March 21, 2012 — A clean, contamination-free semiconductor wafer processing environment is critical to maximize yield. As wafer circuitry and geometries shrink, particles become more likely to cause defects and yield loss. Particles generated during wafer processing by equipment component failure, wafer mishandling, excessive vibration, or other process irregularity can all contribute to defects.
Reducing or eliminating particle contaminants is an important step in tool qualification and chip manufacturing monitoring. Semiconductor tool operators must identify particles inside fab tools, as well as wafer storage and transport environments, for optimum operating conditions.
As part of its ongoing particle monitoring procedures, a high-volume 200mm wafer fab used a standard wafer monitoring method to regularly spot check 10,000 cassettes and lot boxes used in its wafer manufacturing operations. Cassettes and lot boxes were pulled for cleaning every 150 cycles; tools also were monitored for particle residue. Random checks were performed when particle problems were suspected. Using particle monitor wafers to inspect cassettes and lock boxes required two technicians for set-up and execution: placing monitor wafers inside each unit for one hour to collect contaminates, running monitor wafers through a surface scanner, and manually recording results.
 |
| Figure 1a, L to R. Particle monitor wafers, benchtop counter, and handheld counter. While surface scanning is currently the industry standard for particle detection, bench top and hand held airborne particle counters are also used because of their relatively low cost and easy operations. However, these technologies are limited in reach inside equipment and offer little or no information about the location of contamination in the tool. |
 |
| Figure 1b. The airborne particle sensor (APS) wireless particle detection system. |
While accurate, the particle monitor wafers proved time-consuming, resource-intensive and inefficient for the large amount of cassettes and lot boxes that the fab re-qualifies on an ongoing basis. Monitoring each cassette or lot box with a monitor wafer took at least one hour; daily throughput averaged 8 cassettes or lot boxes; that required 16 technician hours. The monitor wafer process did not provide remote real-time feedback, so one operator was needed for tool operation and another for set up, running the surface scan, and recording results.
The 200mm fab initiated a Manufacturing Monitor Program to find an alternative that would reduce qualification time, improve throughput, and provide statistical sampling of production cassettes and lot boxes to meet particles standards.
Portable and benchtop particle counters (Figure 1) offer real-time feedback, but the instruments require either tearing down the fab tool or running a series of test wafers, costing the fab significant downtime and labor. Handheld particle counters are limited by hand reach and require opening the tool.
A wafer-like, wireless airborne particle sensor offered an alternative wafer monitoring methodology for atmospheric tools. It enabled the fab to perform the same manufacturing monitoring tasks in three minutes inside a cassette or lock box that took monitor wafers one hour. Combining the flexibility of monitor wafers and the real-time capability of benchtop counters, the airborne particle sensor (APS) goes deep inside a tool and remotely communicates real-time data to a laptop.
Initially, the 200 mm fab utilized the APS in two different test condition scenarios: cassettes new from washer vs. cassettes in production; and lot boxes new from washer vs. lot boxes in production.
 |
| Figure 2. APS test results when comparing newly washed cassette vs. older production cassette in new lot box. Results show that cassettes remaining in production too long have excessive particles. |
By comparing the particulates gathered on newly washed and production cassettes and lot boxes, the fab determined if process tools attracted more particles from cassettes/lot boxes that were in production for specific periods of time. Figure 2 shows a comparison particle count reading conducted by the APS in a production cassette and a newly washed cassette (Test time: 3 minutes). These results were consistent with output generated with monitor wafers over an hour of test time (see the table). Both metrology methods confirmed that cassettes remaining in production too long have excessive particle contamination.
 |
| Table. Comparison of time required to monitor cassettes and lot boxes using wafer monitors vs. the airborne particle sensor. |
Compared to the fab’s previous monitoring method, the wireless airborne particle sensor more quickly measured numerous combinations of potential particle source elements, such as older production/newly washed cassettes and older production/newly washed lot boxes to determine the presence of particle sources and counts. Since the airborne particle wafer provided real-time feedback, operators could take a 3-minute reading and move on to the next unit for testing. The APS required just one operator vs. at least two for the monitor wafers.
Based on results from its comprehensive qualification project, the fab monitoring team replaced monitor wafers with airborne particle sensors as part of its pre-qualification process in checking cassette and lot box for particles to determine suitability for production.
The combination of fast particle counting, reduced technician operation requirements, and real-time output allowed the fab to increase test throughput by 20x, at half the manpower requirements. After adopting the airborne particle sensor for its manufacturing monitoring process, the fab estimated a reduction of monthly time spent on particle monitoring from 240 hours to 10 hours.
Particle monitoring as a troubleshooting tool at 300mm fab
While the APS proved successfully in increasing throughput efficiency for the 200mm fab in measuring tool particulate contamination as part of its prequalification procedures, the sensor also served as an effective troubleshooting tool for a 300 mm fab experiencing a high wafer defect rate of unknown origin. To determine the source of contamination, the fab wanted an effective troubleshooting and prequalification tool prior to using monitor wafers for final vacuum tool particle qualification.
 |
| Figure 3. The APS discovers the source of particle contamination within a 300mm fab in real-time. |
Benchtop and handheld counters did not offer the reach to identify and isolate the source of particles in this fab scenario. Using monitor wafers during this stage of tool qualification would require segregation and running a separate monitor for each chamber, buffer, and load lock, a time-consuming option. The APS offered accessibility inside the process tools without opening the chambers, and provided real-time feedback on exactly when and where contaminants occurred.
The fab used the APS to scan the wafer path as it went through Chambers A and D, and performed additional spot checks of doors, handoffs, and robot motions. Placed into the tool generating wafer defects, the APS initially moved from the load-port, FI, load-lock and into the buffer. In these sections, the APS deemed particles at an acceptable average of 0.1 cumulative particles per second. When running through the Chamber A door, the APS immediately identified that the door was shedding particles and causing contamination problems (Figure 3). As a result, the fab replaced the Chamber A door slider assembly, which was not yet scheduled for preventative maintenance.
While the fab continues to use monitor waters in vacuum conditions for final qualification, the APS saves significant time in identifying particle contamination sources during tool pre-qualification stages.
Allyn Jackson, CyberOptics Semiconductor, can be reached via the company’s website at www.cyberopticssemi.com.