Microelectronic device and method of its manufacture
This invention relates to a microelectronic device and a method of its manufacture. More particularly, it relates to arbitrarily large Geiger mode avalanche photodiodes with an advantageously low dark counting rate, and a method of producing them.
A diode is formed by the fabrication of an n-type semiconductor layer on a p-type layer, the layers usually being suitably doped silicon. As is well-known, diodes can be used for the detection of light and operated in the following operational modes:
1. Photovoltaic operation. In this mode, a small reverse bias is applied to the diode. Incident light generates electron-hole pairs within the diode. These pairs are separated by the depletion region, so generating a current that increases linearly with the incident light. Current arises only through separation of carriers and this means that photovoltaic mode can be used only for detection of a relatively large light intensity.
2. Avalanche photodiode (APD) mode: In this mode, the diode operates with a reverse bias that is close to but not exceeding the breakdown voltage of the device. In APD mode, the reverse bias sets up an electric field within the depletion region. This electric field causes incident electrons and holes to undergo impact ionisation, causing carrier multiplication as they traverse the depletion region. A single carrier entering the depletion region can generate typically 100-1000 additional carriers, which give rise to an easily-detectable current flow within the detector.
3. Geiger mode avalanche photodiode (GM-APD): This mode operates at reverse- diode voltages in excess of the breakdown voltage of a diode. The large electric field imparted by the high voltage bias causes single carriers entering the depletion region to generate a self-sustaining avalanche of current. Even a
single photon of incident light can generate a self-sustaining current flow in the diode.
Operation in GM-APD can produce a very sensitive detector, capable of detecting single photons. However, detection of current flow in this mode does not unambiguously indicate detection of a photon. Current can also arise from the so-called "dark count". The dark count arises from noise, attributed to thermally-generated or defect-generated carriers.
At present, no shallow junction, low voltage single photon counting avalanche photodiodes exist commercially. Experimentation has shown that shallow junction devices have a dark count that does not scale proportionately with increasing the diameter of the photon-sensitive area. This limits the maximum diameter of useful devices to less than approximately 20μm; above that, noise from the dark count becomes excessive. Of course, it is acknowledged that if the gettering process can be improved, it would make larger active area sizes possible and the invention described herein is not limited to such a size. Below the maximum diameter, performance can be excellent. However, devices this small have restricted applicability. For many applications, for example within chemical, biological astronomical apparatus, devices are required that have large effective area for photon counting. At present, available devices have large depletion regions, deep junction formation, and low-doped epitaxial material. This fabrication method is not compatible with CMOS because the devices operate at a relatively large operational breakdown voltage (in excess of 200V), and the devices typically suffer from reduced sensitivity to short wavelengths of light.
It has been shown that the effectiveness with which process-induced defects in the depletion region can be removed by gettering is the limiting factor for producing a successful GM-APD device. For example, see A. Zanchi, F. Zappa, M. Ghioni, A. Guidice, A. P. Morrison and V. S. Sinnis, "Probe detectors for mapping manufacturing defects", Proceedings of 3rd International Caracas Conference on Devices, Circuits and Systems (ICCDCS2000). The gettering process causes undesirable contaminants and defects within the depletion region to migrate to non-active areas of the device, where the electric field is less, and avalanche multiplication does not occur. Once a defect is moved to a non-active region, its negative effects are decreased.
An aim of this invention is to provide shallow junction Geiger mode avalanche photodiodes (GM-APD) of arbitrary size and shape with sufficiently low dark count for single photon detection and a method for producing such a device.
From a first aspect, this invention provides a photodiode structure operable in Geiger mode, the photodiode structure including an effective area comprising a multiplicity of high-field regions at which avalanche multiplication can occur, the high-field regions being sized and spaced apart by low-field regions in the surrounding substrate.
