Open Access
Add to BibSonomy
Abstract
Smart windows for sunlight control play an important role in modern green buildings. Electrically-controllable light microshutters provide a promising solution for smart windows. However, most of reported microshutters work under on/off binary mode. In this work, an electrothermally actuated microshutter that can achieve analog light control is proposed. The microshutter consists of an array of electrothermal Al/SiO2 bimorph cantilever plates suspended over a through-silicon cavity. The device is fabricated by a combination of surface- and bulk- micromachining processes. Test experiments show that for a single microshutter pixel, the device opening ratio can be tuned continuously from 78.6% (Open state, 0 V) all the way down to nearly 0% (Close state, 8 V) with a small hysteresis. For the entire array of 2 × 5 microshutters, the light transmission ratio varies continuously from 63.3% to 3.6% when the applied voltage is increased from 0 to 7.3 V. Furthermore, the response time, long-term reliability and window-like function of the microshutter are tested.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
The past decade has seen the rapid development of many sustainable designs for the purpose of slowing global warming and saving energy [1]. Among these designs, smart windows in “green buildings” for better utilization of natural sunlight are attracting researchers’ attention [2–5]. When activated, a smart window can be changed from transparent to translucent or opaque through blocking sunlight partially or completely. Smart windows can be achieved based on various principles, such as photochromic effect [6], thermochromic effect [7], electrochromic effect [8,9], liquid crystals [10] and suspended particles [11]. However, several particular problems, including switching speed, long-term stability, and particle settling, are still limiting wide commercial applications of these systems. In recent years, microshutters based on microelectromechanical system (MEMS) technology have been explored as a promising solution for sunlight control. MEMS microshutters with various actuation mechanisms have been reported [12–17]. Electrostatic actuation is the most popular choice. For instance, Pizzi et al. reported the first microshutter structure as early as in 1999 [12]. Without any external forces, the free-end of the metallic petal curled up due to residue stresses, and when the applied voltage was 80V, the metallic petal became flat onto the optically transparent substrate, forming a binary open/close mode response. Mori et al. [13] also proposed an electrostatically actuated microshutter array for house energy management applications, where transparent substrate and electrodes (indium tin oxide, ITO) were employed. Optical transmission measurements showed that the voltages for pull-in (close) and release (open) were 55 V and 38 V, respectively, and the corresponding optical transmittance were about 36% and 53%, respectively. Hillmer et al. [14] also proposed a microshutter array with a driving voltage of 40 V∼100 V. However, these devices can only achieve binary control of sunlight (ON/OFF state), and the driving voltages are high. Mott et al. reported an electromagnetically actuated microshutter that obtained similar characteristics with much lower voltage [15], but it was still a binary mode. Several attempts have been made for achieving analog light control with specially-designed microshutters [16,17]. For instance, Takahashi et al. [17] demonstrated analog light control by employing a vertical electrostatic electrode, but the required voltage was more than 200 V, and an additional latch structure must be used to keep the microshutter plate horizontal as the initial state. In addition to the analog light control issues, there exists a severe hysteresis between the opening and closing actuations in almost all of these microshutters discussed above. For example, in [13], the difference between the pull-in voltage and the release voltage was 16 V, while in [17], the hysteretic voltage difference was as much as 200 V.
On the other hand, electrothermal bimorph actuators are known for their unique advantages of large actuation range and low driving voltage [18–22]. They have been widely used in making MEMS micromirrors for various applications, such as optical phased arrays [18,19], endoscopic optical coherence tomography [20], Fourier transform spectrometers [21] and microlens scanner [22]. Another unique feature of electrothermal bimorph actuation is its large actuation range of quasi-static, meaning the actuation displacement is continuously tunable and controllable in the entire actuation range. This is exactly what is needed for realizing an analog light control of microshutters for smart window applications. Therefore, in this work, we propose a unique electrothermal bimorph cantilever plate design that bends over 90° at rest (open) and can be driven to stay flat (close) at low voltage. Microshutters composed of an array of such bimorph plates have been fabricated and tested. Experimental results show that continuous analog light control is achieved with a maximum driving voltage of only 8 V and the hysteresis of the response curve is as small as 0.15 V, indicating a great potential of this electrothermally actuated microshutter for smart window applications.
This paper is organized as follows. Section 2 describes the design, theoretical analysis and finite element analysis (FEA) of the proposed microshutter. Section 3 gives the fabrication processes and fabrication results. The experimental results and related discussions are presented in Section 4.
