Avoid false alarms with proper gas detection equipment
08/01/1997
Avoid false alarms with proper gas detection equipment
Edward M. Zdankiewicz, Sensor Systems, Keithley Instruments Inc., Chemical Measurements Group, Cleveland, Ohio
Leak prevention, ventilation, and personnel training are important elements of a gas safety program. Another essential element is fugitive gas monitoring, which is required by national and local codes to protect fab personnel from toxic gas poisoning and suffocation from oxygen displacement. It also provides the warning needed to help protect both equipment and personnel from combustion or explosion hazards. Reliability is the most important characteristic of a gas detection and monitoring system. Two of the most critical reliability factors are sensitivity to a specific gas and freedom from false alarms, both of which require appropriate sensor technology. Other reliability factors associated with the sensor and its detector head include low drift, repeatability, resistance to poisoning (overload), low cross-sensitivity to other gases, sampling pump reliability, and filter performance.
Identifying potential sources of leaks provides the necessary information to prevent them and provides clues to where gas detectors should be located. Gas delivery systems usually contain a large number of valves, fittings, and welds. Fittings, filters, and regulators are always potential leak points due to shock, vibration, and the degradation of seals over time. Table 1 lists gases commonly used by semiconductor manufacturers and the hazards they pose.
Of course, the process tool has the potential for unplanned release of gases used. For example, an inadvertent release during cylinder replacement in a gas cabinet sometimes occurs; the plasma-based etching process leaves hydrocarbon deposits on the inner surfaces of the reaction chamber and its components. This buildup is a possible source of emissions when exposed to air and moisture during preventive maintenance.
Preventing and dealing with hazardous gas leaks requires effective safety practices and procedures. Some of the procedures frequently recommended are:
conduct formal safety reviews at least annually, and any time new process equipment is purchased or equipment modified,
implement regular gas safety training programs,
limit the number of cylinders stored on-site through just-in-time delivery, and
contract with gas suppliers to provide on-site technicians for cylinder replacement.
Besides the planning and use of safe handling procedures, good gas delivery system design is essential. Important design features for hazardous gas systems include:
selecting components and materials suitable for reactive gases,
double containment for gas lines, where appropriate,
good ventilation around piping,
leak testing prior to use,
appropriate use of check valves and flow limiting orifices,
automatic shutoff valves,
pressure- and vacuum-cycle purge on process stations, and
backup power for fire protection and exhaust systems.
Monitoring and alarm systems
A final, and equally essential, line of defense against hazardous leaks is a reliable gas detection and alarm system. Such a setup is called for in the Uniform Fire Code (UFC) and by local ordinances in many locales if a gas meets the following criteria:
1. It is a hazardous production material, i.e., has a Class 3 or 4 hazard rating per National Fire Protection Association Standard 49, and
2. Does not have a physiological warning property below the accepted Permissible Exposure Limit (PEL) or the Lower Explosive Limit (LEL).
These codes require a gas detection system that operates continuously and sounds an alarm when the gas concentration exceeds a designated level. Typically, this occurs at the PEL and at 25% of the LEL, triggering alarms in the monitored area and at a continuously staffed emergency control station.
For fixed installations, the monitoring and control system usually consists of multichannel, rack-mounted hardware, including computer-controlled PLCs and annunciators. Typically, diagnostics and reporting functions are included in these computer-based systems. Fixed installations may be augmented by portable gas detection instruments in selected areas.
 |
Reliable detection without false alarms
There are two sides to reliability: detecting gas concentrations above specified levels and ignoring concentrations below those levels. A false alarm can occur when:
an inappropriate sensor is selected for the target gas or for a situation where other (interfering) gases will be present;
a sensor is not calibrated properly;
the sensor drifts outside its intended operating range; or
other detection system components malfunction or do not work together properly.
In a semiconductor fab, false alarms are much more than a simple annoyance. A false alarm that shuts down a line can easily cost $100,000/hour in lost wafer production. False alarms also have the potential to contaminate clean areas if members of the fire department or hazardous materials team enter without taking pre-cleaning precautions that are normally required.
False alarms pose dangers beyond loss of production, too. Frequent false alarms undermine employee confidence in the monitoring system and are frustrating to operating, maintenance, and emergency personnel. Perhaps the most serious danger is that fab personnel may ignore what they think is a false alarm and fail to respond promptly to a truly hazardous condition.
Preventing false alarms requires attention to all elements of the gas detection system. This means:
selecting proper sensors for the gases to be detected,
placing those sensors in appropriate locations,
making sure that sensors are calibrated,
maintaining the system, using effective diagnostics,
monitoring historical concentration trends in critical locations,
using a system design that fits the facility, and
creating a tightly integrated system.
