Open Access
Add to BibSonomy
Abstract
The damage threshold of silicon-based cameras to laser irradiation is measured for continuous wave lasers at both in-band and out-of-band wavelengths. Clarifications about the various kinds of damage reported in the literature are also presented and explained through various tests. For increasing laser intensities, the sequence of laser effects on cameras usually starts from dazzling, to a decrease of the pixel response up to the complete neutralization of its pixel matrix. However, the intensity range for the complete sequence of these laser effects can strongly depend on the laser wavelengths and the tested cameras, particularly for out-of-band laser wavelengths. The proposed definitions of damage thresholds presented in this work offer a basis for the future comparison of results between different studies.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. INTRODUCTION
Imaging sensors are frequently used on autonomous systems and for surveillance purposes, either for security applications or industrial monitoring, with each successive generation exhibiting increased sensitivity and resolution. In parallel, there is also a growing exploitation of lasers ranging from pointers and remote sensing [1,2] up to high power laser sources for countermeasure applications [3,4]. This is the case because several laser technologies are continuously improving their output power, robustness, and compactness, which are widening the range of available laser wavelengths for these applications [5,6]. An example is the development of fiber-based lasers at 2 µm [7–9] and 2.8 µm [10,11] that represent promising tools for processing glass and polymer materials, as well as offering lower eye safety requirements than standard industrial lasers at 1 µm wavelength. Within this context, a better understanding and quantification of laser effects on imaging sensors is required. More specifically, the damage threshold of imaging devices depends on the type of laser beam [continuous wave (CW) or pulsed] and the laser wavelength to which they are exposed. In general, pulsed lasers at low repetition rates do not heat the sensors. Instead, they produce strong electric fields capable of inducing optical breakdown in the sensor materials [12,13]. For imaging devices exposed to a CW laser, the damage is usually due to thermal effects [14–16]. It is generally well accepted that the pulse duration must be over 1 µs to be considered as a CW in laser-induced damage [12–16]. When the pulse lengths are between 1 ns and 1 µs, laser-induced damage can occur either because of thermal effects or by optical breakdown.
Numerous studies have been performed on the interaction of femtosecond to nanosecond laser pulses on imaging sensors [12,13,17–25]. Remarkably, in the case of CW laser exposure, there are only few published reports [13–16,26–28] in the last decade that updated and quantified these laser impacts on imaging devices. These works were mainly done using in-band 1 µm lasers while out-of-band wavelengths were not explored. In the literature, there are also various definitions for the laser damage threshold on a camera. Some definitions are based on single pixel damage, and others refer to line/cross damage in the readout image, while other references consider the damage as the complete neutralization of the matrix sensor. Additionally, there is a large variation for the reported damage threshold (${{\rm{D}}_{{\rm{th}}}}$) of a silicon-based camera, depending if it is a color CCD camera (${{\rm{D}}_{{\rm{th}}}} = \sim 1{{0}}\;{\rm{kW/c}}{{\rm{m}}^2}$) [28] or a monochromatic CCD camera (${{\rm{D}}_{{\rm{th}}}} = \sim 20{{0}}\;{\rm{kW/c}}{{\rm{m}}^2}$) [16,28]. Some publications reported also that single pixel damage is due to melting and/or darkening of the microlens matrix in front of the pixels, while others explained those damaged by the thermal load that melted/disconnected the electrodes performing the pixels’ readout. Han et al. presented the temporal evolution over few seconds of the damage morphology at a fixed laser power density of ${{50}}\;{\rm{kW/c}}{{\rm{m}}^2}$ and associated the line damage observed in the monochromatic CCD output video to the melting of the silicon pixels [27]. Li et al. damaged a color CCD with ${{15}}\;{\rm{kW/c}}{{\rm{m}}^2}$ for 1 ms exposition time and measured a temperature of 1597°C at the laser exposition site [26], which is well above the 1414°C [29] melting temperature of silicon. However, the simulations in [14,15] indicate that the surface temperature of silicon pixel should be below 500°C for similar laser conditions reported in previous experiments [26,27].
In [16], Burgess et al. demonstrated that the damage threshold of silicon-based imaging devices is nearly constant for exposure time greater than 100 µs. This was observed because the dimension of the pixels, thermal conductivity, and diffusivity of silicon allow rapid dissipation of the heat load from the laser exposition site within a time scale around 100 µs. In the same work [16], they demonstrated that, for indium gallium arsenide (InGaAs) cameras, the steady-state between the thermal absorption and diffusion takes more time to reach (around 10 ms), mainly because of the lower thermal conductivity and diffusivity of InGaAs.
Notwithstanding the great advances provided by these previous publications [15,16,26–28], there are still important points that remained unanswered for the interaction of CW lasers with silicon-based cameras, like the impact of the laser wavelength (in-band or out-of-band) on the damage thresholds and on the damage morphologies. It is important to compare these effects for different cameras (CMOS, CCD, color, monochromatic, etc.) and to establish a standard definition of the damage threshold. Such definitions must be independent with the subsequent signal processing algorithms, which can induce erroneous diagnostic of the sensor’s damage threshold by an interpolation of neighboring pixel’s signals. Additionally, it will be important to use laser exposition time representative for outdoor scenarios where atmospheric turbulences modulate the laser beam intensity.
