1 Integration of a passive micro-mechanical infrared sensor package with a commercial smartphone camera system Nathan Eigenfeld Abstract— This report presents an integration plan for a passive micro-mechanical infrared (IR) sensor system with a standard smart phone camera system. The advantage of utilizing a micromechanical system for IR detection is its ability to be optically readout. If mechanically integrated with the smartphone’s camera and light system, a very simple, electronics-free device could be fabricated and packaged to provide affordable IR imaging to the consumer market. I. PROJECT DEFINITION THERMAL imaging has grown in popularity in the last twenty years with the advancement of semiconductor manufacturing for microelectromechanical systems (MEMS). Currently, consumer IR imaging with 320x240 resolution has a market price-point of $2000-$3000. With obvious high-end military applications such as thermal weapon sights, security, and weapons detection, there is a need for even more affordable devices to provide accessibility to the general public. Consumer access to IR imaging, will produce a wave of new applications not limited to personal security systems or night vision for everyone with a cell-phone. Thus far, uncooled thermal detectors have emerged as the dominant technology for marketable IR imaging. They are affordable, low power devices because no cooling is required to maintain their sensitivity. Specifically, the microbolometer has been the pixel of choice for the last ten years for both high-end military applications and expensive consumer products. It has the ability to be batch produced by standard semi-conductor fabrication processes into focal plane arrays (FPA) and coupled with CMOS circuitry. Microsystems Project Final Report, May 2013 However, theoretical noise limitations are impinging on its further improvement. One emerging device to sub-route such limitations is the micromechanical sensor (Fig. 1). The device consists of a bimorph cantilever, which bends when incident IR radiation is absorbed. These devices may be readout capacitively, or optically, the latter being of great advantage for a lownoise device and specifically for this project. For an optical readout, one of the materials in the bimorph must be reflective in the visible light spectrum, upon which an illuminating light is reflected. Devices in the research realm have utilized high-performance CCD camera systems coupled with Fourier Optic methods to measure the pixel deflections in arrays of micromechanical devices based off interference patterns [1]. As phone cameras become more advanced, their capabilities for readout methods will become viable for this readout method. They may already be rudimentarily capable of performing simple readout algorithms sufficient for consumer applications. Fig.1. Demonstration of the bimorph cantilever’s induced bending upon absorbed power by a thermal expansion coefficient mismatch in the materials [2]. 2 The inherent advantage of an optically readout micro-mechanical device for this project is the complete decoupling of the sensor and electronics. The microsystem will rely on the phone’s camera and light as well as software to manage the readout task, making the self-contained microsystem passive, and a completely mechanical package. This simplifies its integration substantially. It is envisioned the system would resemble current commercial lens attachments for smartphones. This report will propose chip design and mechanical package for the IR imager’s integration with current smartphones, address key integration issues regarding the packaged system and also verifying the feasibility of current camera phone performance to sense minute optical interference changes in the pixel arrays. II. DESIGN CONCEPT The housing and function of this microsystem is completely mechanical. However, it must allow for dual-side functionality, i.e. the chip must be able to view IR light from one-side and optical light from the other (Fig. 2.). The advantage of silicon based IR sensor is silicon’s transparency in the IR. Thus, the backside of the chip may face outwards towards the viewing field. It must also be integrated with an IR lens system, which is generally an anti IR reflecting germanium or sapphire window made into a simple optic. The front side of the chip, facing the phone camera and light, must have an optical lens to focus the camera’s LED light onto the sensor. Once this light is reflected, it is re-focused through the LED optic and onto the camera sensor by the inhouse camera lens (Fig. 2). It is assumed, an image analysis algorithm could be created to determine changes in the phone’s camera pixel intensities, which correspond to a mapping of individual bimorph cantilever deflection magnitudes. This readout scheme produces the composite readout image displayed on the phone. As for the sensor pixels, a very simple 320x240 array (current low-price consumer resolution) is acceptable. Since this device is consumer oriented, high-performance is not of initial concern. The bimorph cantilevers should be optimized for sensitivity, not speed of response (two main performance metrics for IR imaging). This is accomplished by substantially thermally isolating the cantilevers from the substrate to achieve the maximum temperature rise in the structure causing the largest deflection. This will help ensure the phone camera can read the optical interference created by deflecting pixels. The FPA may require specific “angling” as well as the LED lens to ensure the light is focused correctly onto the FPA and reflected into the phone’s camera. The FPA package must be transparent on the LED side to allow visible light transmission back and forth from the LED to FPA and FPA to camera. Fig.2. A simple schematic of components in the microsystem mechanical package including a micro-mechanical pixel. III. KEY INTEGRATION ISSUES Several integration issues arise for this design concept for a practical device. One challenge of this project would be retro-fitting of the mechanical housing to a variety of commercial cameras. Different phones have a variety of lens and light arrangement, from vertical to horizontal camera/light orientation, and one, two or multiple lights in different configurations. It is envisioned that the variances in camera and light orientation across different 3 smartphones could be compensated for in the readout algorithm software rather than creating different microsystem packages for individual phones. The mechanical package would be designed to be a universal “snap” on fit