CMOS silicon avalanche photodiodes for NIR light detection: a survey

Analog Integr Circ Sig Process (2012) 70:1–13 DOI 10.1007/s10470-011-9641-6 CMOS silicon avalanche photodiodes for NIR light detection: a survey Afrin Sultana • Ehsan Kamrani • Mohamad Sawan Received: 27 October 2010 / Revised: 15 January 2011 / Accepted: 5 April 2011 / Published online: 17 April 2011 Ó Springer Science+Business Media, LLC 2011 Abstract This paper surveys recent research on CMOS silicon avalanche photodiodes (SiAPD) and presents the design of a SiAPD based photoreceiver dedicated to nearinfrared spectroscopy (NIRS) application. Near-infrared spectroscopy provides an inexpensive, non-invasive, and portable means to image brain function, and is one of the most efficient diagnostic techniques of different neurological diseases. In NIRS system, brain tissue is penetrated by near-infrared (NIR) radiation and the reflected signal is captured by a photodiode. Since the reflected NIR signal has very low amplitude, SiAPD is a better choice than regular photodiode for NIR signal detection due to SiAPD‘s ability to amplify the photo generated signal by avalanche multiplication. Design requirements of using CMOS SiAPDs for NIR light detection are discussed, and the challenges of fabricating SiAPDs using standard CMOS process are addressed. Performances of state-ofthe-art CMOS SiAPDs with different device structures are summarized and compared. The efficacy of the proposed SiAPD based photoreceiver is confirmed by post layout simulation. Finally, the SiAPD and its associated circuits has been implemented in one chip using 0.35 lm standard CMOS technology for an integrated NIRS system. Keywords Silicon avalanche photodiode (SiAPD)  Near-infrared spectroscopy (NIRS)  Transimpedance amplifier (TIA)  Clinical imaging  Guard ring structure A. Sultana (&)  E. Kamrani  M. Sawan Polystim Neurotechnologies Laboratory, Department of Electrical Engineering, Ecole Polytechnique de Montreal, Quebec H3T 1J4, Canada e-mail: afrin.sultana@polymtl.ca 1 Introduction Silicon avalanche photodiode (SiAPD) is a solid state photoconductor device where a p-n junction is operated under high-reverse bias voltage. Incident light photons are absorbed in the depletion region of the reverse biased p-n junction and converted into electron–hole pairs (EHPs). These primary EHPs drift along the applied high electric field (E [ 105 V/cm) and they gain enough kinetic energy to create secondary EHPs along their paths by impact ionization. The impact ionization process leads to an exponential increase in the number of EHPs with traversed distance, that is, in avalanche multiplication gain. Compared to a p-i-n photodiode, SiAPDs have a bias dependent internal gain which makes them compatible for low-level light detection in the visible and near-infrared (NIR) regions. Silicon avalanche photodiodes are more advantageous than photomultiplier tubes (PMTs) which are the most commonly used photo detectors due to the drawbacks of their vacuum tube technology. The PMTs are bulky, subtle, sensitive to magnetic fields, and require high voltage supply [1]. As such, SiAPDs become popular for several applications including light detection and ranging (lidar), photon counting, and fiber optic communication [1]. They are potential candidate for applications like quantum cryptography, profilometry of remote objects, fluorescence spectroscopy, and biomedical imaging systems such as positron emission tomography (PET), single photon emission computed tomography (SPECT), and near-infrared spectroscopy (NIRS) [2]. Near-infrared spectroscopy is an inexpensive, noninvasive and portable imaging technique to monitor the brain function and biological tissues. In this technique brain tissue is penetrated by near-infrared (NIR) radiation and the reflected signal is observed to investigate the 123 2 Analog Integr Circ Sig Process (2012) 70:1–13 2.1 Near-infrared spectroscopy system for clinical imaging To analyze brain function NIR light is emitted on the surface of the head which is penetrated by the brain tissue and passed through the skull and brain. Brain function can be studied by monitoring two major types of signals: neuronal signal and hemodynamic signal [7]. Neuronal signal is fast and arises due to optical changes directly associated with the neuronal activity of brain. Neuronal signal describes changes in light scattering properties of the brain tissues or neurons within 200 ms after the onset of the brain stimulation by NIR light [8, 9]. On the other hand, slow hemodynamic signal arises due to the brain activity which leads to an increase in the local oxygen consumption followed by an increase in blood flow of activated neurons due to neurovascular coupling [7]. The increase in blood flow changes the haemoglobin concentrations and oxygenation of blood. As a result of increase in blood flow in the brain, less light will pass through brain tissues. These changes occur within a few seconds after brain stimulation [9, 10]. Near-infrared spectroscopy is the only neuroimaging system capable of detecting both fast neuronal signal and slow hemodynamic signal [7]. The other brain monitoring methods such as EEG and MEG can only measure the fast neuronal signal, and PET and fMRI can measure only the slow hemodynamic signal. The other advantages of NIRS system are: painless, no need to use ionizing radiation, immune to electromagnetic interference, and capable of real-time, long term bedside monitoring [11]. All these advantages have made NIRS the system of choice for neuroimaging and a hot topic of research. Work is going on in different research groups to improve the performance of NIRS system. A simplified block diagram of such device is shown in Fig. 1. The NIRS system consists of NIR light source, sensor or photodetector, data acquisition and control unit, and a processing unit. The light source is placed on the surface of the head (scalp) and it generates light in the NIR range (wavelength is from 650 to 950 nm). Generally, the light sources used for NIRS system are either LEDs or laser diodes that emit NIR light with optical power around Clinical imaging by monitoring human brain function is considered as the most efficient technique of diagnosis