This structure produces a larger effective area than could be provided by known devices. Avalanche multiplication occurs at each of the high-field regions, distributed across the effective area. The comparatively smaller size of each of the high-field regions allows the gettering process to be carried out effectively on all of those individual regions, thereby ensuring that each has a low dark count. This can provide a photodiode of large effective diameter with an acceptably low dark count rate.
Most typically, a dimension of each high-field region is less than 20μm. In preferred embodiments, the dimension of each high-field region is approximately lOμm. For example, each high-field region may have a polygonal or approximately circular section, with a diameter of 20μm or less (e.g. lOμm).
The high-field regions are spaced apart by low-field regions of size sufficient to receive defects and/or contaminants gettered from the active field regions. For example, the high-field regions may be spaced apart by a distance equal to, less than or greater than approximately lOμm. Thus, defects and/or contaminants removed by gettering can be contained within the low-field regions. Defects and contaminants in the low-field regions have comparatively little detrimental effect on the performance of the device and in particular, its dark count.
Embodiments of the invention may incorporate a wide range of shape, size, configuration and spacing of the high-field regions. In fact, one of the advantages of the invention is that the size and layout of the active areas can be tailored to provide a detector of any size or working configuration. Optimal arrangements for any particular
arrangement may be determined by straightforward experiment. Most typically, the arrangement will be chosen to give an acceptably low dark count.
Each of the high-field regions may be formed by embedding or diffusing material into the substrate. The substrate may be a p-type material, in which case each high-field region may be formed as a p-type dopant implant or diffusion. For example, such a dopant may be boron. Alternatively, the substrate may be n-type material, in which case each high-field region may be formed as an n-type dopant implant or diffusion. For example, such a dopant may be phosphorous or arsenic. In either case, the high- field regions are typically covered by a layer of material of the opposite type. Several or all of the high-field regions may be covered by a continuous layer of such material.
Most typically, embodiments of the invention are implemented as shallow junction diodes. It is in this type of structure that the enhancement in gettering effectiveness is most significant. However, the invention may also be embodied in a reach through diode structure, where multiple high field regions may likewise improve the device dark count.
A photodiode embodying the invention typically has a breakdown voltage significantly lower than 200V. For example, it may be in the region of 30V. Devices may be operable in Geiger mode at or around such voltages. Embodiments of the invention may further be CMOS compatible.
From a second aspect, the invention provides a method of fabricating a photodiode structure operable in Geiger mode comprising forming a low-field region of a first type in a surface region of the substrate, forming a multiplicity of high-field of the first type regions within the low-field region, gettering defects and/or contaminants from the high-field regions, and applying a covering layer of opposite type to cover the high field regions and at least part of the low-field region.
If the first type is p-type, then the covering layer will be n-type. Likewise, if the first type is n-type then the covering layer will be p-type. In this way, a depletion zone is formed between each of the high-field regions and the covering layer, each depletion zone acting as a diode element.
Each high-field region is typically formed by implanting or diffusing material into the substrate.
An embodiment of the invention will now be described in detail, by way of example, and with reference to the accompanying drawings, in which:
Figure 1 is a plan view of a photodiode structure embodying the invention; and
Figure 2 is a sectional view of the structure of Figure 1.
With reference to the drawings, a photodiode structure embodying the invention is constructed on a p-type silicon substrate 10. An upper layer 12 of the substrate 10 is boron-doped silicon to form a p-type epi-layer. This upper layer 12 constitutes a low- field region of the photodiode.
Several high-field regions 14 are formed in the upper layer 12 by implantation or diffusion of a high doping of p-type dopant material, (for example, boron) into a surface of the upper layer 12. Any size and number of active areas can be designed into the device. Ideally the active area size and the number of active areas can be tailored for the particular sensing application. To be commercially viable a total effective diameter of 180um is required. The number of smaller high field regions within the 180um or larger effective area will depend on the gettering effeciency of the process and is process dependent. In any case, the high-field regions 14 thereby formed are separate from one another, being spaced-apart in a pattern that will be described blow.