Within this context, this paper presents a comprehensive analysis of CW damage thresholds on silicon-based cameras, as well as an analysis of the underlying phenomenology. The results of laser damage tests are reported for different cameras: CCD array, 2-D CMOS, and CCD cameras, and for color and monochromatic cameras. Tests were performed at 532 nm, 1072 nm, and at 1970 nm, covering in-band and out-of-band exposure conditions. Finally, a standardization for the definition of the damage threshold is proposed.
2. EXPERIMENTAL SETUP
The evaluation of the damage threshold for those cameras was realized with CW laser beams for 100 ms exposition duration. This laser exposition time was selected in order to be far above the time scale for the thermal steady-state in silicon [16], which occurs around 100 µs, but also to be archetypal of outdoor scenarios where the incident laser intensity fluctuates in turbulent atmosphere. Figure 1 presents the simulation of the laser power entering a camera lens (pupil entrance of 2 cm diameter) after the laser beam propagated 2 km in turbulent atmosphere. The simulations were performed with the WoNat software [30] using the modified Kolmogorov model [31–34] with inner scale of 2 cm and outer scale of 2 m. Figure 1(a) shows the laser power ratio entering the camera lens for two different turbulence strengths. Values for the refractive index structure parameter (${\rm{C}}_n^2$) used for the simulations are ${{1}}{{{0}}^{- 13}}\;{{\rm{m}}^{- 2/3}}$ and ${{1}}{{{0}}^{- 14}}\;{{\rm{m}}^{- 2/3}}$, which can be qualified as being strong and average atmospheric turbulences, respectively. The temporal modulations observed in Fig. 1(a) are dependent on the laser propagation distance and the entrance aperture of the camera lens, but these simulations are fair representations of laser signals observed by cameras for outdoor laser propagations [16]. Figure 1(b) presents the Fourier transform of the laser modulations simulated in Fig. 1(a). We clearly see in these power spectra that the laser modulation periods in turbulent atmosphere are mainly above few tens of milliseconds.
Fig. 1. (a) Simulation of the temporal evolution of the laser power entering a 2-cm-diameter camera lens at 2 km distance for a kW-class laser beam at two different turbulence strengths (${\rm{C}}_n^2$). Initial laser parameters are beam diameter of 6.5 cm at full width at half-maximum, effective focal length of 2 km, and wavelength of 1075 nm. (b) Spectral distribution of the laser power entering the camera lens for the two simulated turbulence strengths.
In order to precisely establish the laser damage threshold of those cameras and considering that the laser intensity fluctuations in Fig. 1(a) are around 100 ms, indoor tests were performed with stable laser expositions at a fixed pulse width of 100 ms (with rising and falling edges under 0.5 ms). It is important to remember here that the thermal steady-state in silicon-based cameras is reached in about 100 µs, and the damage threshold will not change importantly with laser pulse width longer than few tens of milliseconds.
Figure 2 illustrates the laboratory setup used to evaluate the damage induced on cameras by in-band or out-of-band laser exposure. The laser-induced damage thresholds are determined by calculating the damage probability at each laser intensity as described in [35]. A probit fit, based on the Gaussian cumulative density function, was applied to the measured laser intensities where damages were observed for the tested cameras. The damage threshold corresponded to the laser intensity where the probability of damage equals 50%, and its variance is related to the range of laser intensities where this probability passes from 13.6% to 86.4%. The characteristic list of the tested cameras is given in Table 1, and it is important to note that the damage threshold values are determined at the active surface of the cameras, independently of the transmission of optics and filters in front of the sensors. For color cameras, the near-infrared (NIR) blocking filters having a top-hat profile with an optical density (OD) of 2.5 between 700 nm and 1200 nm were removed for these tests. The laboratory setup was designed and built to focus the laser beam on the active surface of the camera with a plano–convex lens of $f/{{6}}$. Each laser could be selected for the damage test by displacing the movable mirrors (see Fig. 2). The camera was installed on a three-axis translation stage to adjust its position relative to the laser focus. To perform tests below the minimum laser power, calibrated neutral density (ND) filters were placed in the optical path. A reflection from a wedge was directed toward a calibrated photodiode (Thorlabs, PDA36A and PDA20H) to confirm the measured laser power and the exposure time. For in-band lasers (532 nm and 1072 nm), the diameter of the laser at the focus was directly measured by the tested camera by calibrating its response with laser signal below its saturation. For the out-of-band (1970 nm) laser focus, its diameter was determined using the knife-edge technique along both the vertical and horizontal axes [35]. For each test, the laser focus has a near Gaussian shape (${{\rm{M}}^2}$ ranging from 1.15 to 1.3), and the measured full width at half-maximum on the camera’s active surface ranged from 40 µm to 75 µm, depending on the laser wavelength and its position in the Rayleigh range.
Table 1. Characteristics List of CCD and CMOS Cameras Used for the Tests
Fig. 3. Example of in-band damage by a 532 nm laser observed (a) at the output image of a color CMOS camera and (b) the corresponding normalized response [horizontal array indicated by the white dashed line in (a)]. Example of out-of-band damage induced by a 1970 nm laser observed (c) at the output image of a color CMOS camera and (d) the corresponding normalized response [horizontal array indicated by the white dashed line in (c)].