and investigation of different neurological diseases, such as, stroke and epilepsy. The commonly used non-invasive neuroimaging techniques are electro-encephalography (EEG), magneto-encephalography (MEG), PET, functional magnetic resonance imaging (fMRI), and near-infrared spectroscopy (NIRS), which is the subject of the present paper [5]. Near-infrared spectroscopy is a technique where brain tissue is penetrated by NIR radiation and the resultant absorption and scattering effects are observed to investigate the brain’s function [6]. Fig. 1 Block diagram of an NIRS system brain’s function. In NIR range, water has relatively low absorption while oxy- and deoxy-haemoglobin have high absorption. Due to these properties, NIR light can penetrate biological tissues in the range of 0.5–3 cm allowing investigation of deep brain tissues, and ability to differentiate between healthy and diseased tissues [3]. Current commercially available NIRS devices are too bulky to be wearable or portable for monitoring brain function [4]. Our goal is to build a novel highly sensitive fully integrated multi-channel wireless front-end receiver for a NIRS system. One of the most important topics of research for NIRS front-end receiver is to design a sensitive photodetector to ensure maximum detection of the reflected NIR light. The fraction of the incident NIR light photons survive to return to the photodetector are strongly attenuated (7–9 orders of magnitude) by the biological tissues. Accordingly, the photodetector requires to be highly sensitive, enabling the reliable conversion of the ultra-low amplitude light signal into a detectable electric signal. SiAPD is a potential candidate for low-level light detection due to its ability to amplify the photogenerated signal by avalanche multiplication. In this paper, first we briefly describe the NIRS system for neuroimaging, and discuss the basic operation and associated circuits of a SiAPD. In the next section we define the design parameters of SiAPDs for NIRS application. Then we discuss the pros and cons of the available SiAPD fabrication technologies and also compare the structure of existing SiAPDs. We propose a device structure as well as a transimpedance amplifier (TIA) circuit suitable for NIRS application and demonstrate the potential performance of the proposed SiAPD by both device and circuit level simulation. Finally we implement the SiAPD and its associated circuits on the same chip using 0.35 lm standard CMOS technology. 2 Background 123 Analog Integr Circ Sig Process (2012) 70:1–13 5 mW. The sensor is a photodetector that monitors the intensity of the reflected NIR signal. The data acquisition and control unit manages all sensors and light source, synchronizes their operation, and acquires corresponding real-time data. The processing unit records, visualizes, and analyzes the acquired data. An integrated NIRS system offers excellent performance and high sensitivity to obtain required information from the brain tissues. 2.2 Operations and circuits of SiAPD The structure of SiAPD along with its depletion region is shown in Fig. 2. Depending on the magnitude of the reverse bias voltage across the p-n junction, SiAPDs can work in two different modes: Linear or proportional mode, and Geiger or single photon counting mode. In linear mode, the reverse bias voltage is kept below the breakdown voltage (VBR) of the junction and the photogenerated charges are amplified with a finite multiplication gain. Here, the statistical variations of the finite multiplication gain produce a noise contribution known as excess noise preventing linear SiAPDs from single photon detection. For linear mode operation, SiAPD requires a TIA to convert the input photocurrent into a voltage signal. Although TIA offers large gain, signal produced by TIA needs to go through LA which boosts the voltage swings and matches the output impedance to drive the output (usually DMUX) [12]. The LA circuit is implemented by cascading two stages of resistive load differential amplifiers and one stage of buffer [12]. The OTA preamplifier is used to selectively amplify the low amplitude signal before it is being filtered and then demodulated. Here we have used OTA proposed in [3]. The OTA differential input pairs are in a class AB configuration, and uses DTMOS devices for input common-mode range enhancement. There are many topologies of TIA reported in literature which has high transimpedance gain and high bandwidth, accompanying with low input noise and low-power consumption [12, 13]. Among them three widely used topologies are: Fig. 2 General structure of SiAPD with its depletion region 3 common-gate TIA, resistive feedback TIA, and capacitive feedback TIA. Figure 3 shows the schematic of three configurations while the detail description of them can be found in [13]. A common-gate configuration is typically chosen as it can tolerate a wide range of SiAPD capacitance. However, resistive feedback architecture has better noise performance and is more attractive when SiAPD models are readily available. Phang et al. proposed a TIA combining a sub 1-V current mirror and a common-gate TIA based on a current-gain amplifier for optical communication [13, 14]. Achigui et al. modified this TIA by adding an OTA with dynamic threshold transistor (DTMOS) for NIRS front-end photo receiver [3]. On the other hand, in Geiger mode, the reverse bias voltage is kept above VBR and SiAPDs work as trigger devices rather than amplifying devices. In this later mode, the electric field is so high that a single photo-generated carrier can trigger a self-sustaining avalanche process that swiftly builds up a macroscopic current in the mA range. The current keeps flowing until the avalanche is quenched by lowering the bias voltage below VBR by dedicated quenching circuits [15, 16]. Avalanche photodiodes operated in Geiger mode are known as silicon single photon avalanche diode (silicon SPAD). Unlike linear mode SiAPDs, silicon SPADs are not concerned with gain fluctuations since here the gain is virtually infinite. Additionally, in Geiger mode, the signal amplitude does not provide intensity information since all the current pulses have the same amplitude. Intensity information is obtained by