A silicon photomultiplier (SPM) is a new type of semiconductor detector that has the potential to replace the photomultiplier tube (PMT) detector in many applications. In common with a PMT detector, the output of an SPM is an easily detectable current pulse for each detected photon and can be used in both photon counting mode and as an analogue (photocurrent) detector. However, the SPM also has a distinct advantage over PMT detectors. The photon-induced current pulse from a PMT varies greatly from photon to photon, due to the statistics of the PMT multiplication process (excess noise). In contrast, the current pulse from an SPM is identical from photon to photon. This gives the SPM a distinct advantage in photon counting applications as it allows the associated electronics to be greatly simplified. Identical pulses also mean that the SPM can resolve the number of photons in weak optical pulses, so-called photon number resolution. This is critical in a number of applications including linear-optics quantum computing.
A silicon photomultiplier is an array of photon counting microcells connected to a common output[1, 2, 3]. Each photon counting microcell consists of a Geiger-mode photodiode connected in series with an integrated quenching element. A Geiger-mode photodiode is a photodiode engineered to operate above the breakdown voltage of the junction4. Under these conditions, the photodiode is sensitive to single photons and a photo-carrier generated in the depletion region can result in the entire junction breaking down — Geiger breakdown — by avalanche multiplication producing a large pulse of current. Once the photodiode has detected a photon, it must be reset such that it is ready to detect the next photon. Resetting the photodiode after Geiger breakdown is known as quenching and can be achieved by passive elements or active quench circuits. In a silicon photomultiplier, quenching is achieved by an integrated polysilicon series resistor.
Each Geiger-output pulse has a duration of typically 40 — 60ns and contains around a million charge carriers and is usually amplified to a useful level (10 — 100mV) by a simple, fast voltage amplifier. The SPM detector can be used in single photon mode, pulse detection mode, and analogue mode for slowly varying or continuous optical signal above the single photon level.
The main performance parameters of an SPM are its photon detection efficiency (PDE), dynamic range, noise rate, and pulse or gain uniformity.
Photon Detection Efficiency
The PDE is determined by the detection efficiency of the individual photon counting microcells and the geometrical efficiency, or fill factor, and can be written as
where η(λ) is the quantum efficiency of silicon at a given wavelength, ε(V) is the carrier Geiger avalanched initiation probability and F is the SPM fill factor. The quantum efficiency of silicon is 80+% over much of the visible spectrum while the avalanche initiation probability of an electron can be as high as 70%. The quantum efficiency and avalanche initiation probability make up the detection efficiency of the individual photon counting microcells. The fill factor is the ratio of the total active area of all the microcells and the total area of the SPM. The SPM fill factor is determined by design constraints and the size of the microcells. In the SPM design there is a fundamental trade-off between the microcell size, the fill factor and the total number of microcells for a given area. The total number of microcells and the PDE determine the dynamic range of the SPM.
The table below summarizes the trade-off between the microcell size, number of microcells, and the fill factor for a 1mm x 1mm SPM area.
Figure 1 shows the typical spectral response of an SPM detector (SensL part number SPMMini1020) with 1144 20μm x 20μm microcells at 3V above the breakdown voltage. The peak PDE occurs at 490nm and is 15.5% at this bias voltage. By further optimizing the SPM fill factor and maximizing the detection efficiency of the microcells, peak photon detection efficiencies of 35% to 40% are achievable.
Dark Count Rate
The dominant noise in SPM detectors is the dark count rate and has units of counts per second (cps). The dark count rate is the number of measured Geiger pulses per second when the detector is placed in complete darkness.
The dark count results from the thermal emission of carriers from generation-recombination sites in the active region of the device. The thermally generated carriers can also initiate a Geiger breakdown and produce Geiger pulses identical to those produced by a photon. The dark rate is typically measured by the threshold crossing method with the threshold set to half the peak amplitude of the Geiger pulses.
