3.1.

Temperature Test

Two different approaches were followed to analyse the impact of the temperature in the MMA technology presented in this study: (1) theoretical investigations on the mirror deflection curves by means of Finite Element Method (FEM) based simulations, and (2) experimental tests by modifying the environmental temperature inside a climate chamber.

Theoretical investigations

The deposition of different layers in surface micromachining and successive processing steps cause at least a residual stress in the metal alloy of the actuator hinges. That layer is responsible for the mirror movement, and this stress impacts directly on the deflection behaviour. This can be seen by a slight change of the natural frequency of torsion modes when changing intrinsic tensile stress. A temperature deviation causes the thermal expansion of the actuator hinges leading to a change of stress. Besides, there is a temperature dependency of the material properties like Young’s or shear modulus, although that can be considered as negligible in the temperature range of interest.

In order to better evaluate these effects, an analysis by FEM simulations was carried out. The electrostatic force at a certain pixel electrode voltage and the resulting deflection state were calculated by using the so-called transducer FEM-elements provided by ANSYS®. A schematic of substrate and hinge materials and their corresponding coefficients of thermal expansion α can be seen in Figure 4. Being E_hinge the Young’s modulus of the hinge, if the environment temperature changes, with ∆T = T – T_ref, then the thermal stress σ_th in the hinge bars can be estimated as follows:

Figure 4

Side view of a hinge layer on Si-substrate with corresponding coefficients of thermal expansion, α_si and α_hinge

Figure 5 depicts a characteristic curve for a single mirror of a 64k MMA and its temperature dependency. It indicates that the influence of the temperature on the electrostatic deflection is negligible over a wide range of voltages. The largest deviations occur at higher voltages: for instance, at 21 V, the tilt angle deviation is 0.03 ° or rather, 2 %, for ∆T values of -200 °C and 200 °C. This is because the torsion stiffness is affected by the stress along the torsion bar direction only to a minor extent, and consequently the electrostatic deflection state at temperatures different from Tref differs only slightly from the reference one.

Figure 5

Electrostatic deflection curves for different temperature variation ∆T

Temperature experimental tests

The DM250 C15 ESS climatic chamber from Angelantoni Test Technologies [20] was used to assess the performance of 64k-MMA in a temperature range from -40 °C to 80 °C at ambient pressure and ensuring a relative humidity < 10 %. The MMA device was covered with a non-hermetic glass window.

A MMA sample was placed in the thermal chamber and the chamber environmental conditions were monitored by means of a Graphtec GL820 data logger with an acquisition frequency of 0.1 Hz. In order to properly measure the temperature of the chamber and of the device, two external platinum resistance thermometers PT100 were employed, one glued to the ceramic frame of the device and the other to the chamber. Operational verification tests were performed stepwise (Figure 6) after a constant temperature had been measured in both sensors for several minutes, in this way guaranteeing a thermal stabilisation of the device tested at each temperature step. During the experiment, we opted for a simple and compact experimental set-up which, although not allowing for an optimal MMA illumination, provided contrast enough to identify single mirror defects, yet fitting inside the climate chamber. The industrial camera Imaging Source DMK 33UJ003 [21] with a CMOS sensor of 3856x2764 pixels, combined with a Xenoplan 1.9/35-0901 lens (f/1.9), was employed to acquire monochromatic images that detect the single-pixel state changes.

Figure 6

Test conditions and procedure for temperature check

To avoid damage of test equipment, two configurations were employed (Figure 7). For high temperatures, MMA and light source were located inside the thermal chamber, while the MMA electronic driver and CMOS camera were outside, with a direct line of sight through a lateral opening in the chamber. This arrangement prevented camera and electronics from operating outside their operational temperature range. On the contrary, for low temperatures the camera with thermal isolation was placed inside the thermal chamber, together with the SLM and a light source to prevent fogging effects in the chamber window.

Figure 7

Temperature test configurations: a) high temperature set-up, b) low temperature set-up; and c) a view of laboratory set-up

During each temperature step, the MMA functionality was tested in the following sequence: firstly, two images were recorded with all micromirrors in ON and in OFF state (white and black picture) by using the CMOS camera. In order to identify non-working pixels, the image difference between white and black was calculated and, if the intensity value of a set of 3x3 pixels was lower than a given threshold, a non-working micromirror was detected. Thanks to the high pixel density of the employed camera, each micromirror can be mapped to at least 3x3 camera pixels. For the analysis, in order to avoid rounding errors, the monochromatic image was converted to floating point with values between 0 and 1; then the subtraction was made. Finally, they were converted to integer values (between 0 and 255) for plotting. The changes in intensity that can be observed by comparing these figures to those obtained in the experiments in following subsections, are due to modifications in the optical alignment and the distance between the elements. Specifically, we could not exactly reproduce the same illumination conditions in all the experiments; this, however, does not affect to the detection of possible defect micromirrors. A calibration procedure was carried out upfront, changing intentionally the state of a single micromirror and it could be confirmed that the camera is able to detect single-pixel failures.

As an example, we report the functionality test at 80 °C in Figure 8. All tests carried out between -40 °C and 80 °C showed the same MMA performance, maintaining full functionality (zero failure rate) for the whole temperature range.

