The present invention relates to radiation sensors, and more particularly, is related to infrared radiation sensors.
Thermopile infrared sensors are well known to possess temperature stability, for example of up to about 300° C., and low power consumption, for example, a U electronic 32×32 microbolometer array with 450 mW power consumption, or an Excelitas 32×32 array with less than 200 mW power consumption. However, since the detectivity of these sensors is known to be low, compared to a conventional resistive microbolometer (a-Si, VOx, herein referred to as “microbolometer”), approaches have been made to compensate this issue by increasing the pixel size. Such sensors with increased pixel sizes include, for example, a sensor having detectivity of 108 cmHz1/2/W and a microbolometer having detectivity of 109 cmHz1/2/W, as well as a thermopile infrared (IR) sensor size of greater than 1 mm×1 mm, compared to a microbolometer of 25 μm×25 μm. Therefore, the etching a cavity in the back side of a substrate below the sensor (“back side etch”) coupled with a front side undercut release process and a bulky absorber on the front-side of the sensor have been incorporated to build sensors, which may be operated without a vacuum.
With the development of new complementary metal-oxide-semiconductor (CMOS) process technology, as known from microbolometer sensors, further miniaturization of the thermopile sensor can be reached by using front-side gas release processes. For example, using XeF2 or Vapor HF, an interferometric absorber, and/or thin film deposition and chip-scale or wafer-level packaging techniques, allows the sensor to reach equivalent microbolometer detectivity capabilities for a relatively low cost price.
Nevertheless, thermopile sensors generally demand high temperature annealing processes in order to activate the dopants of the n-type and p-type silicon, making CMOS readout circuit and common interferometric absorber integration where metal such as aluminum (Al) is used as reflector, impossible.
The sensor of some traditional thermopile sensors may be released from the front-side. However, with these traditional thermopiles, a bulky absorber needs to be implemented, due to the under-etch of the silicon beneath the sensor. Additionally, the lateral etch control is generally imprecise, making an integration of sensors as an array almost impossible.
Therefore, there is a need in the industry to overcome one or more of the abovementioned shortcomings.
Embodiments of the present invention provide a surface micro-machined infrared sensor using a highly temperature stable interferometric absorber. Briefly described, the present invention is directed to a method for manufacturing a surface machined infrared sensor package. Steps of the method include providing a semiconductor substrate having a front side surface and a back side surface, defining a transistor on the substrate front side, implanting a reflector on the substrate front side, forming a sensor on the semiconductor substrate front side adjacent to the reflector, and depositing a thin-film absorber upon the sensor. The thin-film absorber is substantially parallel to the reflector with the sensor disposed there between.
In some embodiments, the transistor may be a CMOS transistor. The distance between the thin-film absorber and the reflector may be approximately one quarter of a radiation wavelength detected by the sensor. The infrared sensor may be a thermopile infrared sensor, a diode-bolometer, or a resistive microbolometer.
Other systems, methods and features of the present invention will be or become apparent to one having ordinary skill in the art upon examining the following drawings and detailed description. It is intended that all such additional systems, methods, and features be included in this description, be within the scope of the present invention and protected by the accompanying claims.
The accompanying drawings are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification. The components in the drawings are not necessarily to scale, emphasis instead being placed upon clearly illustrating the principles of the present invention. The drawings illustrate embodiments of the invention and, together with the description, serve to explain the principals of the invention.
The following definitions are useful for interpreting terms applied to features of the embodiments disclosed herein, and are meant only to define elements within the disclosure. No limitations on terms used within the claims are intended, or should be derived, thereby. Terms used within the appended claims should only be limited by their customary meaning within the applicable arts.
As used within this disclosure, “substantially” means “very nearly,” for example, “substantially uniform” means uniform within normal manufacturing tolerances as would be expected by persons having ordinary skill in the art.
As used within this disclosure, “thin” means less than about 15 nm for TiN, or less than about 15 nm for NiCr, and less than about 200 nm for doped silicon.
As used within this disclosure, a thin film interferometric absorber refers to an interferometric absorber having a thin film formed of a material having a sheet resistance on the order of 377 Ω/sq. For example, the material may be TiN, NiCr, doped silicon, or any other material with a sheet resistance of approximately 377 Ω/sq.
