The Bragg reflector was manufactured by physical vapor deposition (PVD) to avoid degradation of the PZT performance. It consists of several (n) bilayers of amorphous silicon (110 nm) and oxide (305 nm), depending on the target reflectivity and absorption. A two-bilayer (n = 2) Bragg reflector was chosen for the LIDAR application.
To create the Bragg bilayers, an Endura PVD magnetron sputtering chamber from Applied Material (U.S.A.) was used, having been loaded with a silicon target for the 110 nm amorphous silicon layers. The 305 nm silicon dioxide layers were deposited in the same chamber by reactive sputtering of DC-generated plasma.
The evolution of the mirror’s flatness and of the 2D-scan patterns with increasing temperature were experimentally characterized. These characterizations were carried out with the scanner thermalized on a hot plate. This experimental protocol has two principal advantages: it allows precise control of the scanner temperature, and it is compatible with a wide range of temperatures (from ambient to 200 °C).
2.1. Mirror Planarity with Temperature
Due to residual stresses in the multiple layers making up the mirror, the mirror was not perfectly flat. A concave or convex deformation of the mirror is often observed after fabrication, and this deformation affects beam reflection. The deformation can be minimized through technological means, for example, by changing the thickness and stress of the layers on the top and bottom of the mirror. The deposition process was first optimized to minimize internal multilayer stress. The average internal stress of the Bragg reflector bilayers (n = 2) was measured at −155 MPa on a complete wafer after deposition. This low compressive stress makes it possible to minimize mirror deformation induced by stress.
At the same time, the scanner’s Z-deformation was measured after the Bragg process, using an Altisurf 520 tool from Altimet (Fr.).
Figure 3 shows an overview of the scanner’s deformation along the Z axis at ambient temperature.
The mirror clearly displayed a convex deformation at 30 °C. The static radius of curvature (SRC), measured along the diagonal of the mirror over 2 mm and centered on the mirror, was estimated to be 230 mm. The 4 mm-long actuators had a positive deflection of 57 µm, which was induced by compressive stress affecting the whole multilayer stack.
The evolution of Z deflection by the actuators and the mirror, as a function of temperature, was characterized over the temperature range from 30 °C to 200 °C. To carry out this evaluation, the scanner was thermalized on the heating plate integrated into the Altisurf 520 tool. As shown in
Figure 4, increasing scanner temperature resulted in a 10% decrease in the Z-deflection, which reached a value close to 50 µm at 150 °C.
This expected evolution is consistent with published analytical models of multilayers [
11,
12]. The thermal stress
between a film (i) and a substrate (s) is defined by the equation
where
are the thermal expansion coefficients of the film (i) and substrate (s),
corresponds to their biaxial Young modulus,
is the Poisson ratio and
and
are, respectively, the temperatures of deposition and operation.
Taking the values of
and
reported in [
8] and our
values, as shown in
Table 1, the formula clearly shows that increasing the operating temperature,
, from 30 °C to 200 °C induces a decrease in overall stress, which in turn reduces beam deflection.
The radius of curvature of a multilayer beam can be estimated using the following formula [
11]:
where
the thickness of the layer (i) or substrate (s). This last equation is a first-order approximation that omits terms with orders higher than
. This omission is consistent with the
layer thicknesses being at least 20-fold thinner than the substrate thickness. However, these models do not account for any bending moment on the beams. This assumption does not apply in our case because the beams are connected to the mirror. Nevertheless, the model allowed us to determine that the largest
contributions were related to the bottom 1 µm-SiO
2 film, which was present under the substrate. To a lesser extent, the PZT layer contributed as well. Some layers, like TiO
2, may not be taken into account because of their negligible thickness. As a result, the overall radius of curvature of the actuators, and the degree of deflection, is mainly to the result of the compressive 1 µm-bottom SiO
2 film present under the mirror. It should be noted that the initial deformation of the actuators had no direct impact on the optical angle attained; it affected only the Z-position of the mirror.
Figure 5 shows changes to the mirror’s flatness as a function of temperature. The multilayer stack on the mirror is different from that present on the actuator. The mirror tends to flatten as the temperature increases. The increase in Z-signal noise with increasing temperature is linked to the measurement tool and does not reflect a change in the reflector surface, as the reflector was deposited at a higher temperature (250 °C).
Despite this noise, from experimental measurements, the SRC was estimated to be close to 340 mm at 150 °C. In future developments, this mirror curvature will need to be reduced to achieve the Rayleigh criteria: a Z-deformation of less than λ/4. This result could be achieved by, for example, increasing the thickness of the SiO2 layer present under the mirror or by adding a tensile layer on top of the silicon substrate.
2.2. 2D Projected Scan with Temperature
A key parameter to characterize is how the 2D scan of the reflected beam evolves as a function of scanner temperature. This measurement is used to determine and extrapolate the potential variation in the total optical angle depending on the mirror’s temperature in the context of a LIDAR application. The X and Y coordinates of the optical angle determine the size of the scene that the LIDAR can image.
The scanner was thermalized at temperatures between 30 °C and 150 °C, and the optical angles were measured using a class-2 visible laser reflected by the mirror and projected onto a screen. A camera was used to track the position of the laser on the screen by continuously measuring its barycenter. If the optical path (around 130 mm) between the screen and the scanner and the position of the barycenter of the laser on the screen are known, the X and Y coordinates of the optical angle of the spot can be deduced using tan = . The absolute uncertainty of the optical angle, which is linked to how precisely the optical path is measured, was estimated to be 0.2° on our test bench. The scanner temperature was limited to 150 °C because electrical breakdowns are often observed when our technological process is used at temperatures exceeding 200 °C.
