Cost-of-Ownership Survey
Staff -- Semiconductor International, 7/1/1998
Sidebar to "Understanding Equipment Cost of Ownership"
 |
| Fig. 1. The survey respondents are involved with a wide range of semiconductor machinery. |
COO, a concept promoted by SEMATECH and other industry experts to evaluate equipment purchases and prioritize engineering design projects, goes beyond price in analyzing machine performance by incorporating process technologies, information flow, materials consumption, power quality and other variables that ultimately affect throughput and yield.
The survey was underwritten by the Square D Co., a manufacturer of electrical and control products. Questions about the use of COO and the factors analyzed were asked of 2000 semiconductor manufacturers and equipment makers by Bull's Eye Marketing Inc., an independent research firm. More than 80% of the respondents were engineers, project or design managers involved with a wide variety of equipment types.
Nearly half the respondents (49.5%) were equipment makers, while 46.4% were semiconductor producers involved in communications (29.7%), MPU/processors (23.8%) and memory (12.9%) (Table 1, Fig. 1).
| Table 1. How Important is COO to Your Company? | ||||||
| Very important | Somewhat important | Not very important | Not at all important | No opinion | No answer | |
| Semiconductor producer | 51.1% | 42.2% | 2.2% | 2.2% | 2.2% | O |
| Equipment maker | 54.2% | 31.3% | 8.3% | 2.1% | 2.1% | 8.1% |
More than 90% of semiconductor producers think COO is very or somewhat important to their company. Equipment makers put great stock in COO as well: 54% say it is very important, and another 31% believe it is somewhat important. And nearly 75% say their use of COO analyses will increase over the next 12 months.
Key process variable: Facility power quality
One of the most significant findings of the survey was the impact of poor facility power quality on production yields and costs. Power quality is expected to become a major issue for the industry in the United States with the growing deregulation of the utility industry.
Power problems can be reduced throughput and yield and increased equipment maintenance, materials usage and production costs. More than half the respondents said a "brown out" (voltage sags of 10%) would negatively affect production. Power voltage spikes, surges and noise were also seen as disrupting production and increasing costs (Tables 2,3).
| Table 2. What Unfavorable Effect Would a "Brown-Out" Have on Production? | |||||
| Major effect (%) | Moderate effect (%) | Little effect (%) | No effect (%) | No answer (%) | |
| Throughput of product | 33.7 | 33.7 | 18.8 | 7.9 | 5.9 |
| Yield of product | 36.6 | 22.8 | 17.8 | 13.9 | 8.9 |
| Maintenance of equipment | 33.7 | 27.7 | 24.8 | 5.0 | 8.9 |
| Consumables used in process | 17.8 | 25.7 | 30.7 | 17.8 | 7.9 |
| Cost of production | 38.6 | 28.7 | 19.8 | 5.0 | 7.9 |
| Table 3. What Unfavorable Effect Would Power Voltage Spikes, Surges or Noise Have on Production? | |||||
| Major effect (%) | Moderate effect (%) | Little effect (%) | No effect (%) | No answer | |
| Throughput of product | 30.7 | 32.7 | 20.8 | 7.9 | 7.9 |
| Yield of product | 31.7 | 31.7 | 16.8 | 10.9 | 8.9 |
| Maintenance of equipment | 34.7 | 23.8 | 27.7 | 5.9 | 7.9 |
| Consumables used in process | 18.8 | 20.8 | 35.6 | 16.8 | 7.9 |
| Cost of production | 27.7 | 36.6 | 17.8 | 5.9 | 11.9 |
An understanding of the potential impact of facility power quality on production was nearly universal among both semiconductor producers and equipment makers. Yet surprisingly few companies use sophisticated power monitoring and switching systems to actively manage tool performance.
Just 21.4% of the respondents used power quality as a factor in analyzing COO. More than half of those surveyed were uncertain whether having separate systems to manage questionable facility power quality affected the measure of equipment performance (Table 4).
| Table 4. Which do you Consider the Most Effective Way of Coping With Power Quality Problems? | |
| UPS (uninterruptible power supply) | 57.3% |
| Surge suppressors1 | 3.5% |
| Power monitoring/switching equipment | 14.6% |
| Other | 7.9% |
| Power factor correction | 6.7% |
A majority, 57.3%, said they relied on uninterruptible power supply (UPS) equipment to cope with power problems, at best a Band-Aid approach that does not address the root causes. Many expected the utility companies or on-site power stations to control power quality, even though power problems may also be caused by conditions within the plant.
This appears to be a risky strategy for OEMs at a time when a growing number of semiconductor device makers are using performance contracting in purchasing equipment. Monitoring power quality at the tool level would allow an OEM to enter into a performance contract that, among other criteria, would be contingent on the quality of power feeding the tool. Power disturbances that fall outside the guidelines could be time stamped and recorded. For OEMs, the benefit is that performance would be measured only on those conditions outlined in the contract. Fabs, in turn, would gain valuable information linking electrical disturbances directly to throughput, yield and downtime.
Better machine diagnostics needed
Not surprisingly, more than 35% of survey respondents consider the diagnostic capabilities of a piece of equipment an important factor in analyzing COO. Improved diagnostics are a necessary component to continued increases in yield and lower costs.
When asked what type of communication networks they relied on to diagnose equipment and process performance, respondents pointed to sensor buses for I/O, monitoring and reporting, and enterprisewide networks for yield data, machine utilization and productivity (Table 5).
| Table 5. What Improvements in Networks are Most Needed? | |||
| Enterprise (%) | Sensor bus (%) | Neither (%) | |
| I/O type diagnostics | 13.6 | 23.9 | 5.7 |
| Yield data | 18.2 | 12.5 | 5.7 |
| Machine utilization and productivity | 25.0 | 8.0 | 5.7 |
| Synchronization of manufacturing | 23.9 | 4.5 | 6.8 |
| Monitoring & reporting | 22.7 | 14.8 | 6.8 |
| Flexible production | 20.5 | 6.8 | 5.7 |
Given the increased complexity of the fabrication process, respondents said better information gathering and coordination tools will be needed for optimal analysis in the future.
 |
| Fig. 2. COO analysis is expected to see a 74% increase among survey respondents. |
Future use of COO
Although nearly 82% of the respondents currently use COO analyses to evaluate equipment performance, measurement processes vary widely -- from the standardized model developed by SEMATECH to a variety of "homegrown methods" and software. A number of semiconductor producers rely on systems developed by their equipment makers.
Results of the survey indicate a growing interest in COO (Fig. 2), as well as a need for standardized analytical tools and educational programs for the industry.