Criteria for sensor selection
Proper sensor selection is the first, and probably most important, step in preventing false alarms. Use of the right sensor technology maximizes sensitivity and minimizes response times, to provide timely alarms when they are needed. Selection criteria include response time, sensitivity, output signal level, gas selectivity, cross-sensitivity to other gases, and resistance to poisoning. The sensor is usually located inside a detector head, which is placed at or near the point of measurement. The two main types are based on either diffusion or sample draw by direct suction of gas into the detector head. The diffusion type is used for general area monitoring. The suction pump type is used where highly accurate gas sampling is required, or when the gas sample must be extracted from a process exhaust line or duct.
Usually, one of three predominant sensor technologies is selected for a specific target gas, as determined by the sensor performance criteria mentioned earlier. The commonly used sensing principles are:
electrochemical,
physiochemical,
physical,
optical.
Using these four detection principles, sensor manufacturers have created a wide variety of designs that incorporate several device technologies. Some of the more important considerations in selecting one technology over another are described below.
Electrochemical sensors. Electrochemical sensors react to a specific toxic gas. Two implementations of this basic technology are controlled potential electrolysis and galvanic cell electrolysis. A controlled potential electrolysis sensor (Fig. 1) is a three-electrode device that measures the concentration of a gas. It detects the current generated when the gas is electrolyzed at a specific bias voltage applied to the electrochemical cell. The active electrode is made from material selected for the specific gas to be detected. The current between the active electrode and a counter electrode created by electrolysis of the target gas is compared to that between a reference electrode and the counter electrode. The sensor output is a linear voltage differential proportional to gas concentration.
 |
Figure 1. Basic circuit for a controlled potential electrolysis sensor.
Electrochemical sensors are widely used for detecting toxic gases such as CO, B2H6, HCl, HCN, AsH3, Cl2, NF3, and PH3. The controlled potential type is extremely sensitive, with the ability to detect diborane (B2H6) at 100 ppb concentration. The controlled potential cell also has low cross-sensitivity, so it is less likely to cause false alarms or fail to warn in the presence of other gases. Maintenance includes calibration at regular intervals, or just periodic sensor replacement.
Galvanic cell. In some respects, a galvanic cell sensor is similar to the electrolytic type. A diaphragm galvanic cell (Fig. 2) works by measuring the reaction current generated when a gas (i.e., oxygen) is the reactant in the electrolytic cell. Galvanic cells are specifically designed for monitoring low and high concentrations of oxygen, where either situation can be hazardous to personnel and process equipment.
 |
Figure 2. Outline drawing of galvanic cell.
The O2 in air diffuses through a selective membrane into the galvanic cell and reacts with a noble metal (Pt, Ag, etc.) electrode. This reaction creates an electrical potential between this electrode and a base metal (Pb) electrode, causing current to flow though an external circuit. The external circuit uses a resistor to create a voltage drop proportional to O2 concentration. Since the current generated by the cell is also dependent on temperature, this circuit contains a thermistor in series to compensate for ambient temperature changes.
In a galvanic cell sensor, no external power source is needed to detect the gas. Its output is linear and proportional to O2 concentration up to 40% volume. It is also small, light-weight and inexpensive. Maintenance is minimal and recalibration infrequent.
Catalytic sensors. When a combustible gas comes into contact with the catalytic sensor element, the gas oxidizes. Oxidation is enhanced by treating the sensing surface with a catalytic coating and by operating the sensor at an elevated temperature. An embedded platinum wire senses the temperature increase as the oxidation reaction releases heat. The change in temperature is linearly proportional to the gas concentration present and can be accurately measured using a Wheatstone Bridge circuit. A typical sensor operating temperature for gas detectors using catalytic combustion sensors (Fig. 3) is 100?C, with an alarm setpoint at 1/4 LEL.
These sensors are typically located near the source of a potential leak and are widely used for gases combustible in air, such as methane, LPG, acetylene and H2. Their output is linear and proportional to gas concentration up to the LEL. They have a fast response time, excellent repeatability, and high accuracy and are virtually unaffected by temperature and humidity variations. Catalytic sensors have no moving parts to cause mechanical failure, and they require infrequent calibration. When the catalytic bead is expended, normally the sensor is replaced, but it can last for two or three years.
Catalytic sensors are poisoned by silicones, sulfides, and chloride gases. They do not work well in a low-oxygen (<10%) environment and will respond to all combustible gases, not just the one used for calibration. To prevent false alarms, use care in the application of catalytic sensors where interfering gases could be present. However, when other gases in the monitored area are among the combustible types mentioned above, inteference may not be a problem.
 |
Figure 3. Catalytic combustion sensor circuitry.