2. Device design and theoretical analysis
2.1 Concept of the smart window based on electrothermal microshutters
The concept of the proposed analog-controlled microshutter panel for smart windows is sketched in Fig. 1, where the panel consists of a two-dimensional (2D) array of microshutter pixels and a supporting frame. Each microshutter pixel is composed of a cantilever plate over a cavity. Each cantilever plate can curl up, allowing the light to pass through, i.e., the “Open” state (Fig. 1(a)), or becomes flat, blocking the light, i.e., the “Close” state (Fig. 1(b)). Each cantilever plate can also be electrically controlled at any arbitrary position in between the “Open” and “Close” states, i.e., the “Dimming” state (Fig. 1(c)) and the mixed state (Fig. 1(d)). Thus, analog control of light transmission is realized, which can meet the requirement of different light for different areas, as illustrated in Fig. 1(e). In this meeting room, a dimming area is urgently needed for reading and working to avoid glare effects, thus microshutters at the Dimming state are used for these areas. Part of microshutters is actuated at the Open state in the right window as the plant needs a bright area to benefit from intense sunlight. Other complicated functions, such as programmable image display, can be achieved by microshutters at more than one state.
Fig. 1. The concept of the proposed analog-controlled optical microshutter panel for smart windows. (a) Microshutter at the “Open” state (Just represents one case of the released microshutter. The cantilever may be under-curled, over-curled or roll up more than one turn under different residual stresses). (b) Microshutter at the “Close” state. (c) Microshutter at the “Dimming” state. (d) Microshutters at more than one state. (e) Light control in a meeting room.
To implement this idea, we use a silicon substrate as the supporting frame and electro-thermal bimorph plates as the deformable cantilever plates, as shown in Fig. 2. Different from electrostatically actuated microshutters, which typically use transparent electrodes and glass substrate for optical transparency, the microshutter proposed in this work is simply made on a silicon substrate, which has a through-silicon cavity under the bimorph to achieve light transmission. The bimorph mainly consists of two layers that are made of materials with different coefficients of thermal expansion (CTEs). In this work, aluminum (Al) with a relatively high CTE (23.1×10−6/K) and silicon dioxide (SiO2) with a relatively low CTE (0.5×10−6/K) are used. One end of the bimorph is fixed on the silicon substrate while the other end is free. At room temperature, due to the existence of residual stresses in the two layers, the bimorph has an initial curl (Fig. 2(a)). The curl is proportional to the length of the bimorph. With a proper length, the bimorph can form a half circle. When the temperature rises, the CTE difference of the two layers increases the radius of curvature of the bimorph, resulting in the stretching of the curled bimorph (Fig. 2(b)). When the temperature rises to certain value, the bimorph becomes flat (Fig. 2(c)).
Fig. 2. Cross-sectional views of the proposed deformable cantilever plate design based on electrothermal bimorphs. (a) “Open” state. (b) “Dimming” state. (c) “Close” state.
The bimorph temperature can be changed by varying the voltage applied to the embedded heater. Titanium (Ti) is used as the resistive heater material in this work. Note that the Ti heater must be properly designed for the purpose of achieving a uniform temperature distribution on the entire bimorph.
The response of the microshutter to the applied voltage, V, is also illustrated in Fig. 2, where it is assumed that the light comes from the top side of the microshutter. The response of the microshutter can be divided into three states: Open state (Fig. 2(a)), Dimming state (Fig. 2(b)), and Close state (Fig. 2(c)). At V = 0, the bimorph is curled the most and the optical opening is the largest, allowing the light to pass through (Fig. 2(a)). When V increases, the temperature of the bimorph increases, causing the bimorph less curled. Consequently, the optical opening becomes smaller and thus a portion of the light is blocked, i.e., the incoming light is dimmed (Fig. 2(b)). The voltage V can be continuously tuned, so can the light transmittance. Therefore, analog light control is realized. Eventually, the through-silicon cavity will be fully shaded by the bimorph when the applied voltage increases to VOFF, at which the bimorph becomes flat (Fig. 2(c)).
2.2 Theoretical analysis and design guideline of the microshutter
According to the above discussion, the degree of the optical opening is determined by the radius of curvature of the bimorph which is a function of the temperature of the bimorph. The bimorph temperature can be controlled by the voltage applied to the embedded resistive heater. The temperature change induced by the resistive Joule heating is readily given by [23]
Assume that the bimorph curls uniformly along its entire length. Then the tilt angle at the bimorph tip, θtip, can be written as
where lb is the length of the bimorph. The relationship between the optical opening and the radius of curvature is illustrated in Fig. 3. It can be seen that the estimation of the optical opening can be divided into two situations. In the first situation, the light is blocked by part of the backside of the curled bimorph, as shown in Fig. 3(a). In the second situation, the light is blocked directly by the frontside of the curled bimorph, as depicted in Fig. 3(b). The dividing point of these two situations is at θtip =π/2. So, the opening ratio, ηo, defined as the ratio between the optical opening and the bimorph length, can be expressed asFig. 3. Relationship between the optical opening and radius of curvature of the bimorph. (a) θtip ≥π/2, (b) θtip < π/2.
The optical opening can be calculated by plugging Eqs. (1)–(3) into Eq. (4), but the material properties and dimensions of the bimorph must be known. The material properties are listed in Table 1. As Table 1 indicates, the Young’s moduli of Al and SiO2 are approximately equal, so the thicknesses of the Al and SiO2 bimorph layers are set to be equal, leading to the maximum β of 1.5. Note that the bimorph width does not appear in Eqs. (1)–(4), but it determines the microshutter pixel size and also the resistance of the Ti resistor embedded in the bimorph. In this work, the bimorph width is chosen as 600 μm. Meanwhile, RE0 is set to be in the range of 300-400 Ω and the thermal resistance RT is designed to be about 3,300 K/W. With this information, now we can study the dependence of the optical opening ratio on the bimorph thickness and length at various applied voltages.