As used within this disclosure, a thin film thermopile/diode-bolometer/resistive microbolometer refers to a sensor membrane which is generally thinner than 2 μm including a sensing material and an absorber layer.
As used within this disclosure, “temperature stable” refers to stability of a material at temperatures of up to around 1100° C.
As used within this disclosure, “increased detectivity” generally refers to a detectivity improvement over a benchmark of 5×108 cmHz1/2/W for approximately 1 mm×1 mm pixel size. For example, a 5×108 cmHz1/2/W for a 65 μm×65 μm pixel size would have area on the order of 200 times less than the 1 mm×1 mm pixel, while providing similar response. Therefore, increased detectivity indicates an improvement of detectivity of up to fifteen times or more.
As used within this disclosure, highly doped silicon refers to a peak doping concentration after annealing of greater than 1020/cm3.
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings. Wherever possible, the same reference numbers are used in the drawings and the description to refer to the same or like parts.
Exemplary embodiments of the present invention implement a novel, high temperature stable interferometric absorber and an aligned process to incorporate one or more CMOS devices, a thermopile IR sensor, and on-wafer vacuum packaging together. Embodiments include a surface micro machined thermopile sensor having an interferometric absorber based on a highly doped silicon reflector (on a silicon bulk wafer) and an impedance matched, IR absorbing layer, for example, a 7 nm Ti—N approximately 377 Ohm/sq, on the sensor, which are separated by one quarter wavelength distance 310 (
Embodiments include a process for constructing a thermopile IR sensor/diode-bolometer, which enable drastic miniaturization and a detectivity boost. By replacing a traditional microbolometer interferometric absorber which uses metal as reflector, with a highly doped silicon reflector, the thermopile IR sensor can be highly optimized and have similar process freedom as a low temperature processed microbolometer.
As shown by block 210 and
As shown by block 240 and
As shown by block 250 and
The reflective behavior of highly doped silicon can be described by a free electron model derived from the Drude model. Herein, free electrons determine the plasma frequency, which leads to reflection of electro-magnetic waves having frequencies lower than the plasma frequency:
where ωp is the plasma frequency, n is the amount of free electrons, e is elementary charge, m is the mass of the electron and ∈0 is the dielectric constant. In order to reach lower wavelengths, the higher the free electron concentration (doping concentration in silicon), the higher the plasma frequency becomes, for example, far infrared.
The reflector 140 is therefore highly temperature stable and able to survive a dopant activation step of the thermopile sensing materials 160, 165. Due to the miniaturized structure, front-side gas release steps, as mentioned above, can be used and therefore eliminate undercut release steps of traditional thermopiles, preventing loss of pixel area and enabling focal plane array (FPA) integration.
By decreasing the bulk of the IR absorber when compared to traditional thermopiles, the thickness of the doped sensing material 160, 165 can be minimized, which in turn decreases the thermal conductivity λth and the thermal capacity Cth of the sensor 100, in order to keep the thermal time constant τth in the optimal value. This provides a significant advantage in terms of sensitivity over traditional sensors.
With the decrease of the thickness of the sensing material over traditional thermopiles, for example, but not limited to, a thickness below 200 nm, the Seebeck coefficient may significantly increase and the thermal conductivity may decrease (as known from silicon nanowires) and therefore enhance the sensitivity. For example, decreasing the thickness of the sensing material from 300 nm to 150 nm will increase the detectivity by around two times. The relationship of thickness to detectivity is an approximately linear. A decrease in thickness from 300 nm to 50 nm will implement a detectivity increase of nearly 6 times.
The surface micro machined thermopile sensor process 200 is suitable for the concurrent integration of CMOS circuitry and on-wafer vacuum packaging and can be applied to diode-bolometers, and/or novel resistive microbolometers such as poly-crystalline microbolometers or other semi conductive sensing materials in the same manner, which require high temperature processes after forming the reflector. As shown by
As shown by block 540 and
The abovementioned embodiments solve the problem of a bulky absorber and large area pixel(s) for thermopile/diode-bolometer infrared sensors and enable drastic miniaturization. By using a highly doped silicon reflector 140 for an interferometric absorber, high temperature processes to fabricate the thermopile sensor 100 can follow without destroying the reflector 140. CMOS circuitry and on-wafer vacuum packaging techniques can be well integrated within the thermopile IR sensor process 200 using this novel approach of the doped silicon reflector 140. At the same time, the absorber 175 may be used as an electrical connection between the n-type silicon 160 and the p-type sensing silicon 165, thereby increasing the absorption area, since no metal is necessary for the connection. The release step of such a surface micro-machined structure may be done directly before chip scale packaging, therefore decreasing the possibility of damaged sensors due to handling. The same idea can be applied to production of a diode-bolometer and/or a resistive microbolometer, in order to avoid complex wafer bonding techniques and allow high temperature processes after creating the reflector.