counting the pulses during a certain time frame or by measuring the average time interval between successive pulses. For operating the SiAPD in Geiger mode (single photon counting mode) quenching and reset circuits are necessary [17, 18]. Generally three types of quenching circuits have been used for Geiger mode operation: passive, active and mixed quenching circuits (MQCs). Passive quenching circuit (PQC) uses the simplest method of quenching an avalanche. It has two possible configurations: voltage-mode output and current-mode output. The voltage-mode provides longer pulses, which might be convenient to visualize them in the oscilloscope but might hinder high speed detection. On the other hand, the current-mode output configuration allows high detection rates [38, 39]. The schematic of a typical PQC circuit is shown in Fig. 4. Figure 4(a) is configuration with current-mode output and Fig. 4(b) is configuration with voltage-mode output [38, 39]. Here RL is the load resistor (typically between 50 and 500 KX) and RS provides matched termination for a coaxial cable (*50 X). When no current flows in the circuit, the SPAD is reversed biased at Vdd. The difference between this voltage and the photodiode’s breakdown voltage, VBR, is called excess bias, VE, and is given by: 123 4 Analog Integr Circ Sig Process (2012) 70:1–13 Fig. 3 Different TIA structures: a common-gate, b resistive feedback, and c capacitive feedback VE ¼ Vdd  VBR : ð1Þ As soon as a light photon is absorbed in the photodiode, the avalanche multiplication process begins and the avalanche current quickly rises to few mA, generating a voltage drop across the resistance that eventually reduces the reverse bias voltage below VBR, thus quenching the multiplication process. Passive quenching circuit is easy to implement and an effective method of quenching for small area (\50 lm) photodiodes since they have very low photodiode capacitance. Due to low photodiode capacitance, small area photodiodes facilitate high speed operation. PQC can be used for large area photodiodes if high speed operation is not a requirement. In order to achieve faster quenching, active quenching circuit (AQC) is used since it forces the SPAD to drop the reverse bias voltage much quicker. Faster quenching results in lower power loss and hence less heating of the SPAD. However, AQC involves complex circuitry and stringent circuit requirements. In order to relax the circuit design requirements, a hybrid approach, known as MQC, can be employed [18]. Here, passive quenching is used as the first stage to limit the avalanche current to a low value, followed by the application of a quench pulse during the quench delay time and a reset pulse to recharge the SPAD back to the reverse bias voltage higher than VBR. This method adopts a simpler design. In MQC, performance limitations are largely dependent on switching delays, which are directly related to the parasitic capacitances of the circuit [2, 18]. The circuit diagram of a typical AQC mixed with PQC is shown in Fig. 5. This scheme has been originally proposed in [40, 41]. In quiescence condition, the cathode of SPAD is biased to Vdd through R1 and is ready to detect a photon. The onset of the avalanche current starts a passive quenching action and the voltage drop across R1 reduces the voltage at the SPAD cathode. As such, Ssense goes in deeper conduction and the voltage drop caused by R3, turns the quench transistors (Squench1 and Squench2) ON via Sfeedback. This starts the active quenching action by quickly pulling the SPADs cathode down to ground. This brings the reverse bias of the SPAD below breakdown and the avalanche current quickly dissipates. The quench transistors (Squench1 and Squench2) are then turned OFF and the three parallel reset transistors (Sreset1, Sreset2, and Sreset3) are turned ON. The reset transistors are activated by an output pulse from the reset monostable which triggers with the end of the hold-off period. These reset transistors are equivalent of the three low resistance transistors, which resets the quiescent bias of the SPAD and brings the SPAD’s cathode voltage back to detect the next photon. Short duration of the reset time decreases the dead time between photon counts [42]. 3 Design parameters of SiAPD for NIRS application (a) (b) Fig. 4 Schematic of a passive quenching circuit (PQC) with two possible configurations [38, 39] 123 The figures of merit to assess a NIRS system are its size, portability and cost. The main advantages of using SiAPD based detector are small size, ruggedness, low operating voltages, and low cost. With the recent advances in CMOS technology, it is now possible to develop SiAPD with the necessary peripheral circuits on the same chip in order to realize an integrated, ultra-sensitive, portable and inexpensive NIRS system. The design requirements of the SiAPDs for NIRS system such as, area, depletion region Analog Integr Circ Sig Process (2012) 70:1–13 5 3.2 Depletion region thickness of the SiAPD The thickness of depletion region of SiAPD, W, is given by [23]: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   2eS 1 1 ðui  Va Þ: ð2Þ þ W¼ q Na Nd Fig. 5 Circuit diagram of a typical active quenching circuit (AQC) thickness, and signal to noise ratio (SNR) are discussed in next paragraphs. 3.1 Area of the SiAPD According to American National Standard Institute’s Laser Safety Standard (ANSI Z136), the brain tissue exposition limit is typically 200 mW/cm2 at NIR light to prevent any biological hazards [19]. The amount of reflected NIR light photons detected by the photodetector depends on the power of the NIR light source, the attenuation due to biological tissue, and the light source-to-photodetector distance. The level of attenuation with a light source-tophotodetector distance of 4 cm, for a five-layer (scalp, skull, cerebrospinal fluid (CSF) layer, and gray and white matters) head model, can be approximated by 4.12 9 10-4 cm-2 [20]. If we consider an NIR light source of 10 mW, the optical power seen by the photodetector is only 4.12 lW/cm2. Thus, the area of the photodetector should be large enough so that it can capture enough optical photon to generate detectable electric signal. Earlier silicon p-i-n photodiodes were