Table 1. The properties of the employed materials.
Firstly, the bimorph length varies from 500 μm to 1000 μm. Based on Eqs. (1)–(4), the corresponding ηo-V curves are plotted in Fig. 4(a), showing that a “Close” state can be reached by applying 8 V. Note that the “Close” voltage is approximately the same for different lengths because the heaters and the thermal isolation structures are the same. It also shows the initial opening ratio is higher at a larger bimorph length. However, the increase of the bimorph length reduces the resonant frequency of the microshutter, which may cause reliability problems. Thus, 700 μm is chosen as a reasonable trade-off for the bimorph length in this work.
Fig. 4. Analytical results of the opening ratios vs. applied voltage at (a). different bimorph length and at (b). different bimorph thicknesses, respectively.
To further increase the opening ratio, the thickness of the bimorph, which is twice the thickness of Al layer or SiO2 layer, is then investigated. By setting the bimorph length at 700 μm, the ηo-V curves for the bimorph thicknesses of 0.8 μm, 1.6 μm, and 2.4 μm are plotted in Fig. 4(b), showing that the initial opening ratio can reach 88% when the bimorph thickness is 0.8 μm. However, the smaller the bimorph thickness, the steeper the slope of the curve becomes, which may complicate the analog light control. The ultimate goal of the proposed device is to achieve dynamic lighting systems with high resolution pixels, multi-gray-level displays and projections, and light communication systems. Therefore, we hope that the output response, or the light transmission, of the device shows a good linearity and a small curve slope for a given voltage accuracy because of the analog controllability. However, due to the large actuation range of the bimorph, the light control process is divided into two regions, i.e., one before and one after θtip =π/2. For the first region, the light transmission of the device changes very slowly. In the second region, the light transmission changes faster with the voltage, as Fig. 4(a) shows. The voltage accuracy we used here is 0.05 V. If we want to control the light transmission down to 2%, then the threshold slope of the response curve is 2% / 0.05V, or 40%/V. Based on this threshold value, the optimal thickness of the bimorph used in the microshutters is selected as 1.6 μm, that is, the thicknesses of the Al and SiO2 layers both are 0.8 μm.
The final design parameter values of the microshutter are listed in Table 2. Here we choose the typical stress levels for the Al layer (a tensile stress of about 230 MPa) and the SiO2 layer (a compressive stress of about -260 MPa). So the residual stress difference of the two layers is about 490 MPa, which leads to an over 70% opening ratio of the microshutter at the open state. Actual residual stresses of thin films are determined by various factors, including the fluctuations of the chamber pressure and gas flow, grain coalescence, and impurity incorporation [27]. Thus, the process parameters can be experimentally optimized during the deposition to achieve a proper residual stress level. The corresponding ηo-V response is the middle curve with triangle symbols in Fig. 4(b), showing that ηo is about 77% initially and decreases monotonically with increasing voltage. The transition point corresponding to θtip =π/2 is at 4.9 V. ηo decreases relatively slowly before this point and fall much faster after this point.
Table 2. Optimized key parameters of the designed microshutter.
Note that this work is focused on the proof of concept of this electrothermal bimorph microshutter design. Thus, only single microshutter pixels and a small-scale microshutter array composed of 2 × 5 bimorph plates are studied. Also, as each array is a periodic structure, a grating is formed. According to the grating diffraction equation, $d\sin {\theta _m} = m\lambda$, considering the peak wavelength of the sunlight at 550nm and the large grating constant (d) of 780 μm, the angle of the first-order diffraction is only 0.04°. Even the angle of the 5th-order diffraction, which becomes very weak already, is still only 0.2°, which is hardly distinguishable with our eyes. Thus, this grating diffraction effect is negligible.
COMSOL, a Multiphysics software [28], is used to find the resonant modes of one bimorph plate and the simulation results are shown in Fig. 5. The first resonant frequency is 3.036 kHz, which corresponds to a longitudinal bending mode (see Fig. 5(a)). The second resonant frequency is 8.943 kHz, which is a transverse bending mode (see Fig. 5(b)). These results indicate that the designed device can be used in a broad range of devices/systems in motion. For motionless smart windows, there will be a large design space, such as decreasing the bimorph thickness and/or increasing the bimorph length. Note that the higher modes have resonant frequencies greater than 17 kHz, which will have little effects on most applications.
Fig. 5. FEA simulation of first two resonance modes of the bimorph plate. (a) First mode. (b) Second mode.
