A Surface-micromachined Shear Stress Imager

A new MEMS shear stress sensor imager has been developed and its capability of imaging surface shear stress distribution has been demonstrated. The imager consists of multi-rows of vacuum-insulated shear stress sensors with a 300 μm pitch. This small spacing allows it to detect surface flow patterns that could not be directly measured before. The high frequency response (30 kHz) of the sensor under constant temperature bias mode also allows it to be used in high Reynolds number turbulent flow studies. The measurement results in a fully developed turbulent flow agree well with the numerical and experimental results previously published. INTRODUCTION The detection and control of turbulent flows have long been a dream of fluid mechanists because of its potential impact on aerospace industry [1-3]. Up to now, however, little has been done mainly because the unavailability of miniature devices whose sizes are comparable to the feature sizes in high-Reynolds-number turbulent flows. It is our goal to first demonstrate that distributed MEMS sensors and actuators can accomplish active drag reduction. To do this, the first task is to be able to collect, in real time, the information of distributed surface shear stress. This information will then be processed and passed to MEMS actuators to control and reduce the drag. Therefore, the first challenge to achieve the real-time detection of surface shear stress is to build a high-resolution MEMS shear stress imager. In this paper, we report the development of such an imager and the results of its field tests in turbulent flows. SKIN FRICTIONS AND VORTICES The basic understanding of turbulent flow is important for the proper design of the shear stress imager. When a flat plate is placed in a moving flow, it is subject to a skin friction drag. The drag is the integration of surface shear stress τw, which is proportional to the flow velocity gradient near the surface, i.e., where u is the streamwise fluid velocity, y is the axis normal to the surface, μ is the fluid viscosity. High skin friction drag has recently been linked to organized structures in turbulent flows. The high drag region is commonly observed near streamwise counter-rotating vortex pairs (Fig. 1). These vortices, which τw μ u ∂ y ∂ ----= appear randomly in both space and time, bring high velocity fluids down to the walls and create local regions of high shear stress which significantly contributes to the total drag. Therefore, attempts to reduce drag have been focused on methods of either preventing the formation or mitigating the strength of these vortices. The statistical size of the drag-inducing vortex pair streak decreases as the Reynolds number of the flow increases. For a typical airflow of 15 m/s in wind tunnel, the Reynolds number is about 104 and the vortex streaks have a mean width of about 1 mm. The length of a typical vortex streak can be about 2 cm, giving the streaks a 20:1 aspect ratio. The frequency of appearance of the streaks is approximately 100 Hz. The life-time is around 1 ms [4]. SHEAR STRESS SENSORS Many ways exist to do wall shear stress measurement [5]. Among them, the thermal method, which uses hot film sensors to determine shear stress indirectly, has many advantages over other techniques for real time flow measurement and control. For example, it can achieve high sensitivity while keeping the sensor size small. The traditional hot film sensors are thin metal film resistors on substrates, which is electrically heated in operation. Since only the heat convection responds to the shear stress change, it is desirable to thermally isolate the thin film resistor from the substrate to minimize the conductive heat loss, thus increasing the sensitivity. In the past, the only way to partially solve the problem was to use low thermal conductivity materials such as quartz for the substrate. Reasonable good sensitivity could be obtained only when such sensors are used to measure high thermal conductivity fluid such as water. However, they are not sensitive enough for the measurement in low thermal conductivity fluids such as air. Moreover, the size of traditional sensors is typically in the mm range [5]. This may be tolerable in measuring the mean value of shear stress, but is definitely not acceptable in shear stress imaging with reasonable spatial resolution. Thanks to the development of surface micromachining technology, we can optimize both the materials and the structure of the sensors. Fig. 2 shows the cross-sectional structures of a few types of the micromachined shear stress sensors. Type I features a 2 μm deep vacuum cavity with a 0.25 μm thick polysilicon wire embedded in the nitride diaphragm. Here the vacuum cavity is designed for good thermal isolation of the diaphragm from the substrate [6]. Type II has a similar structure to type I except that the polysilicon wire is lifted 4 μm above the diaphragm, thus achieving better thermal isolation. Type III is a conventional polysilicon bridge sitting on the solid substrate [7]. Type IV is basically a micromachined hot wire close to a wall as has been previously reported. The wire is a few microns above the substrate surface and is in the linear velocity distribution region so that it measures the wall shear stress instead of velocity [8]. All four types were fabricated on a single chip to ensure identical thermal and electrical properties of the sensor materials. Fig. 3 shows their calibration results in wind-tunnel. The output changes are proportional to the one third power of shear stress, which agrees with the heat transfer theory [5]. It is obvious that types I and II are the most Fig. 1 Counter-rotating vortex pair. counter-rotating vortices spanwise direction streamwise direction (into the paper) high shear stress region y ratio polysilicon resistor (i. e. larger l/w, where l is the length and w the width). But the diaphragm can not be too thin as it would otherwise break during fabrication or operation. l is limited by the size of the whole device, which is at most 300 μm for the application in shear stress imaging. Therefore, l is chosen to be 150 μm. Also w is limited by the photolithography and etching technology. In this case, it is designed to be 3 μm to ensure good uniformity. SHEAR STRESS IMAGER Fig. 4 shows the photomicrograph of the 2.85 cm x 1.0 cm imaging chip using type I shear stress sensor. It is specifically designed for the study in the turbulent flow with Reynolds number near 104.There are two identical sensor rows 5 mm sensitive ones. Moreover, type I has much simpler fabrication process than type II, it is therefore chosen as the building block of this generation of shear stress imaging chip. After the structure is decided, the geometry of each layer is optimized to give maximum sensitivity. It is found that the sensitivity is higher for thinner diaphragm and larger aspect Fig. 3 Wind-tunnel calibration curves of sensors. 0.4 0.5 0.6 0.7 0.8 τ1/3 (Pa1/3) 0 0.2 0.4 0.6 0.8 R e la tiv e O u tp u t V o lta g e C h a n g e ( % ) type I: Vo = 11.5 V type II: Vo = 7.7 V type III: Vo = 8.7 V type IV: Vo = 8.1 V polysilicon wire vacuum cavity aluminum Si substrate nitride 12345678901234567890123456789012123456789012345678901234567890121234567890123456789 12345678901234567890123456789012123456789012345678901234567890121234567890123456789 12345678901234567890123456789012123456789012345678901234567890121234567890123456789 12345678901234567890123456789012123456789012345678901234567890121234567890123456789 12345678901234567890123456 12345678901234567890123456 12345678901234567890123456 12345678901234567890123456 12345678901234567890123456 1234 78901234567890123456 12345 78901234567890123456 12345678901234567890123456 12345678901234567890123456 12345678901234567890123456 (a) Type I: Hot film. 12345678901234567890123456789012123456789012345678901234567890121234567890123456789