Recent advances in compact, smart vacuum, and gas pressure sensors
01/01/2002
 Ignitorr device |
overview
Miniaturization of sensors and electronics has led to several recent advances in high-vacuum gauges, including extended ranges and smart capabilities, as well as smaller size. For example, the range of Pirani gauges has been extended down to 10-5 torr, response time has been reduced, and electronics are integrated on the same chip as the sensor. Direct pressure sensors based on piezoelectric elements are significantly smaller than conventional capacitance manometers. All vacuum gauges, including ionization gauges, have more control capabilities built into smart controllers. Inverted magnetron ionization gauges are more robust than their hot-filament cousins and have been shown to be equal in accuracy and repeatability.
Gas pressure measurement is critical in many IC fabrication processes and in thin-film coating processes used to fabricate flat panel displays, fiber optics, and MEMS. Process pressures range from the ultrahigh vacuum (~10-9 torr) required for physical vapor deposition (PVD) or evaporation to chemical vapor deposition (CVD) at atmospheric pressure. Other semiconductor processes that require vacuum sensing, measurement, and control include rapid thermal processing (RTP), ion implantation, vacuum-to-atmosphere transfer chambers (i.e., load-locks), and extreme-UV lithography.
Microelectronics technology has enabled a significant size reduction of thermal conductivity and direct-reading diaphragm gauges used for pressure measurement in vacuum and near-atmospheric pressure applications. In addition, microprocessors have greatly expanded the intelligence of control units. For example, calibration factors can be built in and pressure units can be selectable. So-called smart sensors can be designed to auto-calibrate, establish two-way digital communication with tool controllers, and self-diagnose problems.
Different types of pressure gauges measure a different characteristic of the atmosphere under test (Table 1). The demands of semiconductor manufacturing have pushed gauge designers to extend sensor ranges and increase repeatability. For example, high-density plasma processes, such as CVD and etch, operate below 10 mtorr, whereas previous low-density plasmas operated above 10 mtorr. Many thermal CVD and RTP steps operate above 100 torr.
 |
Direct reading gauges measure actual pressure exerted independent of gas composition (e.g., the ordinary mercury barometer), usually relative to a high vacuum or atmospheric pressure. These gauges have the advantage that the indicated pressure of the measured gas is independent of gas composition or mixture. By comparison, indirect reading gauges, such as those based on thermal conductivity or ionization, measure some property of a gas that relates to pressure. Unless one knows the gas composition, errors or inaccuracies may result. A capacitance manometer, on the other hand, will read accurately for all gases and need only be calibrated with one gas to reflect total pressure accurately for all gases and mixtures.
Capacitance manometers
In a capacitance manometer, one side of a thin metal diaphragm is sealed at high vacuum (typically <10-7 torr); the other side is exposed to the pressure being measured. The displacement of the diaphragm toward the low-pressure side is detected by sensing a change in capacitance between the diaphragm and two facing electrodes, one on the low-pressure side and one on the measurement side [1].
Problems plagued early capacitance manometers. When cycled to atmospheric pressure, the diaphragm mountings fatigued and failed prematurely. They were temperature-sensitive and the zero setting drifted with changes in ambient temperature. Also, because an electrically active element was exposed to a process, applications were limited to those using dry, inert gases.
In today's gauges, the curvature of the diaphragm is sensed by two capacitor plates, both located on the reference side of the diaphragm. The capacitor plates consist of metal films on a ceramic substrate — two concentric rings forming a bull's eye. These pressure transducers are called single-sided.
In a single-sided capacitance manometer (Fig. 1), as the diaphragm deflects, both electrodes sense a change in capacitance, but by differing amounts. The difference is sensed by a capacitance bridge and converted to DC voltage output. In an absolute sensor, the reference side of the device is evacuated to high vacuum and sealed off. Usually, an active metal getter that reacts with residual gases and outgassing is included in the reference volume to maintain a low pressure.
 |
The advantage of this design is that the measurement side has no electrodes and can be constructed of clean, corrosion-resistant materials, typically the nickel-chromium alloy Inconel. Inconel is preferred because of its resistance to oxidation and corrosion by halogen gases. Reference side materials, which are isolated from the process being measured, include palladium electrodes on a ceramic plate, and glass-sealed electrical feedthroughs.
For gauge measurements with a capacitance manometer (pressure referenced to atmospheric pressure), the reference side is left open to ambient pressure. Gauge measurements are typically used in gas line pressure measurements or in load-locks.
Application-specific capacitance manometers
Many plasma etch and CVD processes can produce by-products that deposit on exposed surfaces in the vacuum chamber. A film or particle deposit on the diaphragm of a capacitance manometer causes the pressure reading to drift, and requires constant monitoring of the zero setting and frequent re-zeroing.