A high speed, 50,000 frames per second monitoring camera (Photron, MiniWX50) with a 20X microscope objective (MO) was placed at an angle near the front surface of the tested camera in order to verify any visible morphology change, discoloration, or damage during the laser tests. An appropriate notch filter was installed in front of the high speed camera to block the tested laser wavelengths. The NIR thermal emission of the laser heated pixels was also collected with a multimode ZBLAN fiber (core of 600 µm). The collected thermal emission was either detected with a PbSe photodetector (Thorlabs, PDA20H) or its spectral distribution was measured with an InGaAs spectrometer (OceanOptics, NQ512-2.5) calibrated with a blackbody reference source (Infrared Systems, IR-508/301). Notch filters of 100 nm bandwidth were used both from the collecting ZBLAN fiber and at the entrance of the detection devices to block any laser scatterings. For each laser exposure, the output signal/video of the tested camera was recorded before, during, and after the laser exposure. An incoherent white light source was turned on during these tests, and its brightness was adjusted in order to generate a uniform signal across the sensor corresponding to about 20% of the dynamic range of the tested camera. This background signal allowed us to verify any permanent or transient degradation/modification of the pixel response right after the laser exposure. Within the limits of this experimental setup and the error measurements, we did not observe any impact on the measured laser damage threshold by changing this background signal level.
3. DEFINITIONS OF LASER DAMAGE THRESHOLDS
This section summarizes the damage observed for CMOS and CCD cameras, both for monochromatic and color cameras. In the cases of these 2-D cameras, we observed first a local degradation/modification of the pixel response at the laser exposure site [see Fig. 3(a)] before complete line damage occurred. These degradations/modifications at the laser exposed sites were permanent and dependent on the laser wavelength according to the presence of a Bayer filter or a microlens matrix.
Cameras with Bayer filters presented first a lowering or an increase of the pixel response depending of the pixel color (red, green, or blue channel) and the laser wavelength. For example, in the case of a color CMOS camera exposed to ${{150}}\;{\rm{kW/c}}{{\rm{m}}^2}$ at 532 nm [as shown in Fig. 3(a)], we observed a 50% decrease of the green pixel response at the laser focus, but also with an increase of 100% of the red pixel response. On the other hand, when the same CMOS camera was exposed to an out-of-band laser of ${{420}}\;{\rm{kW/c}}{{\rm{m}}^2}$ at 1970 nm, we also observed a permanent modification of the pixel response depending on the channel color as shown in Fig. 3(c). For this last case, both red and blue channels undergo an increase of 45% and 50% of their responses while the green channel response remained almost unchanged. Based on these examples and the various combinations of color channel affected differently by various laser wavelengths, we define here the first level of damage as the first permanent variation by 50% (either positive or negative) of a single pixel’s response, independently of the color channel. For a monochromatic camera, this first level of damage is defined in the same way as a variation by 50% of the pixel response. Hereafter, this definition of the first level of damage is named ${\rm{D}}{{{1}}_{50\%}}$. When increasing the laser irradiation above ${\rm{D}}{{{1}}_{50\%}}$, the damaged area increases, affecting further the pixels’ response and varying radially across the laser exposed site. Usually, ${\rm{D}}{{{1}}_{50\%}}$ can be determined only by analyzing the output image of the tested camera, but it could be associated with surface morphology or discoloration observed under the microscope.
It is important to emphasize that this first level of damage is not due to a deterioration of the silicon-based pixel, but to an alteration of the transmission of the Bayer filter and/or the microlens matrix in front of the pixels. It is nevertheless the first irreversible alteration of the sensor system and as such it is quantified separately. This damage or deterioration of pixel responses were not observed on the LC100 camera simply because this CCD array does not have a polymer layer (Bayer filter or microlens) in front of the pixels.
When a silicon-based camera is exposed to higher laser intensities, a second level of damage appears. As defined very similarly in previous publications [16,27,28], this second level of damage, named here ${\rm{D}}{{{2}}_{{\rm{line}}}}$ corresponds to the apparition of line damage (usually observed for CCD camera) or cross damage (usually observed for a CMOS camera), as shown in Fig. 4. Here, ${\rm{D}}{{{2}}_{{\rm{line}}}}$ is caused by the melting of the electrode in the reading layer due to the thermal load accumulation at the laser exposition site. To demonstrate this, the temperature of the irradiated pixel was estimated by fitting its measured infrared emission spectrum with the blackbody equation as shown in Fig. 4(a). For each camera, the temperature of the heated pixels was extracted at different laser intensities and was summarized in Fig. 4(b). From these measurements, we noticed that the laser intensity range where the second level of damage was observed [grey zone in Fig. 4(b)] corresponded to temperatures between 550°C and 950°C. This temperature range is well below the 1414°C melting temperature of silicon [29]. However it corresponds to the melting temperature of aluminum or aluminum-copper alloys frequently used in microfabrication of camera circuits.
Fig. 4. (a) Examples of infrared emission spectra of pixels from a DCC1545M monochrome camera irradiated with a 1072 nm laser. Dashed lines correspond to fitted blackbody emission curves with the measured spectrum. (b) Summary of irradiated pixel temperatures extracted by fitting their infrared emission spectra at different l072 nm laser intensities and for different tested cameras. (c) Example of second level of damage (line damage) observed on a CMLN-13S2 CCD camera from its output image. (d) Example of second level of damage (cross damage) observed on a DCC1645C CMOS camera from its output image.
In the case of a CCD camera, the image is generated by moving the photoelectrons from each pixel along a single array. Then, each group of photoelectrons from each array of the CCD is moved to the shift register where the signal is amplified. Therefore, the reading layer for a CCD camera is mainly structured along one axis. When the laser intensity is sufficiently high to melt the electrode layer at the laser exposition site, the electrodes short-circuit, causing dark arrays (or unreadable arrays) to appear in the video output as shown in Fig. 4(c).