for all phones (Fig. 3.). Fig.3. A simple layout including the active components in the microsystem. The universal mechanical mount (orange) contains the chip sensor with frontside and backside windows. It mounts directly over the camera and light portion of a given phone. The optic for the visible light interference poses another integration issue. It must be advanced enough to be integrated with the camera’s LED to focus light onto the FPA. Thus, the f-number and area of the lens must correspond to realistic dimensions for the mechanical package. The lens must minimize chromatic and spherical aberrations to optimize the readout by the camera and software. An achromatic doublet lens is feasible by its commercial availability and practical use by optical designers. Next, it must be verified that the LED can provide enough power through the lens, to the array so that the reflected power back to the CCD camera is above the photometric exposure for standard CCD pixels. More specifically, the pixels must be able to sense minute differences in the baseline reflected illuminance from the LED by cantilever deflection in the IR FPA (Sec. IV Analysis). Another important consideration for micromechanical device performance is ambient temperature effect. The bimorph cantilevers deflect even in ambient temperature changes, convoluting the actual signal or IR intensity absorbed in the pixel. The temperature sensor of the phone must be utilized to compensate for this in the software package to obtain correct IR intensities from the pixels (Fig. 3.). Further, these devices are sensitive to mechanical vibrations which induce additional noise into the sensor. The accelerometer in the phone could be utilized to sense jerk or shock events and compensate for them in the readout software for more accurate imaging (Fig. 3.). IV. ANALYSIS To ensure this is a feasible IR detection readout method, simple calculations will be done to demonstrate the variances in reflected power by the deformed cantilevers are detectable by a standard CCD camera in a phone. Below is a table with standard dimensions and light intensities. This analysis will avoid complicated noise considerations as it is outside the scope of the project. LED Radius LED Power LED Lens F-number FPA Resolution Pixel Pitch Reflection Coefficient CCD Lens Radius CCD Lens F-number CCD Pixel Pitch CCD Photometric Threshold Frame Rate 15 mm 25 mW 1 320x240 50 μm 0.9 30 mm 1 5 μm 0.1 lux*sec 30 Hz Table 1: Device parameters for assessment of a basic example. These inputs will provide general dimensions for the optic lens radius/placement and FPA placement. Firstly, the average LED power output is 25 mW, which is focused by the LED lens at f=1 onto the FPA. The f-number is the ratio of lens diameter to the focal point. For the lens to capture all the power of the LED, this means the diameter must be ~30mm and the FPA must be ~30 mm away. This results in ~20 nW of 4 reflected power off each pixel in the FPA. Disregarding the angled reflection needed to reach the camera, the corresponding optical path length is 30mm back to the camera lens and ~10 mm focusing length from the CCD lens to the CCD FPA assuming f=1. Fig. 4: Demonstration of basic optical power shifts by IR pixel deflection and detected interference by CCD pixel Now, the relative shift of the baseline reflection upon cantilever deflection must be addressed for the extreme case (Fig. 4). For micromechanical pixels in research the approximate mechanical responsivity of the bimorph cantilever is 0.25 μm/K [3]. For a device capable of sensing 5mK differences in incident radiation (not including noise), the vertical deflection of the cantilever is ~1 nm. This 1 nm deflection is produces a 30μm shift in the reflected ray on the CCD camera lens. After focusing, this corresponds to ~2 μm shift of the ray on the CCD pixel. Given the 20 nW of reflected power, this corresponds to a relative change in power of ~40%. So for the minimum resolvable temperature difference (excluding noise) of 5mK, the CCD pixel will see a 40% change in incident power levels of ~2 nW. For imaging at 30Hz, this is ~100 lux*sec, which is illuminance*seconds. For commercial 5 μm pixel cameras, the minimum lux*sec is on the order or 10 [4]. Thus, the CCD is very capable of detecting the slight deflections of the microcantilevers. This is without noise included, which could decrease performance by 1-2 orders of magnitude. Even so, the minimum detectable temperature difference would still be ~500mK which is acceptable for a low-cost commercial device. Note this simple analysis does not address the actual software algorithm to readout the relative power shifts per pixel. It was merely a reality check for utilizing the in house camera and light to detect optical power shifts from the deflecting cantilevers. V. CONCLUSION A device design for a purely mechanical IR camera cell-phone attachment was presented and analyzed. This cell-phone attachment could revolutionize IR imaging by its accessibility to the general public. Key integration issues revolve around the universality of the mechanical mount to address various phone designs and the optical LED lens integration. This design is feasible with current camera phone/light technology and the resulting work is opto-mechanical packaging. Based off the analysis of optical path lengths, the active components could be < 50 x 50 mm2, leaving room for a mechanical package that could be < 2 x 2 cm2. This simple phone attachment could bring IR imaging to millions of consumers at a < $500 price point. The main price reduction surrounds the lack of electronic component integration. The applications created by exposure to such an immense user-base are envisioned to be endless. References: [1] J. Zhao, “High sensitivity photomechanical MW-LWIR imaging using an uncooled MEMS microcantilever array and optical readout (Invited Paper),” Proceedings of SPIE, vol. 5783, pp. 506-513, 2005. [2] J. Lai, T. Perazzo, Z. Shi, and a Majumdar, “Optimization and performance of highresolution micro-optomechanical thermal sensors,” Sensors and Actuators A: Physical, vol. 58, no. 2, pp. 113-119, Feb. 1997. [3] Scott R. Hunter ; Gregory S. Maurer ; Gregory Simelgor ; Shankar Radhakrishnan ; John Gray; High-sensitivity 25μm and 50μm pitch microcantilever IR imaging arrays. Proc. SPIE 6542, Infrared Technology and Applications XXXIII, 65421F (May 14, 2007). 5 [4] Joyce Farrell ; Feng Xiao ; Sam Kavusi; Resolution and light sensitivity tradeoff with pixel size. Proc. SPIE 6069, Digital Photography II, 60690N (February 10, 2006).