An objective of the above embodiments is to achieve a new way of constructing a thermopile IR sensor/diode-bolometer, which enables drastic miniaturization and sensitivity boost. For example, compared with a detectivity of 5×108 cmHz1/2/W for approximately a 1 mm×1 mm pixel size for a traditional thermopile IR sensor/diode-bolometer, under the first embodiment detectivity may be approximately 5×108 cmHz1/2/W for a 65 μm×65 μm pixel size, corresponding to a decrease in area of about 200 times less, with a detectivity increase of greater than 15 times. In general, detectors under the first embodiment having significantly less area may provide a similar detectivity response. With the replacement of the traditional microbolometer interferometric absorber, which uses metal as reflector, by a highly doped silicon reflector, the thermopile IR sensor can be highly optimized and provide similar process freedom of a low temperature processed microbolometer. Therefore the performance of this sensor may be considered to be almost equivalent to a conventional microbolometer, having low power consumption and high temperature stability.
Traditional thermopile IR sensors, as mentioned above, use a bulky absorber and back side release processes, which can be eliminated by the first embodiment. Additionally, complex wafer bonding processes for the diode bolometer can be eliminated. New types of novel resistive microbolometers can be created, for example, a poly-crystalline microbolometer, due to high temperature stability of the absorber.
The above embodiments are based on an interferometric absorber made of a highly doped silicon reflector coupled to a thin absorber (TiN, NiCr) within the thermopile/bolometer structure, wherein the distance of both layers should have an optical length of a quarter wavelength. For example, the thin absorber of the first embodiment may be on the order of 7 nm for TiN or less than 200 nm for doped silicon, compared with a traditional absorber, which may be on the order of 3.5 μm thick.
The absorber itself has a sheet-resistance on the order of approximately 200 Ω/sq-1000 Ω/sq, preferably 377 Ω/sq, in order to match the atmospheric impedance. Therefore, reflection on this layer can be minimized and an absorption of approximately 50% can be achieved in the first step. The rest of the IR light will be reflected by the underlying highly doped silicon reflector and again absorbed in its maximum of the electromagnetic wave (in this case the maximum of the reflected wave is at κ/4, which is the optical length between absorber and reflector). With this principle, reflection of up to 85% or more of the IR light within a wavelength of 8-14 μm can be achieved. This results in a highly doped silicon reflector which is highly temperature stable compared to the common metal reflector. Therefore, processes like doping activation of sensing the polysilicon layer can follow without destroying the reflector. Additionally, the amount of absorption can be modified with variation of the doping concentration.
Since no back side release or under-etching of silicon is necessary under the above embodiments, a highly controlled gas release processes can be applied, as known from microbolometers, for example, XeF2 or Vapor HF (VHF) microbolometers, giving control in both the vertical and lateral direction.
By decreasing the mass of the thermopile sensor due to the thin absorber, sensing silicon and suspension arms can be made much thinner, providing advantages in terms of thermal conductivity and Seebeck coefficient. As known from nanowires, the Seebeck coefficient increases with structures thinner than, for example, but not limited to, 100 nm.
Another aspect of the absorber is that the sensor membrane can be released directly before capping of the lid, using chip scale packaging techniques. Therefore, destruction of the sensing membrane can be minimized due to handling or further process steps. Overall, the process flow of the above embodiments is aligned with CMOS integration, so that the integration to an array can be realized. For example, embodiments may include an array of pixels.
In summary, it will be apparent to those skilled in the art that various modifications and variations can be made to the structure of the present invention without departing from the scope or spirit of the invention. In view of the foregoing, it is intended that the present invention cover modifications and variations of this invention provided they fall within the scope of the following claims and their equivalents.