3. Device fabrication
The microshutter is fabricated on a 6-inch SOI wafer, where the handle layer and device layer thicknesses are 550 μm and 25 μm, respectively. The thickness of the buried oxide (BOX) layer is 1 μm, which will be used as the etch-stop of the backside deep reactive ion etch (DRIE) process. The cross-sectional view of the fabrication process flow is outlined in Fig. 6. A 0.4 μm-thick SiO2 layer is firstly deposited through plasma enhanced chemical vapor deposition (PECVD) (Fig. 6(a)), followed by a 0.15 μm-thick Ti sputter deposition and patterning (Fig. 6(b)). Then another 0.4 μm-thick SiO2 layer is deposited by PECVD and vias are formed by HF etching (Fig. 6(c)). After that, a 0.8 μm-thick Al layer is deposited and patterned (Fig. 6(d)). Next, a silicon cavity is etched down to the BOX layer from the backside of the SOI wafer by DRIE. Finally, the BOX layer is removed (Fig. 6(e)), followed by a silicon isotropic etch of the 25 μm-thick device layer to release the bimorph structure (Fig. 6(f)).
Figure 7(a) shows a scanning electron microscope (SEM) image of a released 2 × 5 microshutter array, where all the bimorph plates are curled up and their tip angles are over 150°. Figure 7(b) is a side-view SEM, where slightly different curvatures of the bimorph plates are visible. The non-uniformity of the curvatures of the bimorphs is believed to be mainly caused by the release process. During the release, the 25 μm-thick silicon device layer under the bimorph beams is removed by isotropic etching using Xenon difluoride (XeF2). XeF2 silicon etching has very high selectivity, but the etching front is not smooth. Also, the etching process runs with pulsed XeF2 gas, which may generate some non-uniform distributions of both the gas flow rate and concentration that in turn lead to etching non-uniformity. Further process optimization must be taken to better control the uniformity of the released bimorphs. Figure 7(c) shows an SEM image of the electrical wiring.
Fig. 7. SEMs of fabricated microshutters. (a) A released 2×5 microshutter array; (b) cross-sectional view of the released microshutter array; (c) electrical wire structure.
4. Device characterization and discussion
4.1 Characterization of a single microshutter pixel
4.1.1 Observation of the bimorph plate during the Open-to-Close transition
Figure 8 shows the step-by-step microscopic pictures of the quasi-static response of a single microshutter pixel, where two pixels are present with only the one on the right is activated. The voltage applied to the activated pixel varies from 1 V (corresponding to Fig. 8(a)) to 8 V (corresponding to Fig. 8(e)). It can be seen that the radius of curvature of the activated pixel increases with the increasing voltage. The “Close” state of this pixel is reached at 8 V, as shown in Fig. 8(e), in which the bimorph plate is nearly flat. Also note that a small bending along the width direction of the bimorph plate is observed at this state (Fig. 8(e)), which is due to the uneven temperature distribution on the bimorph. This issue can be solved by optimizing the embedded heater design.
The resonant frequency of the fabricated microshutter is also measured under the optical microscope. By sweeping the frequency of an AC voltage applied to the Ti heater, the first resonance mode is observed at 3.457 kHz, which is in good agreement with the simulated 3.036 kHz. The discrepancy is believed to be mainly caused by the layer thickness variations from microfabrication.
4.1.2 Light transmission experiment
The light transmission performance, or the opening ratio, of a single microshutter pixel was measured using an optical microscope with a back light source and an image-viewing system. The experimental setup is sketched in Fig. 9. The microshutter under test is placed on the microscope stage and a DC voltage is applied to the heater of one of the microshutter pixels. Two multimeters are used to measure both the current passing through and the voltage crossing the resistor for calculating the power consumption of a single microshutter pixel. A light beam is incident on the microshutter from its back, so the opening region of the microshutter is bright (see the inset in Fig. 9), which can be viewed and recorded by the image-viewing system. To check the uniformity of the fabricated devices, more than three single microshutter pixels have been tested individually in this experiment. The light transmission ratio of a microshutter pixel, ηpixel, is simply the ratio of the length of the bright region, Lbright, and the full length of the bimorph plate, lb, which is 700 μm as defined in Section 3, i.e.,
Figure 10 plots light transmission ratios of three microshutter pixels from one 2×5 array as a function of the applied voltage, where the microscope-captured images of the “Open” state, one “Dimming” state and the “Close” state are inserted at their corresponding positions. It can be seen that the three microshutters respond similarly, and the differences of their initial positions are less than 3% and the maximum deviation among the three response curves is also less than 3%. When the applied voltage is 0 V, corresponding to the Open state, ηpixel is 78.6%. This maximum opening ratio is comparable with those of electrostatically actuated microshutters [13]. Analog light control can be achieved simply by electrothermal actuation, i.e., ηpixel can be continuously varied between 78.6% and near 0% by tuning the applied voltage. For example, ηpixel is 69.9% at 4 V and decreases to 31.8% at 6.4 V. It should be noticed that the ηpixel-V relationship is nonlinear. The opening ratio drops from 78.6% to only 62.9% for the first 5 V, with a slope of 3.1%/V, and then it drops quickly from 62.9% to nearly 0% for the last 3 V, with a much larger slope of 21%/V. The voltage to achieve the Close state, VOFF, is about 8 V, which is much lower than those of electrostatically actuated microshutters (> 50 V) [13]. Also, the analytical solution is re-plotted in Fig. 10 as a reference, showing a good agreement between the theoretical prediction and the experimental results. The analytical model can provide a guideline for designing new devices.