In a CMOS camera, the photoelectrons produced at each pixel are converted to a voltage at the pixel site, following which the signals are multiplexed towards an amplifier by transistors and a network of horizontal and vertical electrodes. When the laser intensity is sufficiently high to melt the electrode layer at the interaction site, the electrodes short-circuit resulting in a dark cross pattern at the laser exposition site [see for example Fig. 4(d)].
Fig. 5. Microscope pictures (Olympus, MX63) of camera surface damage taken after the exposition by a 100 ms, 1072 nm laser on color CMOS camera (DCC1645C) for intensity corresponding to (a) ${\rm{D}}{{{2}}_{{\rm{line}}}}$ and (b) for laser intensity 4 times above ${\rm{D}}{{{2}}_{{\rm{line}}}}$. Microscope pictures of a monochrome CMOS camera (DCC1545M) for laser intensity corresponding to (c) ${\rm{D}}{{{2}}_{{\rm{line}}}}$ and (d) for laser intensity 4 times above ${\rm{D}}{{{2}}_{{\rm{line}}}}$.
For reference purposes, the surface morphology of color and monochromatic CMOS cameras at a laser intensity corresponding to ${\rm{D}}{{{2}}_{{\rm{line}}}}$ are shown in Figs. 5(a) and 5(c). We can observe that the silicon pixels are still intact at the center of the laser exposed sites, but the Bayer filter or microlens matrix are burned at the center, and melted in the periphery of the laser focus. When increasing the laser intensity beyond ${\rm{D}}{{{2}}_{{\rm{line}}}}$, the thermal load increases under the pixels, eventually causing thermal damage and mechanical stress to the insulation layer [15]. This in turn induces a short circuit and the complete neutralization of the matrix sensor, causing an unreadable output image. This third level of damage is named here D3 and the corresponding camera output images are completely dark. For laser intensities above ${\rm{D}}{{{2}}_{{\rm{line}}}}$, but below D3, the thermal load at the pixel surface can be sufficient to reach the silicon melting point and cause surface ablation, as observed in Figs. 5(b) and 5(d), for color and monochromatic CMOS cameras. In these cases, especially for CMOS cameras, even if the silicon pixels melted at the laser exposition site, the surrounding pixels of the CMOS matrix remained operational, and D3 was not reached. On the other hand, the third level damage (D3) on CCD camera was easier to reach and the required laser intensities were only 20% to 30% higher than the laser intensity needed for ${\rm{D}}{{{2}}_{{\rm{line}}}}$. The summary for all these tests using different cameras and laser wavelengths are reported in the following sections. Some examples of the evolution and impacts of the thermal load during the laser exposition are also presented.
Fig. 6. (a) High speed sequences of surface images of the CCD array LC100 irradiated at 1072 nm for 100 ms. This sequence was grabbed with high speed camera at 50,000 frames per second, ${{20}} \times$ microscope objective, and NIR filters blocking the 1072 nm laser scattering. (b) Temporal profiles of the incident 100 ms and 1072 nm laser pulse (red line) and the corresponding integrated infrared emission from the irradiated LC100 pixels grabbed with a PbSe detector sensitive in the range of infrared wavelengths between 1.5 µm and 4.8 µm. (c) Corresponding surface image of LC100 taken after the laser exposition with a ${{20}} \times$ microscope (Olympus, MX63).
4. LASER DAMAGE TESTS
A. Temporal Characterization of the Laser Damage
In this section, we will limit the presentation of results to the effects of a 1072 nm laser on the LC100 (CCD array) and DCC1545M (monochromatic) cameras. The temporal evolutions of the laser damage are representative of those observed on others cameras and for others laser wavelengths. Figure 6 presents the temporal shape of the laser pulse (red curve), the corresponding emission of the irradiated pixels from LC100 (black curve), and a series of camera surface images obtained with a MO in front of the high speed camera (see experimental setup in Fig. 2). It is important to remember here that the first level of damage (${\rm{D}}{{{1}}_{50\%}}$) was not observed on the LC100 because this CCD array does not have a polymer layer (Bayer filter or microlens) in front of its pixels. Therefore, the sequence of images taken by the high speed camera in Fig. 6(a) corresponds to laser intensity for the second level of damage (${\rm{D}}{{{2}}_{{\rm{line}}}}$). As seen in images of Fig. 6(a), we observe an increase of the laser heated area during the 100 ms exposure, which explains the increase of the integrated IR signal observed in Fig. 6(b) (black line). In the case of the IR spectral measurements, the infrared emission corresponded to a surface temperature of ${{700}}\;{{\pm}}\;{{50}}^\circ {\rm{C}}$ for this sequence of measurement. Figure 6(c) shows the 20X microscope picture of the CCD array taken after the laser test for an intensity around ${\rm{D}}{{{2}}_{{\rm{line}}}}$. There is no ablation/depth observed at the surface of the CCD array. The ring observed in Fig. 6(c) is due to the microscope light interference with a silicon oxide layer generated at the laser focus on the pixel surface. Thermal oxidation of silicon is known to begin at temperature of 700-800°C [36,37], which corresponds to the estimated temperature from the measured infrared spectrum.
