01 June 2003
Corrosive gas kills product
By Christopher Muller and Brad Stanley
Reactivity monitoring proves environment and tracks episodes.
New semiconductor technologies require much more stringent control of the manufacturing environment. From an air quality standpoint, the control of chemical contamination has become as important as the control of particulate contamination for many processes.
Levels of airborne molecular contaminants (AMC) in ambient air are high enough to be problematic if allowed in the facility. However, the types and numbers of chemical species used in manufacturing semiconductor devices also mean that there is a greater chance of chemical contamination from internal sources.
Chemical contamination control strategies are complex to design and implement because it is difficult to accurately characterize the environment's potential to cause damage to materials and processes.
Further, once one has determined that direct chemical contamination control is required, determining how well, or even if, the implemented control strategies are working is an arduous task.
Reactivity monitoring as ISA Standard S71.04 describes it has been in use for many years to provide accurate environmental assessments for the process industries. Many of the same contaminants of concern to these industries are also of concern to the semiconductor industry.
Thus, semiconductor manufacturers accept reactivity monitoring as a viable environmental monitoring method. An air quality classification scheme based on reactivity monitoring exists and has gained wide acceptance throughout the semiconductor industry.
Reactivity monitors detect the interaction of contaminant gases on copper, silver, and/or other metals by measuring corrosion film thickness. This monitoring technique provides information on many contaminants problematic to manufacturing processes.
It also directly indicates the potential for AMC-related process effects. This technology has proven especially useful for establishing AMC baselines and identifying contamination episodes and their sources.
Facilities can incorporate the technology into their overall preventive maintenance program, and it can help reduce the number of AMC-related incidents by allowing proactive investigation of potential problems.
Performance monitoring of chemical filter systems allows for timely change out of media and filters and reduces the cost of ownership for these systems. Proper maintenance of the systems also ensures air quality goal adherence.
With any AMC control strategy, one must continuous monitor the environment to ensure compliance with air quality requirements. Reactivity monitors are capable of providing information on AMC levels as low as one part per billion (ppb).
A network of monitors that communicates directly with the facility management system can also track the distribution and migration of chemical contaminants in the facility.
Here is a discussion on reactivity monitoring and new advances in monitoring technology.
CONTAMINANT HOT SPOTS
Before one can determine whether airborne molecular contamination control is warranted, the environment's potential for AMC-induced damage or failures must be characterized.
This can happen with real-time gas monitors, although this method does not necessarily indicate the potential effects of these contaminants.
Unless one has characterized a specific contaminant's effect(s) on materials and processes, one can only speculate as to what its presence will mean to product reliability and yield.
Even for a well-characterized contaminant, such as ammonia in lithography processes or chlorine in metallization and salicidation, its control alone does not ensure that the process or device will be defect-free.
Direct gas monitoring does not take into account the possible synergies—positive and negative—between different chemical species.
Reactivity monitoring, however, can show the effects of AMC on materials, and semiconductor fabrication facilities—fabs—use it to perform air quality assessments.
Environmental classifications using reactivity monitoring characterize the destructive potential of an environment. Industry accepts the classifications as a viable alternative to direct monitoring of low-level gaseous contaminants.
Reactivity monitoring now serves to characterize makeup air, identify AMC hot spots, and track AMC incidents within a fab with the ultimate goal of establishing the cause-and-effect relationship between AMC and product yields and losses.
Environmental reactivity coupons (ERCs) now work evaluating long-term air quality trends inside and outside facilities. They perform environmental surveys in order to develop AMC control programs.
They are also useful for differentiating between classes of chemical contaminants and providing estimates of AMC concentrations.
The main limitation of ERCs is their inability to provide a continuous environmental classification. To address this, reactivity monitoring has taken a step further by developing real-time monitoring that employs metal-plated quartz crystal microbalances (QCMs).
These microprocessor-controlled devices can measure the total environmental corrosion attributable to AMC and detect changes in AMC at levels less than 1 ppb. This is probably one of the main requirements for any real-time AMC monitoring protocol used in semiconductor manufacturing applications.
AIRBORNE MOLECULE MONITOR
Air monitoring is central to any environmental control program for achieving and maintaining air quality standards based on the presence—or absence—of gaseous air contaminants.
Such monitoring can also provide the short-term data required to manage and mitigate contaminant-specific episodes.
In addition to applying directly to contamination control programs, one may employ air monitoring data for (1) the evaluation of long-term air quality trends in a facility, (2) research studies designed to determine relationships, if any, between contaminant levels and possible damaging effects, and (3) verification of AMC control equipment performance.