Fig. 10. Opening ratio vs. applied voltage for three single microshutter pixels. The dashed black curve is the prediction of the analytical model. Measurement errors of the opening ratio: ±0.8%.
4.1.3 Light switching hysteresis
A bidirectional open/close test has also been performed on a single microshutter pixel. The applied voltage is firstly increased from 0 V to 8 V, and then it is immediately decreased from 8 V to 0 V. During this process, the opening ratio is recorded. The bidirectional ηpixel-V curves are plotted in Fig. 11, where only a small hysteresis is present with a maximum value of only 0.15 V or less than 2% over the entire range, which is much smaller than those of electrostatically actuated microshutters [13]. This small hysteresis property greatly simplifies the analog light control of electrothermally actuated microshutters.
4.1.4 Electrical characteristics
Due to the Joule heating effect and the positive TCR, the electrical resistances of the embedded Ti heaters increase significantly with the applied voltage. To characterize the electrothermal performances of the proposed microshutter, the resistances of the Ti heaters and the electrical currents under various voltages are measured. The measured resistances of the microshutter pixels #1 to #3 in Fig. 10 at V = 0 are about 346 Ω. The relationship between the resistance and the voltage is plotted in Fig. 12(a) for each of the three microshutter pixels, showing a near-linear relation for voltage greater than 3 V. The resistances increase to about 620 Ω at 8 V. To measure the actual TCR of the sputtered Ti layers, three on-chip Ti resistors (R1, R2, and R3) are designed with different geometric dimensions and arranged around the microshutters. The chip with the three resistors is heated by a hot plate with the temperature range from 25.8 °C to 220 °C. The experimental data is plotted in Fig. 12(b), where the linear fitting equations are marked beside the results. The TCR of each resistance can be calculated as the ratio of the slope to the R-axis intercept of its R-V curve. The average TCR of the resistances is 4.4×10−3/°C.
Fig. 12. (a) The relationship between the resistance and the voltage. (b) The relationship between the resistances and the temperature. Measurement errors of the resistance: ±1 Ω.
By combining the results plotted in Fig. 12(a) and the calculated TCR of the Ti resistances, the electrothermal characteristics, including ηpixel versus temperature and ηpixel versus power, of the microshutter pixel #1 are plotted in Fig. 13. As shown in Fig. 13(a), ηpixel decreases with increasing temperature and reaches near 0%, i.e., the Close state, when the temperature is 200 °C. Note that the high temperature is on the bimorph plates only. Typically, the microshutter array is packaged between two glass plates, so users will have no direct contact to any bimorph plates. Also, it is interesting to notice that ηpixel drops only 5% when the temperature rises from 25 °C to 70 °C, indicating that the microshutter will not be affected much by the ambient temperature change. Even if the ambient temperature climbs to 100 °C, ηpixel is still maintained at about 66%.
Fig. 13. (a) Opening ratio vs. temperature. (b) Opening ratio vs. applied power. Measurement errors of the opening ratio: ±0.8%.
Meanwhile, as shown in Fig. 13(b), (a) maximum 104 mW power is needed to completely close the microshutter, and that is about 1 W for the entire 2×5 microshutter array. This power level will be not acceptable for large-area windows. Nevertheless, this work is focused on the proof of concept of this electrothermal bimorph microshutter design. More encouragingly, there are several ways to drastically decrease the power consumption, such as using bimorph materials with much greater CTE difference such as PVDF (127.8×10−6/K) and Si (2.6×10−6/K), increasing the thermal isolation, decreasing the bimorphs’ total thickness, or even integrating electrothermal bimorph actuation and electrostatic actuation. With a combination of these methods, the power consumption can be reduced by 2 to 3 orders of magnitude, making it possible to make a 1 m2 commonly-used window with a maximum power consumption within 100 W.
4.2 Characterization of microshutter array
4.2.1 Light transmission experiment
Light transmission of a microshutter array is measured through the testing system shown in Fig. 14. A 650 nm laser beam from a HeNe laser, which its beam diameter adjusted by an optical diaphragm, is incident on a microshutter array, the device under test (DUT), and the part of the laser passing through the DUT is picked up by a laser power meter (Mobiken LP1, Sanwa Supply Inc.).