Such applications have led to application-specific gauges for plasma etch systems [2]. A Baratron, for example, has a channel machined into the lower vacuum housing of the manometer that acts as a baffle between the vacuum chamber and the manometer diaphragm. As molecules travel through the channel, they collide with the channel walls thousands of times before reaching the diaphragm. This protects the diaphragm from deposits while allowing the gas molecules to reach the diaphragm for accurate pressure measurement. In addition, a two-stage particle trap prevents particles >250µm from contaminating the diaphragm without affecting vacuum conductance or reducing response time.
Piezo transducer
Sensing mechanical strain or displacement of a thin silicon membrane with a piezo transducer is a recent innovation in direct reading manometers. As with other diaphragm-based vacuum sensors, the piezo transducer output is gas-independent, making one calibration curve applicable to all gases and mixtures. With these devices, a silicon membrane is suitable for noncorrosive environments. Replacing the silicon membrane with a stainless steel PiezoSteel transducer provides corrosion resistance to fluorine and HF gases.
Because the sensing diaphragm area is smaller than that of a traditional capacitance manometer, the sensitivity of a piezo gauge is lower than a capacitance gauge, but is still significantly greater than Pirani and convection Pirani sensors.
Thermal conductivity gauges
Pirani, convection-Pirani [3], thermocouple gauges, and miniaturized direct-reading piezo sensors are useful for vacuum foreline and pump applications, and for monitoring load-locks and roughing line pressure. In addition, thermal conductivity and piezo gauges are useful for slow-pumpdown control, and vacuum load-lock operation.
Conventionally, 10-4 to 1 mtorr was difficult to measure accurately with traditional Pirani or ionization vacuum gauges. Pirani sensors rapidly lose sensitivity above ~10 torr and ionization gauges are not well suited for operation above 10-4 torr. The convection-Pirani gauge has improved high-pressure sensitivity, but suffers from poor accuracy and slow response time, and it must be oriented properly.
Advanced solid-state pressure sensors — putting a MicroPirani sensor on a 1mm silicon chip — can now cover this entire range with 1% reproducibility and 0.1% resolution. In fact, this miniaturization of the traditional Pirani sensor has extended the upper pressure measurement range from 10 to 1000 torr. This is because the geometry of the sensor (i.e., the distance between heat source and heat sink) is less than or equal to the thickness of the thermal sheath around the filament when the system is near atmospheric pressure. In addition, because the sensor mass and volume are so much smaller, the amount of gas heated and the thermal mass of the sensor have been greatly reduced, yielding faster sensor response time.
A MicroPirani sensor puts the vacuum sensor element and temperature compensation sensor on the same chip. This improves the response of the bridge and also makes temperature compensation more repeatable, because the temperature sensor is located on the heat sink. With an on-board temperature compensation circuit and a stable ambient temperature, the measurement range can be extended down to 10-5 torr. This sensor and most other semiconductor-based sensors are designed for use in noncorrosive atmospheres (Table 2).
Combination sensors
By combining piezo and MicroPirani elements in the same "envelope," designers can achieve a very small wide-range pressure sensor (e.g., 10-5 to 1500 or 10-5 to 3000 torr). MicroPirani, piezo, and wide-range Dual Trans combination sensors are available. The PiezoSteel sensor has been designed especially for corrosive environments; here, a stainless steel membrane protects the piezo element and only 316 stainless steel and Viton are exposed to the gas environment. All of these sensors have a high-pressure range up to at least 1 atm and can be operated in any position. The sensors have standard ISO-KF NW 16 flanges and an internal volume of only 0.6 cm3.
Ionization gauges
Hot and cold cathode ionization gauges are both indirect-reading (i.e., the relation between read and true pressures depends on the gas being measured). For example, the gauge sensitivity for hydrogen is one-third to one-half of that for nitrogen. This means that a given pressure reading of a nitrogen-calibrated gauge in hydrogen represents a true pressure that is two or three times higher. In addition, the absolute sensitivity of an ionization gauge can vary up to 25% for the same type of gauge. Sensitivity and variability specifications should be obtained from ionization gauge-tube manufacturers.
The most common hot-cathode gauge is the Bayard-Alpert (B-A) triode, which can measure pressure from ~10-5 torr into the ultrahigh-vacuum range. The gauge can be enclosed in a glass or metal envelope or attached "nude," with the sensor element extending into the vacuum chamber. The latter eliminates conductance limitations between the sensor and the region to be measured and outgassing from gauge walls.
The pressure range of a B-A gauge can be extended into the mtorr range by miniaturization. The major disadvantage is that there is a hot filament that can burn out or be degraded by active gases. The filament also causes outgassing that can cause inaccurate pressure readings and change the composition of residual gases.