Several characteristics of any measurement technique must submit to evaluation to determine the technique's appropriateness for air quality monitoring. Among the more important characteristics are sensitivity, cost, and complexity.
Sensitivity is a particularly demanding parameter for the semiconductor industry, where near-ambient levels of many pollutants occur and control levels approach the parts-per-trillion level.
Likewise, cost may be quite important when deciding on a measurement technique, particularly in large surveys.
A final point of consideration is the complexity of the technique and the degree of skill and training required to obtain quality results. Other factors deserving attention are selectivity and portability.
Even though it is possible to identify and quantify nearly all the chemical species one may encounter in a semiconductor manufacturing facility, the question still remains "What do I do with this information?"
To date, there has been little-to-no data cited or published that provides definitive information of the cause-and-effect relationship between specific levels of gaseous contaminants and the damage they may cause to semiconductor devices and within clean rooms.
Because of this, a number of manufacturers are turning to environmental classification via reactivity (or corrosion) monitoring. The validity for this air monitoring technique lies in the fact many of the contaminants targeted for control are corrosive in nature and, therefore, can be easily monitored through reactivity monitoring.
Coupons showing corrosion.
REACTIVITY COUPONS TESTIFY
Although originally developed for the classification of environments for computers and control rooms, one can use ERCs to indicate the presence of chlorine and fluorine compounds, ammonia, and other corrosive materials including ozone, sulfur dioxide, nitrogen dioxide, hydrogen sulfide.
All of these materials cause a number of AMC-related process effects.
ERCs normally contain copper or copper in combination with other metals to assess an environmental. Studies have shown, however, that although copper coupons are good indicators of corrosive gases in an environment, they are not sufficiently sensitive to many of the contaminants ubiquitous to most semiconductor facilities.
Therefore, the use of ERCs for environmental classifications should include both copper and silver coupons.
ERCs are passive monitors typically exposed to the environment for a period of 30–90 days. Through analysis, the metal coupons provide a realistic assessment of average environmental conditions over time.
ERCs usually serve to evaluate long-term air quality trends inside and outside a facility and to perform environmental surveys as part of the development of an AMC control program. They can also differentiate between classes of chemical contaminants and provide estimates of the airborne AMC concentrations.
One consideration faced in designing an air quality-monitoring program is the choice of passive or active sampling. The relatively immediate feedback of an active monitor is a very desirable aspect, which often precludes the use of passive monitors.
As mentioned previously, the main limitation of ERCs is their inability to provide a continuous environmental classification. To address this, reactivity monitoring has expanded through the development of a real-time monitoring device employing metal-plated QCMs.
These microprocessor-controlled devices measure the total environmental corrosion attributable to AMC. Environmental reactivity monitors (ERMs) employing QCMs can detect and record changes in the levels of AMC in the ambient environment at levels less than 1 ppb.
This capability is one of several main requirements for any real-time AMC monitoring protocol used in semiconductor manufacturing applications.
To date, there is only one commercially available ERM employing copper, silver, and gold-plated QCMs that provides real-time information about the amount of corrosion occurring due to the presence of AMC, that conforms to industry standards, and that is currently operational in a fab.
The device monitors corrosion on a continuous basis and enables preventive action before AMC causes serious damage.
This device wires directly into a central computer system. From there, data can log to historians and certain triggers can be initialized. By using the unit's ability to interface with control systems, the device makes up-to-the-minute information on the levels of corrosive contaminants available. Users can establish and maintain environmental classification databases to provide historical data.
DEFINE SYNERGISTIC EFFECTS
AMC control criteria are often based upon groups of chemical contaminants as opposed to individual chemical species. This is not surprising, as many of these criteria developed from an industry-wide perspective.
However, individual semiconductor manufacturers, due to competitive pressures, tend to use their own contamination control criteria, which they developed through their own experience, capabilities, and expectations.
There is no general consensus on the acceptable levels of airborne chemical contamination, but the ranges have been narrowing over the last several years and this trend will probably continue in the near future.
Two standards for control of AMCs follow. One is an example of the standard in the industry today, SEMI Standard F21-95. The other is an example of the standard other industries use to relate a classification to equipment reliability, ISA Standard S71.04.
The purpose of the SEMI Standard titled Classification of Airborne Molecular Con-taminant Levels in Clean Environments is to classify microelectronic clean environments with respect to their AMC levels.
These standard classifications are used in the specification of semiconductor clean environments (including process tool environments) and of contamination control and measurement equipment performance.
This standard determines environmental classifications by the maximum allowable gas-phase concentration of four specific contaminant categories: acids, bases, condensables, and dopants.
The combination of a quantitative class for each of the four material categories yields a classification describing the environment.