The test experiment is conducted in a dark room to minimize the background noise. The test is done in two steps. In the first step, the DUT is a 2 × 5 microshutter array. In the second step, the DUT is simply an opaque plate with a hole that has exactly the shape and size as those of the microshutter array tested in the first step. The laser power detected in the second step is used as the reference, denoted as Pr. Thus, the light transmission ratio of a microshutter array, ηarray, can be written as
where P(V) is the laser power detected by the laser power meter when the DUT is a microshutter array and V is the voltage applied to the microshutter array. Figure 15 plots the ηarray vs. the applied voltage for the microshutter array, which exhibits a similar non-linear response to those of single microshutter pixels (see Fig. 10). However, if we compare the response curves in Fig. 10 and Fig. 15, we can see the following differences between single microshutter pixels and microshutter array:- (1) At the Open state, the light transmission ratio of the 2 × 5 microshutters array is about 63.3%, smaller than the 78.6% of the single microshutter pixels. This is because part of the light is blocked by the opaque silicon substrate that supports the microshutter array.
- (2) At the Close state, the light transmission ratio is about 3.6% for the microshutter array instead of near 0% for the single microshutter pixels. The greater light leakage from the microshutter array is due to the fact that there are air gaps between the adjacent pixels in the microshutter array. These gaps are necessary to ensure the independent tuning of each pixel. This light leakage at the Close state can be decreased by shrinking the gaps and/or adding opaque structures to block the light leakage from these gaps.
- (3) VOFF of the microshutter array is about 7.3 V, about 9% lower than that of the single microshutter pixels (8 V). This is due to the fact that the microshutter array has 10 pixels. That means the microshutter consumes 10 times as much power as that by one single microshutter pixel. The higher the power consumption, the higher the temperature around the device becomes, resulting in a decrease of the required voltage.
Fig. 15. Measured light transmission ratio versus applied voltage of the microshutter array. Measurement errors of the light transmission ratio: ±0.8%.
The light transmission hysteresis of the entire microshutter array is also measured using a bidirectional open/close test. The measured ηarray-V data is plotted in Fig. 16. It can be seen that the hysteresis remains at the same level (about 0.1 V) when the measurement is extended from a single pixel to a whole 2×5 microshutter array. The response time of the microshutter array is also measured using the same setup shown in Fig. 14 except that the optical power meter is replaced with a photodiode. The measured rise time and fall time of the microshutter are 14.0 ms and 15.2 ms, respectively. Also, the window-like light control function of the microshutter array is demonstrated by transmitting a light from the back of the 2 × 5 microshutter array, and the results are shown in Fig. 17.
Fig. 16. The measured hysteresis loop of the ηarray-V curve of a whole microshutter array. Measurement errors of the opening ratio: ±0.8%.
Fig. 17. Window-like light control function of the microshutter array. (a) Open state. (b) Close state.
Also, as a smart window, the viewing angle matters as well. When the light is incident at an oblique angle from the convex side of the bimorph, the light will be partially reflected by the backside of the bimorphs, partially blocked by the sidewalls of the cavities, and partially transmitted through the silicon cavity, as shown in Fig. 18(a). The light transmission ratios at the open state with the light incident angles at 0°, 15°, 30°, and 45° from the convex side of the bimorphs are measured, and the resultant ηarray-V curves are plotted in Fig. 18(b). Clearly, at the open state, the light transmission ratio drops from nearly 65% to nearly 6% when the viewing angle increases from 0° to 45°. And all the ηarray-V curves show a similar ‘slow-fast-steady change’ tendency. The cases with the light incident from the concave side of the bimorphs will lead to even lower transmission ratios. Also, the mass of a microshutter pixel is only 1.65 μg. With the stiffness of each bimorph beam is in the order of 1 m/N, the maximum deformation of the bimorphs due to gravity is less than 15 nm or less than 0.003% of the entire bimorph length, which means the light transmission ratios of the open state, dimming state and close state will not be affected by gravity in any orientations of the smart window with such microshutters embedded.
Fig. 18. Measured ηarray vs. V curves at various viewing angles. (a) Cross-sectional view of the microshutter when the light is incident obliquely. (b) ηarray-V curves at different viewing angles.
4.2.2 Reliability experiment
The long-term reliability of the microshutter is firstly evaluated by actuating a single microshutter pixel over one million cycles with a 5 Hz, 0-8 V square-wave voltage signal. After the long-term test, the pixel looks intact and functions normally. The ηpixel-V response curves before and after the long-term test are plotted in Fig. 19(a). It can be seen that the initial opening ratio drops from 79.0% to 75.7% and the response curve has a small left shift, which is believed to be caused by the stress relaxation and redistribution in the bimorph plate of the microshutter pixel. Then, the microshutter pixel is actuated to the Close state, and the applied voltage was 8 Vdc. The ηpixel-V response curves before and after 12-hours and 36-hours running are plotted in Fig. 19(b). It can be seen that the initial opening ratio drops from about 79% to about 77% and the response curve also has a small left shift. Note that the response curve after 36 hours does not shift left further but almost overlaps with the curve after 12 hours, which is believed to be that the device becomes stable after the 12 hours’ burn-in effect.
Fig. 19. Long-term reliability test result. (a) 1 million cycles driving. (b) Long-time driving. Measurement errors of the opening ratio: ±0.8%.