The usual filament in a B-A gauge is thorium-oxide-coated iridium, chosen for its burnout resistance. In most gauges today, thorium oxide has been replaced by nonradioactive yttrium oxide. Burnout-resistant filaments may not be a good choice if halogen-bearing gases are present, which degrade the electron-emission ability of both yttrium and thorium oxides. The alternative is a tungsten filament, which quickly fails if it is exposed to air or oxygen. With either type, filament life is unpredictable and depends on abuse incurred by the filament such as exposure dose (i.e., pressure x time) to high pressure or reactive gases.
The cold cathode Penning (or Philips) ionization gauge, sometimes called a glow-discharge gauge, does not have a hot filament to burn out. It is simple, has a metal envelope, a robust electrode structure, and negligible outgassing. The gauge requires a magnet and, if not properly shielded, can have a residual external magnetic field.
Many engineers perceive hot cathode gauges to be more accurate than cold cathode gauges. In reality, the two types of gauges can be equally accurate. The original Penning gauge has evolved into the inverted magnetron (IM) cold cathode gauge with a range from ~10-9 torr to ~5 mtorr. Repeatability (~±5%) and sensor-to-sensor matching (±20-25%) of IM gauges are equivalent to conventional glass-envelope hot cathode gauges [4-6].
 A single-sided capacitance manometer. Px is the measurement side and Pr is the reference side. |
Although a cold cathode gauge is reluctant to start at low pressures (starting depends on field ionization or a random ionization event triggered by a cosmic ray), there is no starting delay at 1 mtorr or higher. The lower the pressure, the longer the starting delay, which can be several minutes at 10-6 torr. To address this problem, UV light (e.g., using the MKS IgniTorr) can be projected into an IM gauge to trigger ionization.
Although the unit cost of a cold cathode IM gauge sensor can exceed that of a glass-envelope B-A gauge, in aggressive gas environments, cost-of-ownership can be lower because of extended lifetime. In addition, a contaminated cold cathode gauge can often simply be cleaned and re-installed, further extending lifetime if cleaning cost is reasonable. A precision hot cathode gauge tube with equal repeatability would cost more, but would still have a limited useful life.
Advanced controllers
Microcircuitry allows several useful functions to be incorporated in compact, electronic controllers, often integrated with the gauge sensor. For example, setpoints can be incorporated so that high and low pressure signals can activate valves, alarms, process steps, relays, etc.
One application for a smart transducer is in vacuum load-locks. Load-locks are often pumped down very rapidly, too fast for a Pirani gauge to track. Also, when a load-lock is vented, a separate vacuum switch is often used to indicate when the chamber has reached atmospheric pressure and the door can be opened. An improved alternative is an integrated transducer combining a MicroPirani gauge and a piezo-based atmosphere switch in a single package. The piezo is a differential pressure sensor that senses when the load-lock's external and internal pressures are equal, and thus immune to variations in external pressure due to weather and elevation. Such an integrated controller can provide analog output and two setpoints for automated control.
Acknowledgments
Baratron is a registered trademark and IgniTorr is a trademark of MKS Inc.; Dual Trans, MicroPirani, Piezo+, PiezoSteel, and PiezoSteel+ are trademarks of Wenzel Instruments ApS; Inconel is a registered trademark of INCO Alloys Intl.; Viton is a registered trademark of DuPont Inc.
References
- J.J. Sullivan, "Development of Variable Capacitance Pressure Transducers for Vacuum Applications," J. Vac. Sci. Tech. A 3(3), pp. 1721-1730, May/June 1985.
- A. Hood, et al., "The Development of Application-Specific Components and Subsystems," Semiconductor Intl. 24(2), Equipment Components and Subsystems Supplement, pp. 10-16, Feb. 2001.
- B. Dort, "Using the Pirani Gauge for Indirect Measurement of Vacuum," Sensors, June 1995.
- D.L. Hyatt, N.T. Peacock, "Long Term Measurement of an Inverted Magnetron Cold Cathode Gauge," Proceedings of the 37th Technical Conference, Society of Vacuum Coaters (1994), pp. 409-412.
- R.N. Peacock, N.T. Peacock, D.S. Hauschulz, "Comparison of Hot and Cold Cathode Ionization Gauges," J. Vac. Sci. Technol. A 9(3), pp. 1977-1985, May/June 1991.
- B.R.F. Kendall, "Ionization Gauge Errors at Low Pressures," Journal of Vacuum Science Technology A 17(4), pp. 2041-2049, July/August 1999.
Neil Peacock received his BA in physics from the University of Colorado. He is staff engineer at MKS Instruments, HPS Products, 5330 Sterling Dr., Boulder, CO 80301; ph 303/449-9861, [email protected].
Robert Waits received his masters of science degree in physical chemistry from Stanford University, Stanford, CA. Waits is currently a vacuum technology consultant and a technical writer in Sunnyvale, CA.