Impact of Environmental Air Pressures on Ion Implant Particle Performance

Numerous methodologies are available to evaluate particle performance for batch implanters. In our case, a single control wafer used to measure sheet resistance for varying dopant species is also monitored for particle adders using the KLA SP1 metrology tool with a frequency of three times per week. In addition, a full batch of 17 wafers implanted with argon is similarly monitored with the same frequency on alternating days. The benefits of using argon include reduced species cross-contamination and lower costs because of gas usage. This dual testing strategy allows us to capture multiple sources of variation, including day-to-day, species-to-species, and slot-to-slot variation. The resulting data allows for detailed analysis of most, if not all, relevant factors involved in a standard process.

Root cause analysis for highly intermittent particle failures within batch implanters can prove difficult. One approach is to use the "divide and conquer" method by first determining whether the particles are added during the actual implant process or during wafer handling. Establishing a baseline for beamless implants is needed for this initial phase of troubleshooting.¹ For beamless failures, the next stage is to test or partition each component of the wafer handler, such as robot, aligner, elevator, etc., until the source of particles is identified. In addition, particle wafer maps can be compared to wafer contact points for each handler component. Repeatable failures will often yield wafer map patterns or "signatures" associated with the root cause.

One such signature observed on a low-energy batch implanter (200 mm) is shown in Figure 1. To highlight the signature, we show a composite map of separate failures with the same edge pattern. Each failure includes all 17 wafers from one batch implant for a total of 68 wafers included in the composite. This edge pattern signature is not observed on all individual wafers, as may be inferred from the column chart of Figure 1 . The pattern typically begins on wafer slot #4 and gradually increases through wafer slot #17, which has the highest number of particle adders. Those fabs that only check one wafer (i.e., slot #1) to determine particle performance may not be aware this type of problem exists.

1. The 68 wafer composite particle “adders” map (above) using a KLA SP1 metrology tool shows wafer edge pattern. At right is the slot number breakdown for the same data shown on the left.

2. The position of the wafer in the dedicated cassette is transferred from a mini-environment to the loadlock.

This particular pattern was extremely challenging to troubleshoot because it could not be duplicated while performing numerous partition tests. In addition, the pattern did not appear to correlate to any contact points within the handler system, although it was observed on a beamless check. As an alternative approach, white vinyl tape was applied to a bare silicon dummy wafer in the shape of the edge pattern. This wafer was then manually loaded, and digital images were taken at each handling position as the wafer passed through the implanter. The most revealing picture is shown in Figure 2 , which depicts the position of the wafer in the dedicated cassette as it is transferred from the mini-environment to the loadlock. The front and rear wafer combs on one side of the dedicated cassette appear to cast a shadow coinciding with the taped pattern on the particle count wafer, as if the contamination source was behind that side of the cassette. Interpretation of the shadow resulted in two theories. Either airborne particles were deposited during wait time in the mini-environment or molecular flow particles were added during the pump downcycle of the loadlock.

3. Negative differential air pressure between the mini-environment and the bay resulted in reverse airflows through the load pod, as evidenced by the cleanroom wipe.

Subsequent testing using deionized water mist revealed an airflow problem from the cleanroom bay into the mini-environment. Negative differential air pressure between the mini-environment and the bay created reverse airflows through the customer cassette load pods, as evidenced by the suspended cleanroom wipe shown in Figure 3 . The previous partition tests had not duplicated the edge pattern because the test wafers were not allowed to wait in the mini-environment for up to two hours, as the standard particle test wafers do when they are loaded behind long-running, low-energy, high-dose implants.

Typical mini-environments include variable-speed fans, which pull air from the surrounding cleanroom bay or chase through multiple HEPA pre-filters before delivering laminar airflow to the wafer handling area. The filtered air then exits back into the chase from vents located at the base of the implanter. Balancing the mini-environment air involves adjusting the fan speeds to achieve a target air velocity and the vent openings to achieve a target differential pressure, as specified by the implanter OEM. But even the best-designed mini-environments depend on the cleanroom recirculation air handler units for proper pressurization.

Air pressure differentials are required to control the flow of particles between areas and minimize cross-contamination within the cleanroom. For a microelectronics fab, the cleanest areas are the operator bays where wafers may be exposed during the load/unload process to and from the mini-environment. Differences in pressure between these two areas typically range from 1.25 to 5 Pa for well-designed cleanrooms.²

Because of unexpected space constraints, the above referenced implanter was installed in a location that was not originally designed for optimal air pressure differentials. Five separate recirculation air handler (RAH) units supply air that flows through the operator bay, service chase, and main corridor surrounding implanter #1, as shown in Figure 4 . The return ducts for each RAH are not shown, but are located in the service chases. In addition, return air vents are located on the wall separating the chase from the main corridor. Despite this non-standard configuration, the particle performance of the implanter was within specification for an extended period of time, and the edge pattern signature was only observed after two years of operation.

4. Five RAH units supply air that flows through the implanter shown above. Return ducts are located in the service chases.

Once the root cause of the particle problem was identified, the relative air pressures were measured (Table). In agreement with the first observation, the mini-environment proved to be at negative pressure with respect to the operator bay. Also problematic was the fact that the operator bay was at negative pressure with respect to the heavily traveled main corridor. Therefore, contaminants could easily migrate from the corridor to the operator bay through the bottom of the load pods and into the mini-environment before contaminating the wafers. Correcting the bay to corridor pressure differential was accomplished by first removing maintenance materials (tables, toolboxes, parts, etc.), which were partially obstructing the return air vents and subsequently reducing power to the corridor RAH units. Altering the mini-environment to operator bay partial pressure would also require multiple corrective actions. Three of the four fans within the mini-environment were defective and needed replacement. In addition, the power to RAH unit #63 was reduced while the power to RAH unit #64 was increased in order to maintain the required face velocity of 75 fpm across the ULPA filters connected to the common pressurized plenum system for this bay. Post-adjustment pressures are listed as "After" pressures in the Table. More importantly, the edge pattern was never observed again, and the particle baseline for this implanter trended down, eventually reaching acceptable levels after approximately three months as shown in Figure 5 . The failure rate decreased by 20%.

5. This chart indicates the downward trend for the implanter after the rebalancing of the air pressures.

Positive air pressure between the mini-environment and surrounding cleanroom areas should not be taken for granted after tool installation. Subsequent changes in cleanroom configuration and equipment location can alter pressure differentials, which should be continuously monitored with real-time sensors. Mini-environment fans can fail without alarm and should be checked on a periodic basis. We believe these subtle changes had combined to create unacceptable particle performance, as evidenced by the worsening edge pattern and baseline.