successfully used for NIRS systems with an active area of *7.5 mm2 [7, 20]. Silicon avalanche photodiodes are also commercially available with comparable active area and are being used in several NIRS systems, such as, Hamamatsu C5460-01 device has a large active area of 7 mm2 [5, 6]. However, these commercial SiAPDs necessitate a dedicated fabrication process, and cannot be fabricated on the same chip with the rest of the NIRS device. As such, fibre optic bundles are used to guide the NIR light photons from the scalp to the photodetector which results in huge loss (up to 40%) of optical signal. On the other hand, the performance of SiAPDs fabricated using standard CMOS process degrades with increase of its area which limits its active area to 0.003 mm2 [21]. Recent advancements in the standard CMOS fabrication process allow producing SiAPDs of active area up to 0.3 mm2 (*200 lm diameter) while maintaining excellent performance [22]. Here, es (= 1.06 9 10-12 F/cm) is the dielectric constant of silicon, q (= 1.6 9 10-19) is the unit charge, Na and Nd are doping concentration for p and n type silicon respectively, ui (=0.7 V) is the built in potential of silicon and Va is the applied bias voltage. W can be varied either by changing the reverse bias voltage across the photodiode or by changing the doping concentration of p and n type silicon. Figure 6 shows the variation of depletion region thickness of a p ? n photodiode (Na = 5 9 1020 cm-3, Nd = 1.28 9 1017 cm-3) as a function of reverse bias voltage. The wavelengths of interest for NIRS system is from 650 nm to 950 nm range. For a NIR light source with optical power Pin at a wavelength k, the input light photon flux, Gin, is given by [1]: Gin ðkÞ ¼ Pin k; hc ð3Þ where h is Planck’s constant, and c is the speed of light in vacuum. The input NIR light photons, Gin(k), are attenuated by the brain tissues. Gatt ðkÞ ¼ Gin ðkÞ  K; ð4Þ where K is the attenuation constant of the brain tissues. The attenuated light photons, Gatt(k) are incident on the SiAPD and absorbed in the depletion region of the photodiode. The fraction of Gatt(k) that are absorbed in the depletion region of SiAPD depends on its photon absorption efficiency, g(k), which is calculated as [24]: gðkÞ ¼ 1  eaðkÞW ; ð5Þ where W is the thickness of depletion region of SiAPD and a(k) is the absorption coefficient of silicon which is a strong function of wavelength of light photons [25]. A very small percentage of light is absorbed in the neutral region of p?. The variation of a(k) of silicon within the NIR wavelength range is shown in Fig. 7. The photon absorption efficiency, g(k), of silicon for different thickness of depletion region can be calculated using Eq. 5 and plotted in Fig. 8. It is apparent from Eq. 5 and Fig. 8 that for a definite depletion region thickness, the photon absorption efficiency is better at shorter wavelength. Again, the higher the thickness of depletion region of SiAPD, the better is its photon absorption efficiency. However, a thick depletion region increases the noise of the SiAPD [24]. To ensure reasonable amount of photon absorption (*70%) in NIR range, it is of primary 123 Analog Integr Circ Sig Process (2012) 70:1–13 16 4.51 4.01 3 Depletion Region Thickness (µm) 14 Absorp. coeff,α (λ) (10 /cm) 6 12 10 8 6 4 2 0 0 5 10 15 20 3.3 Signal to noise ratio of the SiAPD 2.51 2.01 1.51 1.01 0.51 700 800 900 1000 Wavelength, λ (nm) Fig. 7 Absorption coefficient, a(k), of silicon as function of wavelength [25] Photon Absorp. Effic., (λ) importance to design SiAPDs with at least 10 lm thick depletion region. 3.01 0.01 600 Reverse Bias Voltage (V) Fig. 6 Variation of depletion region thickness of SiAPD as a function of reverse bias voltage 3.51 1.2 W_1_um W_10_um 1 W_20_um W_50_um 0.8 For NIRS system, the input optical signal is strongly attenuated by the brain tissues, consequently the photodetector have to detect a low-level light signal. For this purpose, SiAPDs offer substantial advantages over normal photodiodes since the SNR of SiAPD for low level signal is determined by thermal or amplifier noise and not by the shot noise in the photodiode current [26]. However, since avalanche multiplication is caused by carrier impact ionization events which occur with statistically distributed probability, it is intrinsically noisy. As such, the performance of SiAPD is degraded by a factor known as excess noise factor, F. Both excess noise and photocurrent of SiAPD increase as the avalanche gain increase, so the best SNR occurs at a certain gain. A SNR of *40 dB is needed for NIRS application [20]. Silicon avalanche photodiodes with dark current in nA range, and the generated photocurrent in hundreds of lA range confirms SNR of much higher than 40 dB. This special doping scheme hinders SiAPD’s integration in standard CMOS technologies. It will be compact and costeffective to integrate SiAPDs and other electronic circuits of an NIRS front-end receiver on the same chip using standard CMOS process. 4 Design issues of SiAPD 4.1 Comparison between dedicated and standard technology Silicon avalanche photodiodes are commercially available with reasonable characteristics and are being used in several NIRS systems [5–7]. Silicon avalanche photodiodes work in avalanche mode which necessitates application of high voltage across the device. As such, there is a risk of premature breakdown of the device, particularly at the peripheral junction since higher electric field exists at the periphery. To circumvent the risk of device breakdown, generally a guard ring structure is implemented where the peripheral junction has a lower doped region than the active junction because a slightly lower doped region has lower electric field compared to a heavily doped one [27]. There are two approaches of design and fabrication of SiAPDs: one is to use highly optimized dedicated processes to achieve excellent device performance and the other is to adapt SiAPD design to existing standard CMOS processes to reduce cost and to maximize miniaturization [28, 29]. In the sixties and seventies, many active research groups developed various structures of SiAPDs such as the Reachthrough structures and the Beveled-edge structures and fabricated them using dedicated processes with