Three Analytical Techniques for Air Quality and Hazardous Gas Monitoring in Cleanrooms
With three analytical techniques based on proven technologies for the workplace, you can monitor worker safety for nearly all chemicals currently used or likely to be used in the cleanroom.
By William C. Scott
Chemicals used in the semiconductor fab that are identified as health hazards are typically monitored at levels set by the American Conference of Governmental Industrial Hygienists (ACGIH) at the threshold limit value (TLV) of a substance.  Chemicals that are potentially
flammable or explosive are often monitored in the range of their lower flammable limit (LFL) or lower explosive limit (LEL). For monitoring purposes, these chemicals are categorized by their TLVs and LFL\LELs:
TLV at part per billion (ppb) levels: poisonous (toxic) gases, such as metal hydrides, generally having TLVs of less than 1 ppm.
TLV at part per million (ppm) levels: acids, gases, solvents, and oxidizers with TLVs ranging from 1 ppm up to 1000 ppm.
100 percent of LFL\LEL may be reached at room temperature: chemicals that have a propensity to burn or explode are monitored in their LFL\LEL range if the TLV is high and health hazards are the lesser threat.
Three analytical instrument techniques are being used in cleanrooms today to monitor substances according to these categories of health and safety. These technologies are Fourier transform infrared (FTIR) spectroscopy for monitoring chemicals in ppm ranges for both health and safety and for air quality, molecular emission spectroscopy (MES) for monitoring toxic gases at ppb levels, and an acoustic sensing technique for monitoring hydrogen gas in the LEL range.
Using FTIR spectroscopy
Because individual compounds absorb infrared radiation in a characteristic pattern, a graph of absorption versus wavelength (the spectrum) may be used to identify components in a gas mixture by their complex spectra. Furthermore, the amount of light absorbed is proportional to the concentration of the component, providing quantitative analysis. FTIR is superior to other infrared techniques because the signal-to-noise ratios are higher, the resolution is better, the wavelength accuracy is superior, and data acquisition times are significantly shorter.
Infrared spectroscopy is a powerful analytical tool for air monitoring. The FTIR technique offers the advantages of low detection limits, spectral precision, use of mathematical interference correction techniques, and multi-chemical analysis with specific chemical identification.
Using FTIR, low concentrations of nearly all vapor phase chemicals in the air can be detected, including all organic chemicals, by using an encoded qualitative database. Further, complete spectra can be generated to identify unknown odors and suspected leaks. Sampling locations include gas cabinets, exhaust ducts, process tools, valve manifold boxes, scrubber exhaust, plant emissions and worker “breathing zones.” Normal operation provides for automatic, gas-specific monitoring and shut-off control and complete identification of incidents and alarms. (See Table 1)
In most cases, a built-in industrial computer automatically converts the information produced by the FTIR analyzer into concentrations of chemicals using an encoded qualitative database of reference spectra. For example, the monitoring setup might include 13 gases per point, requiring only 15-30 seconds per point for sampling, analysis, and data presentation. The user identifies the specific chemicals in use, and the monitor is set up to detect those chemicals.
FTIR can monitor any chemical in the vapor phase except atomic and diatomic gases (such as He, H2, Cl2, and O2), and it is used in semiconductor facilities to monitor nearly all of the compressed gases and acid and solvent vapors. While the TLV of some toxic gases may be below the detection limit of FTIR, this technique is useful for monitoring for the presence of such gases as plant emissions in the low ppm range.
MES for toxic gas monitoring
Whereas FTIR typically monitors gases at low part per million (ppm) and sub-ppm concentrations, MES reaches down to low ppb levels. MES detects toxic gases at concentrations well below their established TLV. Even at low ppb levels, the MES monitor is protected against cross-sensitivity to solvents , acid gases, and other common chemicals used in semiconductor fabrication processes. Many of these chemicals are known to produce false alarms on other types of detection techniques.
MES uses a hydrogen flame photometric detector, which continuously monitors for toxic substances, e.g., metal hydrides. As a toxic substance in the sampled air enters the flame`s corona, light emissions resulting from chain reactions are passed through two optical filters and are converted to electronic signals. An MES monitor assures it is always working. It performs periodic functional and response tests automatically. Pneumatic, electrical, and detection systems are checked constantly by its computer, which activates relays and prints out data for any specific problem. It uses an industrial computer with multitasking software to communicate with building controls, guard stations, and emergency response stations, as required.
A typical setup using MES would sequentially scan up to 20 locations, drawing air samples through tubes from monitoring points. Operating modes and sampling locations are similar to that of the FTIR monitor. Gas detector interfaces (GDIs) can expand the capability of an MES instrument beyond toxic gases. A GDI allows the monitor to read the output signals from all types of gas detectors, e.g., detectors for chlorine, mineral acids and explosive gases, all as if directly monitored by the MES detector. This greatly expands both the gas types and number of points covered by a single monitor.