Figure 7 presents the results for the monochromatic camera DCC1545M which incorporates a microlens matrix of in front of the pixels. For this camera as well as for others 2-D cameras, it was not possible to characterize the temporal evolution of ${\rm{D}}{{{1}}_{50\%}}$ because the infrared emission is too weak at temperatures corresponding to the melting of polymers (starting around 200°C) [29]. When the laser intensity is increased to ${\rm{D}}{{{2}}_{{\rm{line}}}}$, the infrared emission is initially blocked at the surface by the microlens matrix for 10–20 ms after the beginning of the laser irradiation. As shown in Fig. 7(a) with the first image of the sequence, there is no surface morphology change or emission that was noticeable by monitoring its surface with the high speed camera. After this delay, the temperature rise is sufficient to initiate the combustion of the polymer layer, first at the center of the laser focus (see time sequence at 20 ms) and then at the surrounding of the laser focus (see time sequence at 30 ms). Following this combustion of the polymer layer, we can observe the infrared emission of the heated silicon pixel. It is important to note that similar results were observed with color cameras (Bayer filter instead of microlens matrix) or for CMOS and CCD cameras. Additionally, if the laser intensity is increased well above ${\rm{D}}{{{2}}_{{\rm{line}}}}$, the temperature of the laser exposed area increases more rapidly and the combustion of the polymer layer is initiated earlier. Figure 7(c) shows the ${{50}} \times$ microscope color picture taken after the laser test for an intensity around ${\rm{D}}{{{2}}_{{\rm{line}}}}$. At the center of this picture, we can observe the exposed pixels after the combustion of the microlens at the laser focus. The ring around the laser focus corresponds to melted microlenses.
Fig. 7. (a) High speed sequences of surface images of the monochrome camera DCC1545M irradiated at 1072 nm for 100 ms. This sequence was grabbed with high speed camera at 50,000 frames per second, ${{20}} \times$ microscope objective, and NIR filters blocking the 1072 nm laser scattering. (b) Temporal profiles of the incident 100 ms and 1072 nm laser pulse (red line) and the corresponding integrated infrared emission from the irradiated DCC1545M pixels grabbed with a PbSe detector sensitive in the range of infrared wavelengths between 1.5 µm and 4.8 µm. (c) Corresponding surface image of DCC1545M taken after the laser exposition with a ${{50}} \times$ microscope (Olympus, MX63).
B. Laser Wavelength Impacts on the Damage Threshold
In this section, we present the results on the cameras LC100 (CCD array) for in-band and out-of-band laser exposure. We show how the laser damage occurs relatively to the full-chip dazzling for different laser wavelengths. For this series of tests, the laser intensity was gradually increased at the laser focus, and the number of saturated pixels on the CCD array was measured as a function of the laser intensity. The CCD array was composed of 2048 silicon-based pixels, and their spectral response ranged from 350 nm to 1100 nm. The integration time of the CCD array was fixed to 5 ms, and it was exposed during 100 ms to laser radiation at 1072 nm (in-band) and 1970 nm (out-of-band). The laser intensity required to begin to saturate an irradiated pixel is ${{35}}\;{\rm{mW/c}}{{\rm{m}}^2}$ and ${{6}}\;{\rm{kW/c}}{{\rm{m}}^2}$, at 1072 nm and 1970 nm, respectively.
The number of saturated pixels as a function of the laser intensity at 1072 nm is presented in Fig. 8(a). We can observe that the number of saturated pixels increases slowly with laser intensity up to ${\sim}{0.1}\;{\rm{W/c}}{{\rm{m}}^2}$. Above this laser intensity, the number of saturated pixels on the CCD array augments very rapidly, and the sensor becomes fully saturated around ${{10}}\;{\rm{W/c}}{{\rm{m}}^2}$. It is important to note here that the saturation of the sensor is not due to scattering and reflections from the lens in front of the camera, but mainly by the saturation of the amplification stage by an overflow of photoelectrons from the irradiated pixels. This is confirmed by analyzing the camera output signal as a function of the irradiated time and pixel positions.
Fig. 8. Graphs representing the number of saturated pixels on a CCD array as a function of laser intensity. (a) 1072 nm for 100 ms. (b) 1970 nm for 100 ms. Red dashed lines indicate the laser intensity for which the second level of damage (${\rm{D}}{{{2}}_{{\rm{line}}}}$) occurred. Note that the horizontal scales are the same for these graphs.
When increasing further the laser intensity on the CCD array, complete line damage of the CCD array can be achieved, as indicated be the red dashed line in Fig. 8(a). No transition effect was observed, such as the lowering of pixel response, which would have indicated imminent damage. For each laser exposure, even at few percent below the damage threshold ${\rm{D}}{{{2}}_{{\rm{line}}}}$, the CCD array retrieved its initial response, indicating that there was no local degradation of the pixel response, especially at the laser focus site. Often, the complete damage of the CCD array occurred (i.e., the full array is permanently out of operation) while there was no apparent or surface damage observed at the laser focus site of the CCD array. These observations support the fact that the line damage of the CCD array is due to thermal damage of the electrodes [15,26,27], which short-circuited the reading layer of the CCD array.
Similar tests in Fig. 8(b) were performed by using an out-of-band fiber laser source at 1970 nm. Here, the number of saturated pixels as a function of the laser intensity was also measured, and the integration time of the CCD array was also fixed to 5 ms. We can observe in Fig. 8(b) that the number of saturated pixels increases very slowly with laser intensities up to ${\sim}{{100}}\;{\rm{kW/c}}{{\rm{m}}^2}$. Above this level, the number of saturated pixels on the CCD array augments very rapidly, and the sensor becomes fully saturated around ${{400 - 500}}\;{\rm{kW/c}}{{\rm{m}}^2}$. Such an observation represents a very nonlinear response of the CCD array, with the generated signal exhibiting a quadratic correlation to the incident 1970 nm laser intensity, indicating that this signal was mainly due to two-photon absorption from silicon. However, in opposition to what was observed in Fig. 8(a) at 1072 nm, we observe in Fig. 8(b) that the damage of the CCD array (see red dashed line) arrives soon after the full saturation of the CCD array by the 1970 nm laser.