outstanding characteristics [28]. Silicon avalanche photodiodes fabricated using dedicated process can have low-doped p and n layer resulting in wide depletion region extending from 123 0.6 0.4 0.2 0 600 700 800 900 1000 Wavelength, λ (nm) Fig. 8 Photon absorption efficiency of silicon calculated using Eq. 5 for different thickness of depletion region (W) as function of wavelength Analog Integr Circ Sig Process (2012) 70:1–13 Fig. 9 Schematic of the CMOS SiAPD cross section (not to scale) showing p-well implantation acts as a guard ring, preventing premature edge breakdown [31] the cathode to the anode [28]. Due to the availability of wide depletion region, they are efficient for absorption of red and NIR photons. Nevertheless, the SiAPDs fabricated in dedicated process have two major disadvantages: the production cost is very high due to the specialized fabrication process, and the impossibility to integrate electronic circuits on the same chip. Later several dedicated SiAPD fabrication technologies were proposed which are compatible with the fabrication of CMOS circuits, and therefore, monolithic integration of SPAD devices and CMOS circuits became possible [22]. However, optimizing the performance of both the CMOS devices and the SiAPD is a non-trivial job. To overcome these problems, researchers have investigated the design and fabrication of SiAPDs in a standard CMOS process [16, 30, 31]. The advantages of standard CMOS fabrication process are: the availability of a fully supported, mature and reliable technology at reasonably low cost, and the possibility of developing a complete system on chip with a high degree of complexity [22]. The mandatory requirement for SiAPD fabrication in standard CMOS process is that a suitable subset of CMOS fabrication process flow should be able to build a planar p-n junction without device breakdown at the photodiode periphery [31, 32]. Silicon avalanche photodiodes fabricated using standard CMOS process involves high doped p or n layer resulting in shallow or medium depth depletion region. As a consequence, CMOS SiAPDs are inefficient to detect red and NIR photons, and are not suitable for NIR signal detection in neuroimaging. However, to increase the use of SiAPD based front-end receivers for biomedical applications, integration of the SiAPD and peripheral circuitry on the same chip using standard CMOS technology is highly desired. 7 necessary peripheral circuits on the same chip for an integrated system. However, it is challenging to make SiAPDs in CMOS technology due to lack of special fabrication steps. In a CMOS process, the only available layers are the n? and p? source/drain regions, the n-well and p-substrate (or p-well and n-substrate). In a twin-tub CMOS process, both p-well and n-well regions are available. Figure 9 shows the most commonly used SiAPD structure fabricated using a twin-tub CMOS process. The device consists of a shallow p?/deep n-well junction surrounded by a shallow p-well acting as a guard ring to prevent edge breakdown [31]. By relying on the p-well guard ring, the breakdown voltage of the active area can be above 20–25 V [31]. In a single-tub CMOS process, the guard ring cannot be realized by inserting a narrow p-well at the peripheral junction since p-well layer is not hand-drawn and is present everywhere that there is no n-well. The solution is to split the n-well into two n-tubs separated by a small interval d constituting the guard ring [27, 30]. The guard ring is formed by a weakly doped n-ring due to the n-wells’ lateral diffusion that decreases the breakdown voltage of the peripheral junction (see Fig. 10). Another approach is to use shallow trench isolation (STI) as a guard ring to withstand the high electric fields between the anode and cathode well (see Fig. 11) [33, 34]. Here the edges of the drain implant are confined by the oxide trench and formation of the curved edges is prevented. As a result, a uniform field is achieved more compactly than with a diffused p-well ring. However, STI dramatically increases the density of deep-level carrier generation centers at its interface [34]. Since the active region of the SiAPD is in direct contact with the STI, the injection of free carriers into the sensitive region of the detector results in a very high dark count rate (DCR) and degrades the performance of SiAPD (Fig. 12). 4.3 Comparison of the CMOS SiAPDs performance Several research groups have fabricated SiAPDs using standard CMOS technology. The area and design of the SiAPDs for different CMOS technologies are different 4.2 Existing CMOS SiAPD structures The fabrication of SiAPDs in standard CMOS technology permits fabrication of both the photodetector and the Fig. 10 Schematic of the CMOS SiAPD cross section (not to scale) with guard ring developed by lateral diffusion n-well [27] 123 8 Analog Integr Circ Sig Process (2012) 70:1–13 Fig. 11 STI-bounded SPAD [33] Fig. 12 Block diagram of proposed front-end photoreceiver working in linear and Geiger mode which results in a wide range of performance matrix of SiAPD as depicted in Table 1. It is evident from the Table that photon detection efficiency is better for larger area SiAPDs. Silicon avalanche photodiodes with larger area can more suitably be designed using older CMOS technology. However, use of older CMOS technology will increase area and power consumption for rest of the electronic circuits of the NIRS front-end receiver. On the other hand, doping concentration levels in CMOS increase as the technology advances, causing an increase in the peak electric field in the depletion region and decrease in the breakdown voltage of the diodes. It is clear from the Table that advanced CMOS technology offers SiAPDs with low VBR ensuring safer operating condition for biomedical applications. Therefore, we proposed following front-end photo receiver structure consists of SiAPD and its associated circuits for both modes. We can vary the bias voltage (Vbias) across the SiAPD and operate in either of the modes. 