The impact of the air conditions must be considered as well. The change of the air temperature affects the temperature on the bimorph, and thus, as analyzed in Section 2.2, the opening ratio will decrease when the ambient temperature increases. Since the ambient air temperature is usually well below 70 °C, it can be seen from Fig. 13(a) that ηpixel drops only 5% even when the temperature rises from 25 °C to 70 °C. Therefore, the opening ratio changes very slowly when the air temperature varies, making it insensitive to any ambient temperature changes. Meanwhile, the influence of the air speed change on the dynamic thermal behaviors of bimorphs has been investigated in [29]. The results indicate that the heat conduction dominates the heat transfer process of electrothermal bimorph actuators, while the convection has little impact on the overall heat transfer even under strong forced convection with a mini fan. So the operation angle, or the opening ratio of the microshutter, will not significantly be impacted by the air temperature and air speed. And in order to avoid the potential impact of these two factors on the response time, the microshutter is typically packaged between two glass plates and installed on the indoor side. More study will be carried out.
4.3 Comparison with previously reported microshutters
Microshutters based on different actuation methods have been reported previously. A comparison of the device proposed in this paper to those previously reported microshutters is summarized in Table 3.
Table 3. Characteristics comparison with previously published microshutters.
5. Conclusion
The proposed analog-controlled light MEMS microshutter based on electrothermal actuation has been successfully fabricated with a simple micromachining process. Experiments have shown that light transmission ratio can be smoothly controlled between about 80% and near 0% by varying the driving voltage between 0 and 8 V. It has also been experimentally verified that the MEMS microshutter has low hysteresis (0.15 V) and high reliability. Also an analytical electrothermal actuation model has been developed to predict the behavior of the proposed device and the prediction matches well with the experimental results. Although only a small 2 × 5 array is demonstrated, this microshutter design can be easily extended to large arrays. Therefore, this unique analog-control microshutter design brings in a bright future for smart windows. This microshutter design also has potential applications such as dynamic lighting systems with high resolution pixels, multi-gray-level displays and projections, and light communication systems. Note that currently the maximum power needed to completely turn off one microshutter pixel is about 100 mW, which is not suitable for practical use, but the power consumption can be reduced by orders of magnitude through using much thinner bimorph layers and much larger thermal isolation, Future work will focus on the array size scaling and the power consumption reduction. Stress relaxation in the bimorph plate will be investigated as well in the near future.
Funding
National Natural Science Foundation of China (62074015, 61774015, 61574016); the 111 Project of China (B14010); National Science and Technology Planning Project (2018YFF01010904).
Disclosures
The authors declare no conflicts of interest.
References
1. S. Sorrell, “Reducing energy demand: A review of issues, challenges and approaches,” Renewable Sustainable Energy Rev. 47, 74–82 (2015). [CrossRef]
2. G. A. Niklasson and C. G. Granqvist, “Electrochromics for smart windows: thin films of tungsten oxide and nickel oxide, and devices based on these,” J. Mater. Chem. 17(2), 127–156 (2007). [CrossRef]
3. C. Meng, E. Chen, L. Wang, S. Tang, M. Tseng, J. Guo, Y. Ye, Q. F. Yan, and H. Kwok, “Color-switchable liquid crystal smart window with multi-layered light guiding structures,” Opt. Express 27(9), 13098–13107 (2019). [CrossRef]
4. C. Meng, M. C. Tseng, S. T. Tang, C. X. Zhao, S. Y. Yeung, and H. S. Kwok, “Normally transparent smart window with haze enhancement via inhomogeneous alignment surface,” Liq. Cryst. 46(3), 484–491 (2019). [CrossRef]
5. J. Wang, C. Meng, Q. Gu, M. C. Tseng, S. T. Tang, H. S. Kwok, J. Cheng, and Y. Zi, “Normally Transparent Tribo-Induced Smart Window,” ACS Nano 14(3), 3630–3639 (2020). [CrossRef]
6. C. Goebbert and M. A. Aegerter, Sol-Gel Technologies for Glass Producers and Users (Springer, 2004).
7. A. Taylor, I. Parkin, N. Noor, C. Tummeltshammer, M. S. Brown, and I. Papakonstantinou, “A bioinspired solution for spectrally selective thermochromic VO2 coated intelligent glazing,” Opt. Express 21(S5), A750–A764 (2013). [CrossRef]
8. S. Bao, X. Zhang, P. Jin, and K. Yoshimura, “Optical properties and switching durability of TiO2 top-coated magnesium–nickel thin-film switchable mirrors,” Jpn. J. Appl. Phys. 54(4), 045501 (2015). [CrossRef]
9. J. Wang, L. Zhang, L. Yu, Z. Jiao, H. Xie, X. W. Lou, and X. W. Sun, “A bi-functional device for self-powered electrochromic window and self-rechargeable transparent battery applications,” Nat. Commun. 5(1), 4921 (2014). [CrossRef]
10. D. Cupelli, F. P. Nicoletta, S. Manfredi, M. Vivacqua, P. Formoso, and G. D. Filpo, “Self-adjusting smart windows based on polymer-dispersed liquid crystals,” Sol. Energy Mater. Sol. Cells 93(11), 2008–2012 (2009). [CrossRef]
11. R. Baetens, B. P. Jelle, and A. Gustavsen, “Properties, requirements and possibilities of smart windows for dynamic daylight and solar energy control in buildings: A state-of-the-art review,” Sol. Energy Mater. Sol. Cells 94(2), 87–105 (2010). [CrossRef]
12. M. Pizzi, V. Koniachkine, O. D. Martiis, and R. Marbach, “New type of micro-mechanical shutter array for replacement of infrared photodetector arrays,” in 2003 IEEE/LEOS International Conference on Optical MEMS (IEEE, 2003), pp.173–174.