Table 2. Dazzling and Damage Thresholds of Tested Cameras
Fig. 9. Graphs summarizing the results for (a) the full-chip dazzling and the first level of damage (${\rm{D}}{{{1}}_{50\%}}$), and (b) the second and third level damage of the tested cameras with in-band (532 nm or 1072 nm) and out-of-band (1970 nm) lasers. The color zones are eye guidelines to regroup the symbols associated to the same laser effects (dazzling or damage). Note that the vertical scales are the same for these graphs.
Comparably to what was observed for the damage at 1072 nm, no transition effects (i.e., a lowering of the pixel response) were observed on the CCD array with the 1970 nm laser. For each laser exposure, even at few percent below the damage threshold, the CCD array retrieved its initial response, indicating that there was no local degradation of the pixel, especially at the laser focus site.
These results are an example showing that the damage threshold ${\rm{D}}{{{2}}_{{\rm{line}}}}$ for the CCD array remains in the same order of magnitude when changing the laser wavelengths. However, the complete sequence of the laser effects on the camera (full-chip dazzling, ${\rm{D}}{{{1}}_{50\%}}$, ${\rm{D}}{{{2}}_{{\rm{line}}}}$, D3) with the laser intensity strongly depends on the laser wavelengths and cameras. All these damage thresholds are now recapitulated in the next section.
C. Summary for the Laser Damage Thresholds
All values for the full-chip dazzling, ${\rm{D}}{{{1}}_{50\%}}$, ${\rm{D}}{{{2}}_{{\rm{line}}}}$, and D3 are summarized in Table 2 and are presented graphically in Fig. 9. It is important to note that these values are determined at the active surface of the cameras, independently of the transmission of optics and filters in front of these sensors. For color cameras, the NIR blocking filters (${\rm{OD}} = {2.5}$ between 700 nm and 1200 nm) were removed in order to have sufficient incident laser power at 1072 nm to damage the sensors. For these tests, if the NIR blocking filter was put back in front of the sensors, the incident laser power at 1072 nm would have to be increased by a factor of ${{1}}{{{0}}^{2.5}}$ to achieve the same laser intensities for damaging the camera’s active surface as those presented in Table 2 and Fig. 9.
For in-band laser exposure, the first damage threshold (${\rm{D}}{{{1}}_{50\%}}$) is achieved at laser intensities 2–3 orders of magnitude greater than laser intensities for full-chip dazzling. For the color camera, the first level damage (${\rm{D}}{{{1}}_{50\%}}$) at 532 nm occurs at a laser intensity 10 times lower than its second level damage (${\rm{D}}{{{2}}_{{\rm{th}}}}$). However, for monochrome cameras, the ${\rm{D}}{{{1}}_{50\%}}$ is observed for in-band laser intensities slightly lower than its ${\rm{D}}{{{2}}_{{\rm{line}}}}$. The lower damage threshold for color cameras is due to higher absorption of the Bayer filter as compared to the transparent microlens matrix used for monochrome cameras. The second damage threshold (${\rm{D}}{{{2}}_{{\rm{line}}}}$) was observed for similar order of magnitude laser intensities for all cameras and laser wavelengths tested. These similarities are associated to the thermal load required to melt the reading layers, mainly composed of aluminum or aluminum-copper alloys for those cameras. For the third level of damage (D3), it was observed on CCD cameras for laser intensities at least 5 to 10 lower than for CMOS cameras. Finally, for the out-of-band laser, the sequence of full-chip dazzling, the first, second, and third level damage occurs for very close laser intensities, as shown in Fig. 9 at 1970 nm.
The damage observed with these CW lasers are due to photothermal effects. No correlation was observed between the pixel size/area and its damage threshold, most probably because the laser focus diameters used in our tests were larger than the pixel size. However, using a smaller laser focus of similar size to the camera pixel would probably increase the measured damage threshold due to the higher radial dissipation of the thermal load.
As summarized in Fig. 9, full-chip dazzling of all tested 2-D cameras (CMOS, CCD), occurs for in-band laser intensities between 5 and ${{30}}\;{\rm{kW/c}}{{\rm{m}}^2}$, which are 2–3 orders of magnitude lower than the laser intensities for their damage threshold (${\rm{D}}{{{1}}_{50\%}}$, ${\rm{D}}{{{2}}_{{\rm{line}}}}$). Conversely, for out-of-band (1970 nm) exposure, full-chip dazzling occurs for intensities between 200 and ${{540}}\;{\rm{kW/c}}{{\rm{m}}^2}$, near the laser intensities for their damage thresholds. Finally, for all silicon-based cameras tested (linear, CMOS, CCD, monochrome, and color), permanent damage occurs for laser intensity between ${{150}}\;{\rm{kW/c}}{{\rm{m}}^2}$ and ${1.5}\;{\rm{MW/c}}{{\rm{m}}^2}$ on the sensor chip.
5. CONCLUSION
Damage thresholds of silicon-based cameras to CW laser radiation has been measured for in-band and out-of-band laser wavelengths. Clarifications about various kinds of damage reported in the literature have also been explained and presented in this work for different laser intensities and wavelengths.