Following creation of the high-field regions 14, they are subjected to a gettering process whereby defects and contaminants are removed from them, to be accumulated within the surrounding layer 12. The gettering process is a well reported process by which contaminants and defects in one region of a device move, or getter, to new locations. Typically the sinker implant provides a site for the defects and contaminants to move to. The act of placing high field regions separated by low field regions allows the gettered defect or contaminant an area outside of the high field region with which to move to. As has been discussed, it is possible to carry out a gettering process that effectively removes defects and contaminants from a high-field region of diameter less than 20μm. In embodiments of the invention, each of the high-field regions is sufficiently small to
enable the gettering process to clear each depletion zone of defects to a satisfactory extent.
A field cover layer 16 of n-type material such as phosphorous or arsenic-doped silicon is then implanted or diffused to cover all of the high-field p-type regions 14. An annular sinker 18 of p+ material is formed to surround the cover layer 16. The entire structure is covered by an oxide layer 20. Metal connections 24 are formed through the oxide layer 20 to make electrical contact with the cover layer 16 and further metal connections 26 make contact with the sinker 18.
As will be seen, a p-n junction is formed between each of the high-field regions 14 and the cover layer 16, each one acting as a diode element, and each one surrounded by a low-field region. The structure thereby provides what is effectively a multiplicity of diode elements in parallel, the high-field p-type regions 14 being anodes and the n-type cover layer 16 being a common cathode and serving as a guard ring between the junctions.
In this embodiment, each of the p-type regions 14 is approximately hemispherical within the upper layer 12, projecting an approximately circular plan of lOμm diameter at the surface of the upper layer. The high-field regions 14 are arranged in a regular array with a spacing of approximately lOμm between adjacent regions. In the illustrated embodiment, the high-field regions are arranged in rows, spaced at a constant pitch, and each row is offset from the next by half of that pitch. This is just one of many possible configurations.
The lower doping of differing regions, either n-type or p-type, sets the breakdown voltage of the device as the depletion region spreads largely through the lower doped region. The applied voltage of the reverse biased pn diode is dropped completely across the depletion region that is set by the junction of the n-type and p-type materials. In the present embodiment a highly doped n-type layer is used as the guard ring layer and cathode. The p-type or anode layer is doped with an opposite species type to the cathode and a lower doping concentration per cmΛ-3. The depletion region, which is caused by the diffusion of holes and electrons to opposite sides of the junction, primarily spreads in the lower doped side of the junction, in this case the p-type side.
The depletion region thickness therefore, is set by the doping level of the lower doped side of the pn junction. The lower the doping of the low doped side, the larger the depletion region width and the larger the breakdown voltage since the electric field, given as V/cm, is reduced. Likewise an increase in doping level for the low doped side, gives rise to an increase in the electric field and a lower breakdown voltage. Typically, doping is implemented to set the breakdown voltage at approximately 30V.
It will be appreciated that an alternative embodiment could readily be constructed with an n-type substrate, n-type high-field regions and a p-type cover layer.
In operation, application of a large reverse bias in excess of the breakdown voltage of the diode elements causes a large electric field to be generated within the depletion region of each diode element. An incoming photon causes an electron-hole pair to be formed within a depletion zone associated with one of the high-field regions. The electric field within the depletion regions separates electrons and holes that enter into the depletion region, imparts energy to them causing them to accelerate and generate extra electron-hole pairs through impact ionisation. This leads to avalanche multiplication in accordance with GM-APD mode operation. The relative absence of defects and contaminants that results from effective gettering ensures that there are comparatively few electron-hole pairs formed by defects or thermal effects, therefore the current arising from dark count is acceptably low.
It should be noted that the invention is not limited to Si detectors and that the invention could work in detectors of different materials. For example, it will be seen that the structure of the invention could be used in, for example but not limited to: SiGe or GaAs detectors.