Thermal conductivity sensors. This type of sensor (Fig. 4) measures the resistance change caused by a temperature decrease in a heated (embedded) platinum wire, resulting from its exposure to the target gas. The temperature decrease is a function of the thermal conductivity of the gas. Each gas has a unique thermal conductivity.
The sensors compare the change in the gas sensing element temperature and its resistance to a temperature compensated reference element filled with a standard gas. A bridge circuit measures the sensor element`s resistance change and gives an output linear voltage signal in proportion to gas concentration.
Thermal conductivity sensors can be calibrated for a wide range of gases, including Ar, He, Ne, CO2, H2, and combustible hydrocarbons. Gases with conductivities similar to air cannot be detected because air or N2 typically is used in the reference element. Sensor output is linear up to 100% volume, making these sensors suitable for high concentrations.
Since their detection principle is based on the physical (i.e., not chemical) parameter of thermal conductivity, these sensors are immune to poisoning and remain stable over long periods. Measurement does not depend on the presence of oxygen, as is the case for catalytic sensors. Maintenance and calibration requirements are low.
 |
Figure 4. Conceptual view of thermal conductivity sensor bead.
Semiconductor electrical conductivity sensors. Semiconductor chemical sensors rely on a change in electrical resistance within a chemically sensitive semiconductor element, corresponding to a change in chemical concentration. These are high-resistance devices with operating resistance levels ranging from a few kW to several MW. Most semiconductor sensor designs incorporate an embedded heating element to tune the sensor thermally to the target gas of interest.
Tin oxide (SnO2) is an n-type, metal oxide semiconductor material commonly used for the sensing element. Typically, the element is manufactured by depositing a thin film of SnO2 onto a substrate or by sintering SnO2 powder to form a small semiconductor bead (Fig. 5). In operation, two counterbalancing reactions occur. Without a target gas present, ambient oxygen molecules chemisorb onto the element, removing electrons from the tin oxide grain surfaces, so element resistance increases due to electron depletion. When a target chemical gas is present, it produces an oxidation reaction at the element`s surface, releasing electrons back into the SnO2 grain surfaces, and causing a corresponding increase in film conductance (i.e., a resistance decrease.)
One version uses the heating element as part of the detection device. Two other designs work in the manner of a field effect transistor: the conduction channel is narrowed by the charge redistribution created when molecules of the target gas are adsorbed. Construction details vary to allow each sensor to be calibrated for the target gas. In one case, the sensor is calibrated for H2. In another, the sensor can be calibrated for the common combustible (hydrocarbon) gases, CO, and many fab process solvents. A third implementation can be calibrated for Cl2, H2S, or ethylene. In some sensor designs, membrane filters improve selectivity.
 |
Figure 5. Typical thin film semiconductor sensor.
Sensitivity varies by design, but most provide a high output at low concentrations. Because of their detection method, small size, and low power consumption, the initial stabilization period for most of these sensors is short, as is response time. They tend to have long lives (up to 10 years), excellent long-term stability (low drift), and are resistant to poisoning. Since they work at relatively high temperatures (250-400?C), the semiconductor surface tends to be self-cleaning, so repeatability is good and maintenance low.
Infrared sensors. In optical sensors, infrared radiation (light) passes through the gas sample in a measurement cell and also through a comparison cell (Fig. 6). In the measurement cell, the gas absorbs infrared light. The light passing through the two cells is measured by two photodetectors, which create a differential signal proportional to the concentration of the target gas in the measurement cell. The detector head then amplifies and conditions the signal.
The specific infrared wavelength absorbed is different for different gases, so the sensor can be tuned for a specific target gas. Typically, infrared sensors detect CO, CO2, refrigerant gases (CFCs and HCFCs), and combustible gases. These sensors can be accurately tuned for hydrocarbon molecules and ignore nonhydrocarbon combustible gases, such as H2. Therefore, infrared sensors have minimal cross-sensitivity and are impervious to poisoning by other gases or by high concentrations of the target gas.
Other performance features are similar to those of catalytic sensors. However, due to the optical nature of these sensors, they may not perform well in environments with high (condensing) humidity or very dusty environments.
Paper tape detectors. Paper tape instruments are another type of optical device and use chemically impregnated strips of paper to detect toxic gases. When exposed to the target gas, the paper changes color in a manner similar to a piece of litmus paper. A photocell detects the color change and translates it into a concentration value. While this technology can detect various toxic gases, it cannot be used for most combustible gases.
 |
Figure 6. Detection principle for an infrared sensor.
Paper tape instruments are relatively expensive, so they are typically placed in a central location and connected to the measurement points by long lengths of sample tubing. Samples are pumped from each measurement point sequentially. With sequencing and sample tubing running as long as 300 ft, there can be significant lag times between the time at which a leak occurs and its detection at the instrument. Also, reactive gases, such as Cl2, HCl, HF, and NH3, are easily absorbed within the tubing. This can prevent the instrument from detecting the correct concentration of such gases at the measurement point.