The proposed definitions of damage thresholds represent a sensor-agnostic basis for future comparison of results between different studies and for categorizing the phenomenology of laser-induced sensor damage. The suggested description of the first level of damage (${\rm{D}}{{{1}}_{50\%}}$) corresponds to the laser intensity range for the first possible permanent alteration of the sensor’s response, occurring through degradation and darkening of microlens matrix or Bayer filters in front of the pixels. While ${\rm{D}}{{{1}}_{50\%}}$ does not apply to imaging sensors lacking this layer, it is nevertheless paramount to consider due to the low thermal resistance of the materials used in this layer.
When the laser irradiance is further increased, there is a sudden shutdown and irreversible damage of the exposed pixels, due to the short circuit of the reading layer below the laser exposed site. It is correlated with the melting temperature of the metal alloys present in the pixel’s structure. This second level of damage (${\rm{D}}{{{2}}_{{\rm{line}}}}$) corresponds to the laser intensity range for inducing the first line damages on CCD or cross damage on the CMOS camera’s output images.
Finally, when increasing the laser intensity above ${\rm{D}}{{{2}}_{{\rm{line}}}}$, the thermal load increases under the pixels, eventually causing thermal damage and/or mechanical stress to the insulation layer [14,15]. Such impacts can induce a short circuit and the complete neutralization of the entire matrix sensor, resulting therefore in an unreadable output image. This third level of damage (D3) corresponds to camera output images that are completely dark.
For the cameras tested in this work, the observations can be summarized as follows:
- • ${\rm{D}}{{{1}}_{50\%}}$ for color cameras is significantly lower than ${\rm{D}}{{{1}}_{50\%}}$ for monochromatic cameras due to the stronger absorption of the Bayer filter;
- • ${\rm{D}}{{{1}}_{50\%}}$ for monochromatic cameras is similar, but slightly lower than ${\rm{D}}{{{2}}_{{\rm{line}}}}$ due to the weak absorption of the microlens matrix;
- • ${\rm{D}}{{{2}}_{{\rm{line}}}}$ is of similar order of magnitude for CCD and CMOS cameras (color and monochromatic) because it is associated with the melting point of aluminum or aluminum-copper alloys frequently used in microfabrication of camera circuits;
- • D3 for CCD cameras is significantly lower than D3 for CMOS cameras due to the pixel reading structure; and
- • For out-of-band laser wavelengths, ${\rm{D}}{{{1}}_{50\%}}$, ${\rm{D}}{{{2}}_{{\rm{line}}}}$ and D3 occur sequentially, but for close laser intensities.
As mentioned previously, the measured damage thresholds in Table 2 are for CW laser expositions, and, due to the thermal conductivity and diffusivity of silicon, they will increase if the laser focus diameter is similar to the pixel size.
The values presented in Table 2 are specific to the cameras tested here, but the damage evolution with the laser intensity and the observed damage morphologies can be generalized to other silicon-based cameras and laser wavelengths [14–16,26–28], as well as other sensor types. It is, therefore, very important to establish standard definitions of the damage induced on imaging sensors in order to better compare, in the future works, their resilience and their vulnerabilities under different laser exposition scenarios.
Funding
Defence Research and Development Canada.
Acknowledgment
This work was supported by the Defence Research and Development Canada Program. The authors acknowledge technical support from Nancy Bérubé and Christian Samson.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
REFERENCES
1. J. F. Daigle, D. Pudo, and F. Théberge, “Standoff high energy laser induced oxidation spectroscopy (HELIOS),” J. Appl. Phys. 122, 173102 (2017). [CrossRef]
2. L. Mei, P. Guan, and Z. Kong, “Remote sensing of atmospheric NO2 by employing the continuous-wave differential absorption lidar technique,” Opt. Express 25, A953–A962 (2017). [CrossRef]
