial temperature distribution would have to be “re-balanced" by moving the shaft mounting location closer to midway between the pedestal center and its outside diameter.
2. Thermal stress. For shafts which attach to the heater near the radial midpoint, there was concern about thermal stresses being high at the shaft weld joint locations because of a macroscopic thermal strain effect known as coning, whereby the warm end of the shaft expands more than the cool end.
3. Constraints. In addition to uniformity remediation and thermal stress mitigation, the design improvement effort would have to account for numerous constraints. These constraints included the bending limitations of sheathed heating elements, nichrome (NiCr) wire watt density limitations, shaft manufacturability, heater “castability," welding limitations, construction material limitations, and a 50°C (122°F) design margin for the o-ring sealing the shaft bottom flange.
The leading design options which emerged from concept ideation are illustrated in Figure 3. The inverted cone (Option A) was preferred from a relative ease-of manufacturing perspective.
However, due to lingering concerns about thermal stress, the radial bellows alternative known as a reentrant post (Option B) was developed in parallel as a fallback. To advance a suitable design into production prototyping, Watlow began by detailing 20 solid model versions spanning the aforementioned two geometries. These versions were funneled down to a final embodiment through the two-stage, multi-physics finite element analysis (FEA) process depicted in Figure 4.
This process was comprehensive in scope, accounting for all heat conduction paths subjected to the process heat generation and radiation and convection boundary conditions established in the chamber, as well as the coupled inelastic thermal strains and associated stresses in the entire pedestal heater subassembly.
As a result of numerous FEA simulations, one of the originally-conceptualized versions of the inverted cone geometry ultimately emerged as the preferred shaft redesign alternative for improving the pedestal heater’s uniformity.
The illustrative thermal FEA result case presented in Figure 5 suggests the theoretical feasibility of a one percent surface temperature uniformity goal for a hypothetical process scenario. The inelastic thermo-mechanical FEA results verified the reentrant post was superior in terms of thermal performance and long-term structural robustness.
However, the inverted cone provided the best overall combination of temperature uniformity, stress margins and manufacturability (lathe operation only versus lathe and mill or weld for the reentrant post).
Although the results also indicated the top weld joint might undergo significant creep deformation as the temperature exposure time approached 10,000 hours, the 70 percent higher yield stress limit at the 100 hour mark was large enough for the inverted cone to retain adequate elasticity during thermal cycling intervals more representative of actual use. Heating a wafer to achieve a surface temperature “super-uniformity" approaching one percent of the application set point temperature is not trivial from the heater design perspective.
Comprehensive, FEA informed trade studies accounting for coupled physics effects, design and manufacturing rules, as well as material limits are beneficial to achieving aggressive performance goals, particularly when legacy components are being considered for retrofitting into next generation systems.