5.1 Proposed device structure CMOS SiAPDs with wide depletion region ([10 lm) are appropriate for NIR light absorption which necessitates designing the SiAPDs with low-doped layers. In this paper, we proposed a new SiAPD structure using relatively lowdoped layers available in standard 0.35 lm CMOS technology. The design specifications of SiAPD for NIRS application is described in Sect. 3 and the design parameters for the proposed SiAPD are summarized in Table 2. 5.1.1 Simulation results of the proposed siapd device 5 Implementation of the proposed front-end Our aim is to design a highly sensitive SiAPD for NIRS front-end receiver to ensure maximum detection of the reflected NIR light. We would like to investigate the performance of the SiAPD in both Linear and Geiger mode. 123 Optimization of the performance of SiAPD is done by device level simulation using Sentaurus TCAD software. The active junction of the photodiode exists between p? (Na = 5 9 1019 cm-3) and deep n-well (Nd = 1.28 9 1017 cm-3). The doping concentrations for these layers are Analog Integr Circ Sig Process (2012) 70:1–13 9 Table 1 Comparison of different CMOS SiAPDs Technology Operating Mode Area Photon detection efficiency Breakdown voltage, VBR 0.13 lm [35] Geiger Octagonal 5% at 800 nm 10 V 0.18 lm [2] Geiger Octagonal 1% at 700 nm 10.2 V Circular 10% at 750 nm 35 V 5 lm diameter 5% at 840 nm Circular For 400 lm SiAPD 25 to 800 lm diameter 30% at 650 nm Circular 0.5% at 950 nm 10 lm diameter 0.35 lm high voltage (HV) [36] 0.7 lm HV [28] Geiger Linear 12 V Gain: 10–40 20% at 800 nm 0.8 lm HV [37] Geiger 55 V 12 lm diameter Table 2 Design specification of the proposed SiAPD Parameter Value Fabrication technology CMOS 35 lm Supply voltage 3–5 V Gain in linear mode [100 Wavelength 700–950 nm Photon absorption efficiency 85% at 700 nm 30% at 900 nm Area *200 9 200 lm2 Depletion layer thickness [10 lm fixed for 0.35 lm CMOS technology. We created the masks for the SiAPD structure using Ligament Layout Editor and created an input command file for Ligament Flow Editor. The input command file emulates the fabrication process and creates the structure and its doping data. The output from Ligament Flow Editor serves as an input for Sentaurus Process, which produces doping profile and electric field distribution of the diode. Figure 13 shows the doping profile of the device. Figure 14 shows the electric field distribution of the device under reverse bias. It also shows that the maximum electric field (*105 V/cm) appeared in the active p?-deep n-well junction and the device is able to withstand the electric field without breakdown. 5.1.2 Layout of the SiAPD device We have submitted the layout of p?/n-well avalanche photodiode with guard ring for fabrication in a 0.35 lm CMOS technology. We have designed SiAPDs with two different shapes: square and octagonal. For the octagonal SiAPD, guard ring is realized by low-doped p-well around p? active area. For the square shape SiAPD, guard ring is realized by low-doped n-ring due to n-wells’ lateral diffusion. Fig. 13 Doping profile of the SiAPD using the Sentaurus TCAD software 5.1.3 Proposed transimpedance amplifier For linear mode operation, the first circuit element is a TIA. We have proposed a new TIA with combination of common-gate and resistive feedback techniques, depicted in Fig. 15 to achieve a low-noise, low-power and high-gain front-end. This circuit is a modified version of the previously developed TIA in our lab [3]. The design consist of a current amplifier implemented in a transimpedance configuration [43]. In this circuit, we have simply used the combination of two transistors (M6 and M7) to work as a feedback resistor to minimize the output ripple and omit the extra drawn current. The dc transimpedance gain is given by: Vout gM5 Rf  1 ¼ ; gM4 þ gM5 Iin ð6Þ where gM4 and gM5 are the transconductance of transistors M4, M5, and Rf is the feedback resistance implemented by M6 and M7 biased in the linear region. The frequency where the loop gain of the TIA is unity: 123 10 Analog Integr Circ Sig Process (2012) 70:1–13 Fig. 14 Device simulation of the SiAPD under reverse bias using the Sentaurus TCAD software Input Noise (A/ Hz) 7E-10 6E-10 5E-10 4E-10 3E-10 2E-10 1E-10 0E+00 1E+00 1E+02 1E+04 1E+06 1E+08 Frequency (Hz) Fig. 16 Input noise of the proposed TIA as a function of frequency Fig. 15 Schematic of the proposed TIA xt  A gm1 R1 ;  Rf C D Rf CD ð7Þ where CD is the photodiode capacitance. The closed-loop bandwidth of the TIA is approximately equal to the unitygain frequency: BW  1 þ A gm1 R1  xt :  Rf C D Rf CD 6 Conclusion ð8Þ Bandwidth of TIA increases by decreasing the CD. We have used the CD = 1 pF in our simulations as this is the commonly reported value [13, 18, 24]. 5.1.4 Simulation results of the proposed TIA We have used CADENCE schematic editor and Virtuoso layout editor to design and simulate our proposed TIA in 0.35 lm CMOS technology. In order to optimize the performance of the amplifier, we have analyzed the sensitivity 123 of each circuit component on the output. Figure 16 shows the input noise of the proposed circuit as a function of input current’s (Iin) frequency. Figure 17 shows the transimpedance gain of the proposed circuit as a function of its frequency. The input noise is very low (&0.13 nA/HHz at 100 Hz and 1 fA/HHz at 1 kHz) compared to existing TIAs. Measured output noise is also about 15 lv/HHz at 1 kHz. The power consumption of the front-end circuit is also very low (0.8 mW). The TIA has high transimpedance gain (*250 MV/A) and high bandwidth (*10 kHz). Silicon avalanche photodiodes are promising photodetector for low-level light detection. In this paper, we present a review of constraints and challenges to implement CMOS SiAPDs for an integrated NIRS system. We proposed a novel SiAPD along with its peripheral circuitry for NIRS application. Our designed SiAPD has large area (*200 9 200 lm2) to capture enough NIR signal and wide depletion region ([10 lm) to increase the absorption of NIR light. Device level simulations result shows that the device can withstand the high electric field (*105 V/cm) for avalanche without device breakdown. We also designed Analog Integr Circ Sig Process (2012) 70:1–13 11 3.0E+08 Gain (V/A) 2.5E+08 2.0E+08 1.5E+08 1.0E+08 5.0E+07 0.0E+00 1E+00 1E+02 1E+04 1E+06 1E+08 Frequency (Hz) Fig. 17 Transimpedance gain of the proposed TIA as a function of frequency a novel TIA appropriate for amplifying and filtering the signal coming from the SiAPD. The main advantages of the TIA are low-power consumption (0.8 mW), high transimpedance gain (up to 250 MV/A), and very low input noise (1 fA/HHz). We confirmed the efficacy of our proposed TIA by circuit level post layout simulation. We implemented SiAPD and its associated circuits using 0.35 lm standard CMOS technology. Such excellent performance of SiAPD substantiates its feasibility to serve as a highly sensitive photodetector of an integrated NIRS device for portable biomedical applications. Acknowledgments We gratefully acknowledge financial support from the Heart and Stroke Foundation of Canada (HSFC) and the Canadian Institutes of Health Research (CIHR). References 1. Refaat, T. F., Elsayed-Ali, H. E., & DeYoung, R. J. (2001). Driftdiffusion model for reach-through avalanche photodiodes. Optical Engineering, 40(9), 1928–1935. 