13. K. Mori, K. Misawa, S. Ihida, T. Takahashi, H. Fujita, and H. Toshiyoshi, “A MEMS Electrostatic Roll-Up Window Shade Array for House Energy Management System,” IEEE Photonics Technol. Lett. 28(5), 593–596 (2016). [CrossRef]
14. H. Hillmer, B. Al-Qargholi, M. M. Khan, N. Worapattrakul, H. Wilke, C. Woidt, and A. Tatzel, “Optical MEMS-based micromirror arrays for active light steering in smart windows,” Jpn. J. Appl. Phys. 57(8S2), 08PA07 (2018). [CrossRef]
15. D. B. Mott, S. Aslam, K. A. Blumenstock, R. K. Fettig, D. Franz, A. S. Kutyrev, M. J. Li, C. J. Monroy, S. H. Moseley, and D. S. Schwinger, “Magnetically actuated microshutter arrays,” Proc. SPIE 4561, 163–170 (2001). [CrossRef]
16. W. A. de Groot and P. D. Floyd, “Switchable Windows With MEMS Shutters,” U.S. patent 8,724,202 B2 (13 May 2013).
17. T. Takahashi, M. Mita, K. Motohara, N. Kobayashi, H. Fujita, H. Toshiyoshi, and N. Kashikawa, “Electrostatically Addressable Visored Shutter Array by Electroplating for Astronomical Spectrography,” in 16th International Conference on Optical MEMS and Nanophotonics (OMN), (IEEE, 2011), pp. 79–80.
18. K. Jia, S. R. Samuelson, and H. Xie, “High-Fill-Factor Micromirror Array With Hidden Bimorph Actuators and Tip-Tilt-Piston Capability,” J. Microelectromech. Syst. 20(3), 573–582 (2011). [CrossRef]
19. J. Cheng, W. Liu, Q. Chen, N. Xu, Q. Sun, Y. Liu, W. Wang, and H. Xie, “A MEMS variable optical attenuator based on a vertical comb drive with self-elevated stators,” Sens. Actuators, A 271, 398–408 (2018). [CrossRef]
20. T. Xie, H. Xie, G. K. Fedder, and Y. Pan, “Endoscopic optical coherence tomography with a modified microelectromechanical systems mirror for detection of bladder cancers,” Appl. Opt. 42(31), 6422 (2003). [CrossRef]
21. F. Han, W. Wang, X. Zhang, and H. Xie, “Miniature Fourier transform spectrometer with a dual closed-loop controlled electrothermal micromirror,” Opt. Express 24(20), 22650–22660 (2016). [CrossRef]
22. L. Zhou, X. Yu, P. X.-L. Feng, J. Li, and H. Xie, “A MEMS lens scanner based on serpentine electrothermal bimorph actuators for large axial tuning,” Opt. Express 28(16), 23439–23453 (2020). [CrossRef]
23. S. T. Todd and H. Xie, “An electrothermomechanical lumped element model of an electrothermal bimorph actuator,” J. Microelectromech. Syst. 17(1), 213–225 (2008). [CrossRef]
24. H. Xie, S. Todd, A. Jain, and G. K. Fedder, “Single-Crystal Silicon Based Electrothermal MEMS Mirrors for Biomedical Imaging Applications,” in MEMS/NEMS Handbook Techniques and Applications (Springer, 2007), Chap. 3.
25. W. Peng, Z. Xiao, and K. R. Farmer, “Optimization of thermally actuated bimorph cantilevers for maximum deflection,” Nanotech 1, 376–379 (2003).
26. K. Jia, P. Sagnik, and H. Xie, “An Electrothermal Tip–Tilt–Piston Micromirror Based on Folded Dual S-Shaped Bimorphs,” J. Microelectromech. Syst. 18(5), 1004–1015 (2009). [CrossRef]
27. L. B. Freund and S. Suresh, Thin Film Material: Stress, Defect Formation and Surface Evolution (Cambridge Univ. Press, 2003).
28. . “COMSOL Multiphysics”, https://cn.comsol.com/
29. L. Zhou, Z. Chen, J. Cheng, Q. Chen, Y. Ding, and H. Xie, “Investigation of dynamic thermal behaviors of an electrothermal micromirror,” Sens. Actuators, A 263, 269–275 (2017). [CrossRef]