These environmental classifications are very useful for describing the levels of contaminants in the cleaned space, but there is no correlation to equipment reliability, product reliability, and/or process yields. There is also the complexity of interpreting what the synergistic effects of those contaminants will be.
ISA COVERED AIRBORNE
Industry well knows that electronic and electrical equipment failure is possible due to the corrosive effects of certain contaminant gases. ISA—The Instrumentation, Systems, and Automation Society developed a classification system for the protection of electronic equipment employing reactivity (or corrosion) monitoring.
S71.04-1985 Environmental Conditions for Process Measurement and Control Systems: Airborne Contaminants covers airborne contaminants that affect electrical/electronic equipment and establishes airborne contaminant classes.
The standard establishes environmental classifications according to the type of contaminant. Within each classification, severity levels are also described.
The synergistic effects of all contaminants present manifest as the total angstroms of corrosion on the reactivity coupons. This count in turn relates to reliability of the electronic equipment in that space.
These reliability statements do not directly relate to semiconductor processing, and semiconductor facilities desire to operate at much lower levels than are shown in this standard.
However, this shows how reactivity monitoring can determine the aggressiveness of an environment and allow the prediction of process details and product quality from environmental classifications.
This has already happened to some extent.
INDICATE TARGET LEVELS
While the SEMI and ISA standards provide classification schemes for airborne molecular contaminants, they do not correlate cleanliness classes to semiconductor product yields or reliability.
The Semiconductor Industry Association's (SIA's) "National Technology Roadmap for Semiconductors," and SEMATECH's Tech-nology Transfer 95052812A-TR present recommendations that can be used to develop gas-phase contaminant specifications.
The current SIA roadmap states there is now consensus that, as device geometries approach 0.18 micrometers and beyond, the percentage of process steps affected by nonparticulate or molecular contamination will increase.
Because of this trend, environment control technology requirements indicate target levels of ambient acids, bases, condensables, and dopants for specific process steps.
This environmental analysis method is currently working for a number of semiconductor manufacturers and has played in the formal literature. Reactivity monitoring classification schemes are pending that directly correlate corrosion rates to environmental classifications. They are currently being refined based on the results of testing and the specific needs of the microelectronics industry.
Generally speaking, the copper and silver corrosion rates should be class C2/S2 or better unless otherwise agreed upon.
The individual corrosion films, which ERCs can quantify, also may serve to characterize the environment and to determine the proper AMC control strategies.
Based upon the recommended control levels and test results from laboratory and field-exposed ERCs, acceptance criteria relevant to these applications have been determined. These criteria take into account total corrosion as well as the relative contribution of each individual corrosion film.
Where companies employ chemical filtration to maintain the interior concentrations of gaseous pollutants as low as possible, they have routinely attained total corrosion rates less than 15–20 Angstrom units per 30 days. Subsequent gas monitoring indicates pollutant levels are at or below the limits of detection for the analytical techniques employed.
This "no detectable pollutants" scenario is being used to set up environmental classification systems based on reactivity monitoring. Consensus opinion is that if an environment exhibits corrosion rates corresponding to a C1 or S1 classification, there is nothing within economic bounds that can improve the environment.
Typical uses of reactivity monitoring with ERCs and ERMs have been for the characterization of outdoor air used for ventilation, the identification of "hot spots" within a manufacturing facility, and the effectiveness of various preventive measures.
Companies also use reactivity monitoring to track AMC "episodes" in semiconductor fabs for the purpose of developing the cause-and-effect relationship between AMC and product yields/losses.
InTech refers the reader to the charts and tables on the ISA Web site at www.isa.org/intech/semiconductors. See the numbers and graphs starting on page 4.
These graphics come from semiconductor manufacturers that have developed AMC assessment and monitoring programs centered around the use of environmental reactivity coupons (ERCs) and monitors (ERMs). The data is from applications illustrating the varied use of, and experience with, this monitoring technology. IT
Behind the byline
Christopher Muller is an expert on airborne molecular contamination control in clean room applications and has published numerous papers, articles, and handbooks on the topic. He is the manager of technical services at Purafil. In addition, he is a member of ISA's SP71 committee on Environmental Conditions for Process Measurement and Control Systems, a senior member of the Institute of Environmental Sciences, chairperson of ASHRAE's Standard Project Committee 145P, and is involved with SEMATECH's International Technology Roadmap for Semiconductors. Brad Stanley has a B.S. in chemical engineering and works for Purafil, Inc., where he performs research and development pertaining to the manufacture, testing methods, and application of Purafil's gas-phase filtration media. He actively authors and presents papers for the subjects of odor control, emergency gas scrubbers, electronic equipment corrosion, and reactivity monitoring.