Advanced Instrumentation for Optimizing Semiconductor Strain Gage Performance

Abstract

Semiconductor strain gages (SSGs) have very high change in resistance when strained. On a comparative basis, the sensitivity of semiconductor strain gages is fifty to one hundred times (50X-100X) the sensitivity of foil gages. To deliver the impressive performance benefits of homogeneous semiconductor strain gages, M. has optimized characterization of gages as a function of temperature to account for the thermal effects on resistance and to maximize strain sensitivity. We describe the motivation, challenges, and approach to optimizing the performance of SSGs through advanced instrumentation.

Integrated scientific resources, Inc., designed and built a new instrumentation system successfully replacing the legacy system that has been used successfully by M. since the mid eighties. The new system consists of two multimeter/scanner systems from Keithley instrument (3760A) which concurrently measure the devices under test. A Watlow controller is used to control the over while data are collected.

Semiconductor Strain Gages

Semiconductor strain gages were discovered during the transistor era and became commercially available early in the 1950’s. These gages may be homogeneous, diffused or silicon on sapphire (SoS) deposited. Diffused gages have a variety of problems that limit their useful life and affect their performance. SoS deposited gages have a much lower gage factor and are less corrosion resistant. Homogeneous gages have good performance and higher gage factors.

There are two main elements from which homogeneous semiconductor gages can be made. These elements are Germanium and Silicon. Both can be P- or N-doped. At M. Instruments, the P-doped (Boron) Silicon material is selected for use in our strain gages. The N-doped Silicon is selected for use in our temperature gages. Silicon gages have been proven to be more stable and corrosion resistant than Germanium gages.

M. Instruments Miniature P-Doped Silicon Semiconductor Strain Gages

M.’s strain gage is manufactured from a Boron-doped Silicon ingot grown as a single crystal. The strain gage crystalline axis is selected to maximize the longitudinal to transverse ratio. For the Silicon temperature gage, the reverse is true. This axis selection results in finished strain gages that will have high strain sensitivity along the longitudinal axis and low strain sensitivity along the transverse axis. For the temperature gage, the axis selection results in a low strain sensitivity along the longitudinal axis.
The gage shape is application dependent. Semiconductor strain gages are normally bar-shaped. The length and resistance are varied to suit the application. The width is kept nominally at 0.005 inches1 while the thickness is kept nominally at 0.0005 inches (See Figure 1). Gold leads are bonded to the ends of the gage for electrical connection. For very small sensors, it may be necessary to have the gold electrical leads oriented in the same direction as they come off the gage. In these cases, a U-shaped gage geometry is used. Since the U-shaped gage has twice the length of material as a bar-shaped gage for the same overall length, the U-shaped gage has twice the resistance making it desirable for small areas of high strain or for wireless applications where higher resistances are important. There are also M-shaped gages that provide even higher resistances in the same physical space as a bar gage.
Such gages are difficult to handle because of the very small diameter gold leads that measure 0.002 inch-es in diameter. Handling these gages without breaking them during gauging operations requires operator skill and experience.

Challenges in Temperature Compensation for Homogeneous SSGs

SSGs have higher temperature coefficients of resistance (TCR) than foil gages. For this reason, obtaining consistent sensor readings with a single semiconductor gage design is very difficult unless the temperature is constant. The use of semiconductor gages in full Wheatstone bridges of four gages or two gages in half bridges eliminates the temperature-dependent problem if the gages are thermally matched so they have the same resistance at all temperatures over the temperature compensated range. In other words, when used in a two- or four-element bridge, the null balance does not change if each of the strain gages changes resistance at the same rate as a function of temperature.

Most semiconductor gage manufacturers do not accurately match their gages over temperature because of the cost. This is due to the difficulty of handling very small gages which, at M. Instruments, can be as small as 0.018” long by 0.012” wide by 0.0004” thick and the equipment needed to accurately measure resistance and temperature for production quantities of strain gages.

Figure 3 shows the printed circuit board used to mount forty-eight (48) individual strain gages. Each gage is soldered to a position on the board. Since the gages are attached only by their gold leads, it is necessary to protect the gages from the air currents that circulate in the conditioning chamber.

Figure 4 shows a four-board baffle used to protect the populated boards from air currents. The baffle is inserted into the conditioning chamber and each board is inserted into its respective chamber. Thus, one gage test system can characterize one hundred and ninety two (192) gages at a time.

The baffle attenuates the temperature change rates that the gages would otherwise see as the chamber heater and cooling gas are modulated to achieve the commanded temperature. A thermal gradient still exists though vertically from board to board and laterally and in depth across each board. To know what the temperature is at each gage location at the moment its resistance is being read, each column of twelve gages (Figure 3) is instrumented with a leading and trailing Platinum RTD that has an accuracy of 0.5°F, NIST traceable. The temperature at each RTD location is measured and is used to interpolate the temperature at each gage location. M. has increased RTD measurement accuracy beyond 0.5°F through weekly calibrations under known isothermal conditions. The calibration procedure allows the calculation of correction factors that, when applied to the RTDs measurements, provide a temperature measurement accuracy of +/- 0.05°F.

The data acquisition units employ a four-wire measurement technique to compensate for any line resistance changes. Resistances are measured using a high impedance digital meter to six decimal-place accuracy. Novel digital filtering attenuates the effects of mechanical vibrations from the conditioning chamber and 60 Hz noise from the electrical system.

In M. terminology, a matched set of strain gages means the resistances of a group of individual gages fall within a given tolerance over a set of specified temperatures. M. typically matches gage resistances at three specific temperatures covering 0°F, 78°F and 278°F. Since a temperature gradient exists in the conditioning chamber and across the gage test board at each of the nominal conditioning temperatures of 0°F, 78°F and 278°F, the measured resistance for each gage is temperature corrected to 0°F, 78°F and 278°F using the interpolated RTD temperatures and the measured temperature coefficient of resistance for each gage.

Thus, the gage test system shown in Figure 2, containing the data acquisition units, conditioning chamber, baffling, gage test boards and programming, is a result of 25 years of effort to yield a test system that is capable of measuring the temperature at each gage location to better than 0.1°F accuracy and each gage resistance to better than 0.01%.

For the customer, properly bonding M. strain gages that are tightly controlled in terms of geometry and closely matched in terms of resistance will result in a sensor product that will have a static error band of less than 0.05 % full scale with almost no long-term drift and be operable for billions of full-scale cycles with no change in characteristics.

The demand for force, pressure or torque sensors capable of wireless transmission is increasing. As a result, there is a need for higher resistance strain gages to reduce the power demand of the sensor. M. is well-positioned, with its extensive experience in semiconductor gages and its unique gage test system, to design and manufacture matched sets of high resistance gages to meet customer size, resistance, temperature and power requirements.