Single Photoelectron Spectrum
The pulse or gain uniformity of an SPM can be assessed by producing the single photoelectron spectrum (SPS). The SPS allows the gain and gain uniformity of the SPM microcells to be calculated. The SPS is a histogram of the output charge from the SPM in response to a fast, weak pulse of light. In the experiment, the optical signal is provided by a fast LED and the Geiger pulses from detected photons are amplified and fed into a charge to digital converter (QDC). The QDC is set to integrate the charge for a certain period, known as the gate, and is triggered to begin recording a short period before the optical flash from the LED. The total charge recorded during the gate is then stored and passed to a PC for analysis. The process is then repeated in order to build up a histogram of the total charge recorded for each pulse of light.
Figure 3 shows a typical single photo-electron spectrum from a 1mm2 SPM. Due to the statistical nature of the optical pulse and of the detection process, there is a non-zero probability that no photons will be detected when the LED flashes.
Another source of spurious Geiger pulses is optical crosstalk5. Optical crosstalk can occur between neighboring microcells as a result of photon emission by a microcell that is undergoing Geiger breakdown. Optical crosstalk occurs practically instantaneously and results in a pulse with amplitude two, three, etc., times the height of a single Geiger pulse. The optical crosstalk is defined in terms of the crosstalk probability. The crosstalk probability is the probability that a microcell undergoing Geiger breakdown will trigger a second microcell to breakdown. The optical crosstalk can be virtually eliminated by blocking the optical path between adjacent microcells. This can be achieved by etching a trench between the microcells and filling it with an opaque material. SPM devices with such optically isolated microcells have been fabricated and assessed. Figure 4 shows a scanning electron microscope image of a trench etched between the microcells of an SPM. The optical crosstalk probability of an SPM with trenches was found to be reduced to approximately 1%.
One application of SPM detectors is confocal microscopy. A confocal microscope uses fluorescence to image biological specimens. It uses a focused laser spot, which is scanned across the specimen, and a pinhole aperture located near the image plane to block fluorescent signals originating from regions of the specimen that are not under illumination by the laser spot. By eliminating the stray light, the image quality is improved but the amount of signal is reduced. As a result the photodetector used in confocal microscopes must be extremely sensitive and respond quickly to rapid changes in intensity. Silicon photomultipliers are an ideal photodetector for confocal imaging microscopes as they are fast, have single photon sensitivity, a peak optical response well matched to the emission wavelengths of the fluorescent probes, and low noise when cooled to moderate temperatures.
Additional applications for SPM detectors can be found in the areas of single photon counting, LIDAR, high-energy physics, homeland security, nuclear medicine, fluorescence detection, and in a multitude of other low-light applications where linear mode APD or PMT detectors are currently used.
For applications that require detection areas larger than 1mm2, SPM detectors with a 9mm2 active area are available. Arrays of SPM detectors can also be produced by flip chip bonding of detectors onto inexpensive glass or ceramic substrates5.
1mm2 SPM detectors are sensitive to single photons at room temperature, provide photon number resolution, and have photon detection efficiency and dark rate performance similar to PMT detectors when cooled to the moderate temperatures attainable with thermo-electric coolers. In addition, SPM detectors have a number of advantages over PMT detectors, namely small size, low bias voltage operation, magnetic field insensitivity, and immunity to damage from exposure to ambient light conditions. SPM detectors can be used in single photon counting mode, pulse mode, and as an analogue detector. The SPMMini1020 has a peak PDE of 15.5% at 490nm and at 3V above the break-down voltage has a dark count rate of the order of 600kcps at -20°C.
- A. G. Stewart, V. Saveliev, S. J. Bellis, D. J. Herbert, P. J. Hughes, J. C. Jackson, Performance of 1mm2 Silicon Photomultipliers, IEEE J. Quantum Electron., Vol. 44, No. 2, pp. 157, Feb 2008
- V. Golovin and V. Saveliev, Novel type of avalanche photodetector with Geiger mode operation, Nucl. Instr. Meth. A, vol. 518, pp. 560-564, 2004
- J. C. Jackson, Silicon Photomultiplier Detectors for Low Light Detection, Photonics Spectra, Dec 2007
- J. O’Keeffe and J. C. Jackson, Laser Focus World, Jan 2006
- P. J. Hughes, D. Herbert, A. Stewart, J. C. Jackson, Tiled Silicon Photomultipliers for large area, low light sensing applications, Proc. Of SPIE: Semiconductor Photodetectors IV, Vol. 6471, 2007