2. Faramarzpour, N., Deen, M. J., Shirani, S., & Fang, Q. (2008). Fully integrated single photon avalanche diode detector in standard CMOS 0.18-lm technology. IEEE Transactions on Electron Devices, 55(3), 760–768. 3. Achigui, H. F., Sawan, M., & Fayomi, C. J. B. (2008). A monolithic based NIRS front-end wireless sensor. Microelectronics Journal, 39(10), 1209–1217. 4. Bozkurt, A., Rosen, A., Rosen, H., & Onaral, B. (2005). A portable near infrared spectroscopy system for bedside monitoring of newborn brain. Biomedical Engineering Online, 4, 29. 5. Franceschini, M. A., & Boas, D. A. (2004). Noninvasive measurement of neuronal activity with near-infrared optical imaging. Neuroimage, 21(1), 372–386. 6. Soraghan, C., Matthews, F., Markham, C., Pearlmutter, B. A., O’Neill, R., & Ward, T. E. (2008). A 12-channel, real-time nearinfrared spectroscopy instrument for brain-computer interface applications. In Proceedings of 30th Annual International IEEE EMBS Conference (pp. 5648–5651). 7. Haensse, D., Szabo, P., Brown, D., Wolf, M., et al. (2005). New multichannel near infrared spectrophotometry system for functional studies of the brain in adults and neonates. Optics Express, 13(12), 4525–4538. 8. Wolf, M., Wolf, U., Gratton, E., et al. (2003). Fast cerebral functional signal in the 100-ms range detected in the visual cortex by frequency-domain near-infrared spectrophotometry. Psychophysiology, 40(4), 521–528. 9. Wolf, M., Wolf, W., et al. (2002). Different time evolution of oxyhemoglobin and deoxyhemoglobin concentration changes in the visual and motor cortices during functional stimulation: A near-infrared spectroscopy study. Neuroimage, 16, 704–712. 10. Ruben, J., Wenzel, R., & Villringer, A. (1997). Haemoglobin oxygenation changes during visual stimulation in the occipital cortex. Advances in Experimental Medicine and Biology, 428, 181–187. 11. Wolf, M., Morren, G. D., & Bucher, H. U. (2008). Near infrared spectroscopy to study the brain: An overview. Opto-electronics Review, 16(4), 413–419. 12. Yang, Q. (2005). Design of front-end amplifier for optical receiver in 0.5 micrometer CMOS technology. MSc Thesis, University of Hawai. 13. Phang, K. (2001). CMOS optical preamplifier design using graphical circuit analysis. PhD thesis, University of Toronto. 14. Peluso, V., Vancorenland, P., Steyaert, M., & Sansen, W. (1997). 900 mV differential class AB OTA for switched opamp applications. Electronics Letters, 33(17), 1455–1456. 15. Zappa, F., Tisa, S., Gulinatti, A., Gallivanoni, A., & Cova, S. (2005). Complete single-photon counting and timing module in a microchip. Optics Letters, 30(11), 1327–1329. 16. Lawrence, W. G., Christian, J. F., Augustine, F. L., Squillante, M. R., & Entine, G. (2005). Development and characterization of CMOS avalanche photodiode arrays. In Proceedings of SPIE, 5726 (pp. 122–132). 17. Dalla Mora, A., Tosi, A., Tisa, S., & Zappa, F. (2007). Singlephoton avalanche diode model for circuit simulations. IEEE Photonics Technology Letters, 19, 1922–1924. 18. Tisa, S., Zappa, F., Tosi, A., & Cova, S. (2007). Electronics for single photon avalanche diode arrays. Sensors and Actuators A, 140, 113–122. 19. American National Standards Institute (2010). http://www. ansi.org/. Accessed 12 Oct 2010. 20. Normandin, F., Sawan, M., & Faubert, J. (2005). New integrated front-end for a noninvasive brain imaging system based on nearinfrared spectroreflectometry. IEEE Transactions on Circuits and Systems I, 52(12), 2663–2671. 21. Gulinatti, A., Rech, I., Maccagnani, P., Ghioni, M., & Cova, S. (2005). Large-area avalanche diodes for picosecond time-correlated photon counting. In Proceedings of ESSDERC (pp. 355–358). 22. Ghioni, M., Gulinatti, A., Rech, I., Zappa, F., & Cova, S. (2007). Progress in silicon single-photon avalanche diodes. IEEE Journal of Selected Topics in Quantum Electronics, 13(4), 852–862. 23. Sze, S. M. (1981). Physics of semiconductor devices. New York: Wiley. 24. Zappa, F., Tisa, S., Tosi, A., & Cova, S. (2007). Principles and features of single-photon avalanche diode arrays. Sensors and Actuators A, 140, 103–112. 25. Optical properties of silicon. (2010) http://pvcdrom.pveducation. org/APPEND/OPTICAL.HTM. Accessed 12 Oct 2010. 26. McIntyre, R. J. (1966). Multiplication noise in uniform avalanche diodes. IEEE Transactions on Electron Devices, 13(1), 164–168. 27. Moutaye, E. R., & Tap-Béteille, H. (2008). CMOS avalanche photodiode embedded in a phase-shift laser rangefinder. IEEE Transactions on Electron Devices, 55(12), 3396–3401. 28. Charbon, E. (2008). Towards large scale CMOS single-photon detector arrays for lab-on-chip applications. Journal of Physics D. Applied Physics, 41, 094010. 