A 2D scan was performed with a 200 Hz sinusoidal driving signal (X-axis) for the fast axis and a 4 Hz ramp for the slow axis (Y-axis). The PZT supply voltage was swept from 0 V to a maximum of 20 V with
. The total optical angle
and
was measured at y = 0 and x = 0, respectively. In addition, for each of the four temperatures tested, the total duration of the measurement was at least 30 min, including thermalization and the time required for 2D scanning.
Figure 6 shows the optical angles measured for one of our scanners at 30 °C and 150 °C. A slight decrease can be observed in both axes of the scanning pattern. Additionally, it is worth noting a minor distortion in the 2D scan at 150 °C. This distortion will require further examination.
Four identical Bragg mirrors were used for this test, as shown in
Figure 7. These mirrors have exactly the same design, the same Bragg reflector and the same technological process flow. The evolution of the effective optical angle,
, as a function of changes in scanner temperature from 30 °C to 150 °C is reported. The changes observed correlated well with previously published results [
8,
13,
14,
15].
Mirror deflection thus results from the movement of the PZT actuators and, consequently, of the effective transverse piezoelectric coefficient
, based on the inverse piezoelectric effect. Indeed, the piezoelectric stress (
can be expressed as
[
13], where
is the transverse electric field. In addition,
is proportional to
, the saturation polarization, and
, the relative permittivity measured out of plane, as
. This coefficient is highly dependent on the operating temperature of the mirror. Theoretical calculations [
14] predict that
will increase with temperature. In contrast, other authors [
8,
15] have reported experimental results showing that
increases to a maximum (near 40 °C) and then gradually decreases at higher temperatures (up to 200 °C).
After deposition, ferroelectrics like PZT may not be polarized, in which case they will have poor piezoelectric properties. Poling treatment, which involves applying a voltage well above the coercive voltage, orients the piezoelectric domains in the same direction. However, after poling, the piezoelectric coefficient slowly decays due to rearrangement of the domains towards their equilibrium state. If poling is performed at a high temperature (hot poling), the mobility of the ferroelectric domains increases significantly, allowing a better alignment in the direction of the poling field [
10]. Nevertheless, it should be noted that this increase in
after hot poling has been observed only at ambient temperature. Another study [
15] with an initial poling step of 25 V at ambient temperature and intermediate poling steps at higher temperatures found that
decreases with increasing temperature. The two curves converged at high temperatures (near 200 °C).
This case is different, as the temperature of the scanner increases progressively with continuous polarization of the PZT. A similar case has been reported elsewhere [
8]. Dahl-Hansen et al. followed the deflection tilt of a unipolar micro-mirror driven by a 20 V signal at a frequency of 1.5 kHz while varying the mirror temperature stepwise from ambient temperature to 175 °C. Our study shows a similar trend, with an increase in the optical angle between 50 °C and 100 °C and a decrease at 150 °C. The value of the maximum is more difficult to determine given the uncertainty of the measurements. In addition, one of our samples underwent an electrical breakdown at 100 °C, reflecting degradation at high temperature. This decrease in optical angle with increasing temperature has been attributed mainly to the decrease in
as the scanner temperature approaches the Curie temperature (near 350 °C in our case) [
8,
16,
17]. Alteration of film stress is also reported to have a potential impact [
8] but is considered to contribute only slightly to the evolution of
[
17].
The decrease in optical angle observed in
Figure 7 was close to 5%, which is consistent with values reported elsewhere [
8]. However, this evolution of the optical angle, which is linked to the actuation power, even if it is limited to a change of few percent, must be taken into account in the context of future use within a LIDAR system. This value has direct implications for the global area scanned from the scene. Limiting the temperature increase by means of a low-absorption reflector is a first response to this issue, but that solution could also be coupled to others, like those based on the influence of PZT poling or hot-poling conditions, as has been reported [
9]. These and other studies, such as investigations focusing on the evolution of
, should be explored further in the future.
The last parameter checked was the stability of the 2D scan at an operating temperature of 150 °C. To conduct this investigation, an image of four spots, as shown in
Figure 8, was continuously projected onto the screen for one hour. An image of the screen was taken every minute by a 720-by-540-pixel camera (X and Y) and the barycenter of each of the spots was calculated. The shift in the X and Y positions of the barycenter, is reported in pixels.
The standard deviation of the barycenter positions of the four spots, calculated in pixels for both X and Y axis, was estimated using the following formula: , where x is the value for one sample, is the mean value of all the samples and n is the number of samples. In our case, n = 60. The maximum value of is close to 1 pixel. This value is the smallest deviation that can be measured using our test bench. Thus, in our case, the relative deviation of the optical angle was close to 0.03°.
In conclusion, no significant deviation of the optical angle at a scanner temperature of 150 °C was observed for continuous one-hour measurements, demonstrating the stable performance of the PZT actuator.
Further studies are needed to verify that PZT performance is not affected by a return to ambient temperature after operation at different temperatures. The potential impact of a hot-poling step remains to be investigated as well.