Figure 8

Functionality test at 80°C: a) recorded white image; b) recorded black image; c) calculated intensity difference between a) and b); d) matrix representation that would show defect mirrors, if there were any.

3.2.

Vacuum Test

Vacuum experiment

We assessed the robustness of a MMA sample without any glass cover after being exposed to a low pressure environment, by means of a vacuum chamber with a custom pumping station featuring an Edwards XDS10 Dry Scroll primary pump and a Pfeiffer Balzers turbo pump (Figure 9). Before starting the experiment, the vacuum chamber and the MMA support were cleaned in a flow box (with clean room classification ISO 5) to remove dust and other contaminants. To ensure stable experiment conditions, the tests started after the pressure in the chamber remained constant (< 10^-5 mbar) at least for 2 hours at ambient temperature.

Figure 9

Test set-up with MMA in the vacuum chamber

Due to limited space for all relevant components inside the vacuum chamber, no MMA control and functionality test was applied due to limited space for all relevant components, but rather the operational MMA test was performed immediately after removing the device from the chamber. The functionality test is similar to the one used for the temperature experiment: two MMA patterns were applied, the relevant images recorded and the positions of non-working mirrors were identified, if any. Test results showed a zero failure rate of the MMA (Figure 10). This means all micromirrors were able to switch their state from ON to OFF before and after having been exposed to vacuum.

Figure 10

Functionality test after vacuum experiment

Challenging vacuum conditions entail an extensive and complex analysis, since changing damping conditions lead to different mirror oscillation times when the tilt angle is changed. At lower pressure the mirror damping decreases and the oscillation takes longer until the mirror remains in a stable position, therefore overall operation speed decreases. In addition, mirror overshoot with temporary high deflections during the mirror settling could occur, which might cause mirror failures in a worst case scenario. Hence, it seems sensible to keep standard atmospheric pressure, encapsulating the MMA device inside a hermetical package. Moreover, a hermetical solution would enhance the long-term reliability of the device, making possible a strengthened protection of the array surface against moisture and particles [16]. Operational tests in vacuum environment will be carried out in future activities by using a space-suitable hermetical MMA package.

Theoretical investigations: Hermetical package in vacuum

Pressure differences between inside (atmospheric pressure 1 atm) and outside (almost vacuum conditions) the MMA package are challenging for package components, especially for the window cover. For this reason, a FEM model for a given hermetical package was implemented to calculate the stress level and displacement on the window cover under vacuum environment. Figure 11 shows the FEM Model with a sapphire window in a Kovar frame on a ceramic chip carrier. No bonding layers to fix the window or other components were included in the model. Thanks to device symmetry, and in order to increase computational speed, only one quarter of the MMA package was considered. The assumed pressure difference of air between the encapsulated domain and the vacuum environment is p = 1000 hPa.

Figure 11

FEM Model for MMA hermetical package

The displacement of the window center u_z as well as the maximum total amount of the x-directed tensile stress component σ_xx is displayed in Figure 12 as function of the thickness of a sapphire cover window. Decreasing cover glass thickness increases glass displacement and tensile stress. Therefore, in order not to reach a stress level of approximately 270 MPa (Table 1) and to keep the risk of glass cracks very low, the glass thickness must be larger than 0.3 mm.

Figure 12

Maximum tensile stress (σ_xx) and center displacement (u_z) vs. window-cover thickness (t_glass)

Table 1

Mechanical properties of Kovar (Fe/Ni/Co alloy) and Sapphire

3.3.

Vibration Test

The vibration test determines the dynamic load that the MMA can withstand. To generate this load, in this test only random vibration between upper and lower frequency limits were exerted, as deemed the most critical for the device in a space application. Random vibration is more characteristic of modern-field environments produced by rocket engines [7]. To apply the required vibration levels, a LDS555 electro-dynamic shaker with an LDS PA1000L linear power amplified was employed. The MMA device was tested along the three axis directions: X, Y and Z, for 2 minutes each. The MMA was mounted in a specific bracket onto the shaker; this bracket can accommodate the device in different orientations as Figure 13 depicts. Additionally, a control accelerometer was mounted onto the shaker to control the vibration level applied onto the system.

Figure 13

Setup for vibration test along the three axis: a) X-axis, b) Y-axis and c) Z-axis

Various levels (Table 2) were tested in order to find a potential limit for acceptable vibrations and these levels correspond to those defined by the MIL-STD-883 specification method 2026 [22]. The root mean square acceleration (G_rms) is the square root of the area under the acceleration spectral density curve in the frequency domain. The G_rms value is typically used to express the overall energy of a particular random vibration event and is a statistical value used in mechanical engineering for structural design and analysis purposes [23].

Table 2

Test condition letters for Random vibration Test according to MIL-STD 883 Method 2026

In Figure 14 the target (green) and the actual (black) measured acceleration profiles in frequency domain are represented. It can be seen that the measured profile falls inside the tolerances (top and bottom yellow lines). MMA operational verification tests were performed before and after the test of each axis and they were successful (Figure 15), assuring the MMA functionality at least for the vibration condition 1-G. In order to check higher vibration levels in the future, additional testing on a more powerful shaker will be required.

Figure 14

Example of a random spectrum for the test condition 1-G along X-axis

Figure 15

Functionality test after vibration experiment