29. Kim, Y., Jun, I., & Kim, K. H. (2008). Design and characterization of CMOS avalanche photodiode with charge sensitive 123 12 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. Analog Integr Circ Sig Process (2012) 70:1–13 preamplifier. IEEE Transactions on Nuclear Science, 55(3), 1376–1380. Rochas, A., Pauchard, A. R., Popovic, R. S., et al. (2002). Lownoise silicon avalanche photodiodes fabricated in conventional CMOS technologies. IEEE Transactions on Electron Devices, 49(3), 387–393. Rochas, A., Gisin, N., et al. (2003). Single photon detector fabricated in a complementary metal–oxide–semiconductor high-voltage technology. Review of Scientific Instruments, 74(7), 3264–3270. Rochas, A., Rigler, R., et al. (2003). First fully integrated 2-D array of single-photon detectors in standard CMOS technology. IEEE Photonics Technology Letters, 15(7), 963–965. Finkelstein, H., Hsu, M. J., & Esener, S. C. (2006). STI-bounded single-photon avalanche diode in a deep-submicrometer CMOS technology. IEEE Electron Device Letters, 27(11), 887–889. Gersbach, M., Niclass, C., & Grant, L. (2008). A single photon detector implemented in a 130 nm CMOS imaging process. In Proceedings of the 38th Eur. Solid-State Device Research Conference (pp. 270–273). Niclass, C., Gersbach, M., Henderson, R., Grant, L., & Charbon, E. (2007). A single photon avalanche diode implemented in 130-nm CMOS technology. IEEE Journal of Selected Topics in Quantum Electronics, 13(4), 863–869. Xiao, Z., Pantic, D., & Popovic, R. S. (2007). A new single photon avalanche diode in CMOS high-voltage technology. In Proceedings of Transducers and Eurosensors (pp. 1365–1368). Tisa, S., Tosi, A., & Zappa, F. (2007). Fully-integrated CMOS single photon counter. Optics Express, 15(6), 2873–2887. Razeghi, M. (2010). Technology of quantum devices (p. 2010). London: Springer. Cova, S., Ghioni, M., Lacaita, A., Samori, C., & Zappa, F. (1996). Avalanche photodiodes and quenching circuits for singlephoton detection. Applied Optics, 35(12), 1956–1976. Zappa, F., Giudice, A., Ghioni, M., & Cova, S. (2002). Fullyintegrated active-quenching circuit for single-photon detection. In Proceedings of the 28th European solid-state circuits conference, 2002, pp. 355–358. Zappa, F., Lotito, A., & Tisa, S. (2005). Photon-counting chip for avalanche detectors. IEEE Photonics Technology Letters, 17(1), 184–186. Cronin, D., Moloney, A. M., & Morrison, A. P. (2004). Simulated monolithically integrated single photon counter. In Proc. IEEE high frequency postgraduate student colloquium 2004, pp. 9–14. Phang, K., & Johns, D. A. (2001). A 1 V 1 mW CMOS front-end with on-chip dynamic gate biasing for a 75 Mb/s optical receiver. In IEEE international solid-state circuits conference, San Francisco, CA, USA, pp. 218–220. Afrin Sultana received her B.Sc. degree (with honors) in Electrical and Electronic Engineering from Bangladesh University of Engineering & Technology, Dhaka, Bangladesh, and M.A.Sc. and PhD degrees in Electrical and Computer Engineering from the University of Waterloo, Waterloo, Canada in 2002, 2004, and 2009, respectively. Her PhD dissertation focused on amorphous silicon based large area detector for protein crystallography. She has been working as a postdoctoral fellow with Professor Mohamad Sawan in Polystim Lab at Ecole Polytechnique since 123 November 2009. She is currently working to develop highly sensitive photodetector for near-infrared spectroscopy (NIRS) system for monitoring brain function. She has received a number of government and university scholarships including Ontario Graduate Scholarship, University of Waterloo President’s Scholarship, and the Best Student Paper Award in the Symposium on Circuits, Devices and Systems of IEEE Canadian Conference on Electrical and Computer Engineering, 2008. Ehsan Kamrani received his B.Sc. degree in Biomedical engineering from Shahid Beheshti Medical Science University, Iran, and Masters degree in Electrical and Control Engineering from Tarbiat Modares University, Iran, in 2002 and 2005 respectively. From 2005 to 2009 he has been an Academic Member-Instructor in the Department of Electrical and Electronics Engineering, University of Lorestan, Iran. His expertises are on wireless networked sensors, web-based control systems and biomedical signal/image processing. Since 2009 he has been doing his PhD on Biomedical Engineering under supervision of Prof. Sawan at Polystim neurotechnologies Laboratory, Ecole Polytechnique, Montreal, Canada. He is working on design and implementation of an fNIRS photo receiver for real-time brain monitoring. Mohamad Sawan received the B.Sc. degree in 1984 from Laval University and the Ph.D. degree in 1990 in electrical engineering, from Sherbrooke University, Canada. He joined Polytechnique Montréal in 1991, where he is currently a Professor of Microelectronics and Biomedical Engineering. His scientific interests are the design and test of mixed-signal (analog, digital, RF, MEMS and optic) circuits and Microsystems: integration, assembly and validations. These topics are oriented toward the biomedical and telecommunications applications. Dr. Sawan is a holder of a Canada Research Chair in Smart Medical Devices. He is leading the Microsystems Strategic Alliance of Quebec (ReSMiQ) receiving membership support from 11 Universities. Dr. Sawan is founder of the Eastern Canadian IEEE-Solid State Circuits Society Chapter, the International IEEE NEWCAS Conference, and the Polystim Neurotechnologies Laboratory at Polytechnique Montréal. He is cofounder of the International Functional Electrical Stimulation Society, the International IEEE conference on Biomedical Circuits and Systems (BioCAS), and the BioCAS committee in the IEEE circuits and systems society. He is also cofounder, Associate Editor and member of the steering committee of the IEEE Transactions on BioCAS (TBCAS). He is Deputy Editor-in Chief of the IEEE Trans. on circuits and systems II (TCAS-II), Editor and Associate Editor of several other International Journals. Dr. Sawan is member of the board, editor, guest editor and associate editor of several other prestigious scientific Journals. Dr. Sawan published more than 500 papers in peer reviewed journals and conference proceedings, two books and more than ten book chapters, and he offered more than 100 invited talks. He was awarded nine patents pertaining to the field of biomedical sensors and actuators. Dr. Sawan received several Analog Integr Circ Sig Process (2012) 70:1–13 prestigious awards; the most important of them are the Medal of Honor from the President of Lebanon, the Bombardier Award for technology transfer, the Barbara Turnbull Award for medical research in Canada, and the achievement Award from the American University of Science 13 and Technology. Dr. Sawan is Fellow of the IEEE, Fellow of the Canadian Academy of Engineering, Fellow of the Engineering Institute of Canada, and Officer of the Quebec’s National Order. 123