Acoustic sensing to monitor hydrogen in the cleanroom
Acoustic sensing detects hydrogen in the workplace from concentrations as low as 100 ppm (0.25% LEL) to as high as 40% (1000% LEL). In some instruments, one tube has sample air flowing constantly, and one tube is filled with reference air and sealed. Sound is produced and clocked as it travels through the air in each tube. Results are expressed as ratios of clock counts so that cancellation eliminates any sources of drift. Electrical pulses generated as sound bursts are sent and received and are used to start and stop two electronic counters. Digital filtering allows for rapidly changing hydrogen concentrations during this interval but will not allow erroneous measurements, due to a malfunction, to be communicated to the system computer. Low levels of hydrogen in the sampled air greatly alter the time-of-flight of acoustic waves, making it easy to detect hydrogen at low levels. Helium is nearly as light as hydrogen and causes a similar (but lesser) change to the acoustic waves. Since helium is also used in semiconductor facilities, it is necessary to distinguish hydrogen from helium and other gases. Hydrogen is readily removed by palladium, but helium and other gases are not.
The proper use of a palladium getter, assures specificity to hydrogen by automatically validating all readings exceeding a warning level. Traditional monitoring for hydrogen has been done with gas detectors, either the catalytic bead or electrochemical type. Gas detectors have to be checked and calibrated regularly, and facilities using numerous hydrogen detectors have discovered that detector upkeep is manpower-intensive. Gas detectors are subject to interferences from a variety of chemicals and have no means of distinguishing those chemicals from hydrogen. As a result, interfering chemicals in the background air can generate false alarms. The acoustic approach checks itself, never requires calibration, and validates all alarms against interferences and other anomalies.
Preventing false alarms and identifying gases
Certain analytical techniques do not require gettering (or other means of chemical separation) to identify and validate gases. The FTIR technique identifies gases in the air by their infrared absorption spectra, which is like a “fingerprint.” On occasion, the spectral peaks used for the identification and measurement of a chemical might exactly overlap peaks of another chemical used in the same facility. Knowing this, an application chemist can set up the FTIR analytical method to account for such overlap by using other unique peaks of both chemicals and advanced statistical data fitting.
The MES detection method as well as the acoustic detector do not have the inherent “fingerprint” identification of FTIR. Nevertheless, both methods have a high degree of inherent chemical specificity. The MES uses a pair of narrow bandpass optical filters, one centered on the peak light region of toxic gas emissions and the other centered on the peak light region of interfering chemicals. This allows the MES to compensate for interferences so that they do not cause a positive reading.
The acoustic detector is inherently specific to gases having very low molecular weights, such as hydrogen and helium, because the time-of-flight of the acoustic wave becomes faster the more these gases mix with air.
The inherent specificity of the MES and acoustic detector is good but not good enough. Semiconductor facilities can suffer large losses when gases and processes shut down upon the detection of gas leaks. Therefore, it is imperative that the detected gas leaks are not caused by background chemicals that are not especially hazardous, e.g., Isopropyl Alcohol (IPA).
While these monitors are inherently protected against response to IPA and most of the other background chemicals, it is possible that a few chemicals might cause an upscale response. Certainly, helium will cause such a response on the hydrogen monitor. It is also possible that a transient condition might effect the monitor in a way that it registers an upscale response. It is important to the user that such conditions do not result in false alarms. That is why gettering is utilized.
In the MES monitor, gettering is triggered when a warn level (low-level alarm) is exceeded due to either a toxic gas or a malfunction or other interference. To verify a toxic gas, the air sample is diverted through a getter tube. If the signal diminishes, the getter has removed the toxic chemical that is being monitored at that sample point. If the warn level (or greater) signal is restored after successful gettering, the warn is valid. Relays and printouts are activated. If the getter fails to lower the reading, an interfering chemical could be the cause. If the reading should happen to drop when the getter is switched on but fails to return when it is switched off, the probable cause is a transient. Either condition, if left unchecked, might have caused a false alarm and shut down production. Gettering can prevent that from happening.
Gettering takes only a few seconds. For example, if the instrument samples a point and detects 35 ppb arsine with a warn signal set at 25 ppb (1/2 the TLV), its monitor dwells for approximately 6 seconds while the getter reacts with and removes the arsine from the air sample. If it is arsine, the detection level will drop to near 0 ppb. Then, the monitor switches out the getter and rechecks for arsine, requiring the reading to rise at least above the warn setpoint. Passing this process validates the arsine leak.
Faster, simpler, more reliable, and sensitive monitoring–with greater specificity–for air quality, toxic gases, and explosive gases, such as hydrogen, is becoming a reality. Monitors work automatically, 24 hours a day, with minimal equipment downtime and virtually no loss of production. Centralized monitoring utilizing powerful analytical techniques is user-friendly and can be operated by non-technical personnel. These analytical techniques are fast and specific and are more reliable and sensitive than individual sensor-based systems, which require costly labor-intensive maintenance and consumable parts. n
1. The TLV is an eight-hour-time-weighted average airborne concentration of a sbustance and represents the conditions under which it is believed that nearly all workers may be repeatedly exposed day after day without adverse effect.
William C. Scott has served as vice president of TeloSense Corp. since its founding in 1981. Previously, he was manager of analyzer systems at United Controls and a product specialist at the former Beckman Instruments. For more than 25 years, Scott has been involved in product development and marketing and sales of process and environmental analyers. He earned his bachelor of science degree in chemistry from the Univeristy of Cincinnati.