3. D. H. Titterton, Military Laser Technology and Systems (Artech House, 2015).
4. N. Dalloz, T. Robin, B. Cadier, C. Kieleck, M. Eichhorn, and A. H. Dhollande, “55 W actively Q-switched single oscillator Tm3+, Ho3+-codoped silica polarization maintaining 2.09 µm fiber laser,” Opt. Express 27, 8387–8394 (2019). [CrossRef]
5. I. Elder, “Performance requirements for countermeasures lasers,” Proc. SPIE 7836, 783605 (2010). [CrossRef]
6. I. Mingareev, F. Weirauch, A. Olowinsky, L. Shah, P. Kadwani, and M. Richardson, “Welding of polymers using a 2 µm thulium fiber laser,” Opt. Laser Technol. 44, 2095–2099 (2012). [CrossRef]
7. P. F. Moulton, G. A. Rines, E. V. Slobodtchikov, K. F. Wall, G. Frith, B. Samson, and A. L. G. Carter, “Tm-doped fiber lasers: fundamentals and power scaling,” IEEE J. Sel. Top. Quantum Electron. 15, 85–92 (2009). [CrossRef]
8. C. Gaida, M. Gebhardt, T. Heuermann, F. Stutzki, C. Jauregui, and J. Limpert, “Ultrafast thulium fiber laser system emitting more than 1 kW of average power,” Opt. Lett. 43, 5853–5856 (2018). [CrossRef]
9. B. M. Anderson, J. Solomon, and A. Flores, “1.1 kW, beam-combinable thulium doped all-fiber amplifier,” Proc. SPIE 11665, 116650B (2021). [CrossRef]
10. V. Fortin, M. Bernier, S. T. Bah, and R. Vallée, “30 W fluoride glass all-fiber laser at 2.94 µm,” Opt. Lett. 40, 2882–2885 (2015). [CrossRef]
11. G. A. Newburgh and M. Dubinskii, “Power and efficiency scaling of Er: ZBLAN fiber laser,” Laser Phys. Lett. 18, 095102 (2021). [CrossRef]
12. X. Wang, Z. H. Shen, J. Lu, and X. W. Ni, “Laser-induced damage threshold of silicon in millisecond, nanosecond, and picosecond regimes,” J. Appl. Phys. 108, 033103 (2010). [CrossRef]
13. B. Schwarz, G. Ritt, and B. Eberle, “Impact of threshold assessment methods in laser-induced damage measurements using the examples of CCD, CMOS, and DMD,” Appl. Opt. 60, F39–F49 (2021). [CrossRef]
14. Z. Li, X. Wang, Z. Shen, J. Lu, and X. Ni, “Mechanisms for the millisecond laser-induced functional damage to silicon charge-coupled imaging sensors,” Appl. Opt. 54, 378–388 (2015). [CrossRef]
15. M. Han, X. Wang, J. Nie, K. Sun, and M. Zhang, “Temporal evolution characteristics and damage threshold of CCD detector irradiated by 1.06-µm continuous laser,” Optik 157, 1282–1291 (2018). [CrossRef]
16. C. Burgess, L. Hill, L. Lewins, and C. Westgate, “Modelled and experimental laser-induced sensor damage thresholds to continuous wave infrared sources,” Proc. SPIE 10797, 10797T (2018). [CrossRef]
17. J. Shao, J. Guo, T.-F. Wang, and M. Wang, “Experimental research on femtosecond laser damaging array CCD cameras,” Proc. SPIE 8796, 87960D (2013). [CrossRef]
18. J.-Y. Lin, R. Shu, G.-H. Huang, K.-M. Fang, and Z.-X. Yan, “Study on threshold of laser damage to CCD and CMOS image sensors,” J. Infrared Millimeter Waves 27, 172–478 (2008). [CrossRef]
19. J. Dai and Z. Wang, “New analysis on laser-induced damage mechanism of CCD device,” J. Laser Appl. 2009, 258 (2009). [CrossRef]
20. J. J. Jiang, F. Luo, and J. G. Chen, “Research on femtosecond laser induced damage to CCD,” High Power Laser Part. Beams 17, 515–517 (2005).
21. S. Y. Huang, Y. S. Zhang, and B. Q. Tang, “Damage effect on CCD detector irradiated by 500 fs laser pulse,” High Power Laser Part. Beams 17, 1445–1448 (2005).
22. X. W. Ni, J. Lu, and A. Z. He, “Measurement of laser damaging thresholds of CCD devices,” Laser Technol. 18, 153–156 (1994).
23. Z. Zhen, X. Zuodong, C. Deyan, S. Yubin, and J. Zhang, “Effect of FT-CCD damaged by nanosecond pulse laser,” Infrared Laser Eng. 46, 1003001 (2017). [CrossRef]
24. F. Guo, R. Zhu, A. Wang, and X. Cheng, “Damage effect on CMOS detector irradiated by single-pulse laser,” Proc. SPIE 8905, 890521 (2013). [CrossRef]
25. B. Schwarz, G. Ritt, M. Koerber, and B. Eberle, “Further investigation on laser induced damage threshold of camera sensors and micro-optoelectromechanical systems,” Proc. SPIE 11161, 111610A (2019). [CrossRef]
26. M. Li, G. Jin, Y. Tan, M. Guo, and P. Zhu, “Study on the mechanism of a charge-coupled device detector irradiated by millisecond pulse laser under functional loss,” Appl. Opt. 55, 1257–1261 (2016). [CrossRef]
27. M. Han, J. Nie, K. Sun, X. Wang, and X. Dou, “Experiment on the temporal evolution characteristics of a CCD multilayer structure irradiated by a 1.06 µm continuous laser,” Appl. Opt. 57, 4415–4420 (2018). [CrossRef]
28. B. Schwarz, G. Ritt, M. Koerber, and B. Eberle, “Laser-induced damage threshold of camera sensors and micro-optoelectromechanical systems,” Opt. Eng. 56, 034108 (2017).
29. Handbook of Chemistry and Physics (CRC Press, 2017).
30. https://www.iiviad.com/portfolio/mathematical-simulation-techniques/.
31. V. I. Tatarski, Wave Propagation in a Turbulent Medium (Dover, 1967).
32. E. Brookner, “Improved model for the structure constant variation with altitude,” Appl. Opt. 10, 1960–1962 (1970). [CrossRef]
33. A. Ishimaru, Wave Propagation and Scattering in Random Media (IEEE, 1999).
34. D. H. Tofsted, S. G. O’Brien, and G. T. Vaucher, “An atmospheric turbulence profile model for use in army wargaming application I,” Army Research Lab report ARL-TR-3748 (2006).
35. C. Westgate, How to Determine the Laser-Induced Damage Threshold of 2-D Imaging Arrays (SPIE, 2019).
36. B. E. Deal and A. S. Grove, “General relationship for the thermal oxidation of silicon,” J. Appl. Phys. 36, 3770 (1965). [CrossRef]
37. C. H. Lin, “Oxidation (of silicon),” in Encyclopedia of Microfluidics and Nanofluidics (Springer, 2008).