One advantage of a paper tape instrument is its ability to verify visually that a color change has occurred on the paper. They are also good at detecting nonreactive toxic gases at low levels of concentration and will detect a wider range of these gases than many electronic sensors. Due to their mechanical design, paper tape instruments are more prone to failure than electronically based sensors. The paper tape cassette drive can jam, optics can be fouled, and flows can become unbalanced. Therefore, these instruments require regular preventive maintenance. Also, the paper tape is a consumable that must be replaced every two to four weeks.
 |
Matching the detector to the application
Table 2, which lists some typical sensor and detector head applications, shows that different ambient conditions may dictate a different sensor type and sampling scheme for the same gas or location. The table also suggests that no single detector technology works reliably for all the gases found in a typical fab. To get reliable detection without false alarms, you must evaluate each measurement point and use the sensor/detector head combination that provides the best sensitivity, selectivity, response time, accuracy, and immunity to prevailing environmental conditions.
Sensor and system reliability are intertwined
Detector head design and dependability are also an integral part of gas detection system reliability. For example, size is important when tight spaces are involved at the measurement point. In suction types, the pump must be highly reliable and all parts of a detection head must be designed to withstand a wide range of environmental conditions. The electrical output must be immune to EMI and RFI. A filter used to protect the sensor from particulates and moisture must strike a balance between low permeability and high contaminant capacity to prevent clogging. A gas (membrane) filter must exhibit good selectivity with adequate flow rate to ensure fast response to the target gas.
The ability to detect and diagnose sensor and detector head problems is essential for maintaining reliability. This distributed function is contained partially within the detector head itself and partially within the gas detection network controller. Detector heads typically monitor the gas sensor for open or short circuits, out-of-range response, and low (or loss of) sampling pump flow rate. The controller continually checks for data communication errors and network problems between itself and the detector heads.
Overall system design and layout also have a large impact on reliable detection and the avoidance of false alarms. A general design characteristic of a multipoint gas detection system is its sampling and data communication scheme. The types of detectors used will influence this design. For example, with a paper tape sensor, an alarm occurring at the central instrument location must be transmitted back to the monitored area, typically requiring point-to-point wiring from the instrument to each area. Similarly, many electronic detectors use point-to-point wiring from the monitored area to a central emergency station.
Cost and other system design issues
The current trend uses a system design with a distributed detector architecture and data communication bus. The detector heads are placed at the measurement point and connected to a central location through an RS-232 bus, power line carrier, or other data communication scheme. The monitoring and control system typically is PC-supported, with all supervisory control functions accessible through a graphical user interface (Fig. 7).
Once beyond a few monitored points, a detector network architecture can significantly improve reliability and lower wiring installation costs by reducing the number of cables between the measurement points and the controller. Another advantage is a high level of system flexibility and scaleability, which allows simultaneous monitoring of as many different gases and detection points as necessary.
 |
Figure 7. Typical networked gas detection system architecture.
This architecture also simplifies system integration by using a common data communication interface and protocol throughout the system. It opens the door to highly effective supervisory control and monitoring software, such as Wonderware`s InTouch, with its intuitive human-machine interface (HMI). Other software features to look for are real-time alarm and gas level trend displays, floor layout displays, historical alarm and gas level trends, real-time diagnostics, an event database, and easy compliance report generation.
Although reliability is paramount, users still want the best detection scheme per dollar. For example, reliability affects cost of ownership by the effect it has on maintenance and downtime costs. And false alarms have a huge negative impact on production when they cause unwarranted shutdowns. Poor diagnostics can also increase downtime and maintenance costs; so can frequent calibration requirements. These and other life cycle costs are listed in Table 3, which is based on SEMATECH`s well-known Cost of Ownership model.
 |
Conclusion
When evaluating gas detection system designs and the components that go into them, it is important to quantify the costs and benefits of reliability. In particular, examine costs associated with false alarms and maintenance of gas detection systems, along with their potential effects on equipment utilization and production yields.
EDWARD M. ZDANKIEWICZ holds a BSME from Case Western Reserve University, Cleveland, OH, an MSME from the University of Toledo, and is a Registered Professional Engineer in Ohio. He is an engineering manager for the Chemical Measurements Group at Keithley Instruments; his responsibilities include the design, development and integration of gas detection systems. Previously, he was an engineering manager in Keithley`s Semiconductor Division, involved in the design and development of automated parametric testers. Keithley Instruments Inc., 28775 Aurora Rd., Cleveland, OH 44139; ph 800/552-1115, fax 216/248-6168, e-mail [email protected].