PDE/SiPM Interactive Calculator

The Photon Detection Efficiency (PDE) of Silicon Photomultipliers (SiPMs) quantifies the probability that an incident photon will produce a detectable avalanche pulse in the device. PDE is the product of three independent factors: quantum efficiency (QE), geometric fill factor, and avalanche triggering probability. This calculator enables precise analysis of SiPM performance across multiple wavelengths and operating conditions, essential for applications in medical imaging (PET scanners), LiDAR systems, quantum key distribution, and high-energy physics detectors where single-photon sensitivity determines system performance.

📐 Browse all free engineering calculators

Visual Diagram

PDE/SiPM Interactive Calculator

Governing Equations

Theory & Practical Applications

Physical Principles of SiPM Operation

Silicon photomultipliers operate as arrays of single-photon avalanche diodes (SPADs) connected in parallel through integrated quenching resistors. Each microcell functions as an independent Geiger-mode photodiode biased above its breakdown voltage. When a photon generates an electron-hole pair in the depletion region, the intense electric field (typically 3×10⁵ V/cm) accelerates the carrier to energies sufficient for impact ionization. This initiates a self-sustaining avalanche that rapidly spreads across the entire microcell junction within nanoseconds, producing a macroscopic current pulse containing approximately 10⁵-10⁶ carriers regardless of whether one or multiple photons triggered the avalanche.

The photon detection efficiency represents a cascade of three independent quantum processes. First, quantum efficiency quantifies the wavelength-dependent probability that an incident photon generates an electron-hole pair before being reflected, transmitted, or absorbed in non-sensitive regions. Silicon's bandgap of 1.12 eV establishes a long-wavelength cutoff near 1100 nm, while surface recombination and entrance window absorption reduce QE at short wavelengths. Peak quantum efficiency occurs in the blue-green region (400-500 nm) where silicon's absorption depth matches the depletion region thickness, reaching 85-95% for devices with anti-reflection coatings.

Fill factor arises from the geometric necessity of isolating adjacent microcells with optical and electrical barriers. Trenches etched between cells prevent optical crosstalk but create dead regions insensitive to photons. Additionally, quenching resistors, metal interconnects, and guard rings consume active area. Modern high-density SiPMs with 15-25 μm cell pitch achieve fill factors of 60-75%, while larger 50-100 μm cells may reach 80-85%. Trenched isolation architectures improve fill factor by confining optical barriers to narrow regions while maintaining electrical separation.

Avalanche triggering probability depends exponentially on overvoltage according to P_avalanche = 1 − exp(−V_ov/V_char), where V_char represents a device-specific characteristic voltage typically 1.5-2.5 V. This relationship reflects the stochastic nature of impact ionization: carriers must gain sufficient energy between collisions to generate secondary electron-hole pairs. At low overvoltage, many carriers scatter before reaching the ionization threshold energy, resulting in self-terminating sub-Geiger discharges. Increasing overvoltage enhances the electric field, shortening the acceleration distance required to reach ionization energy and increasing avalanche probability. Beyond 5-6 V overvoltage, P_avalanche saturates near unity but at the cost of dramatically increased dark count rate and afterpulsing.

Spectral Response and Wavelength Optimization

The wavelength dependence of PDE exhibits a characteristic peak structure determined by silicon's optical and electronic properties. Short-wavelength photons (300-400 nm) are absorbed within nanometers of the surface where high surface recombination velocity annihilates carriers before they reach the depletion region. UV-enhanced devices employ special doping profiles or surface treatments to extend sensitivity below 400 nm, critical for Cherenkov light detection in particle physics experiments where the emission spectrum peaks near 350 nm.

Infrared sensitivity degrades beyond 800 nm as silicon's absorption coefficient decreases exponentially with wavelength. Photons penetrate deeply into the substrate, generating carriers far from the high-field region where avalanche multiplication occurs. These carriers diffuse slowly and most recombine before reaching the junction. Near-infrared optimized SiPMs use thick epitaxial layers (50-100 μm) to maintain reasonable quantum efficiency at 850-950 nm for LiDAR and optical communication applications, though PDE rarely exceeds 20% in this regime compared to 40-50% at peak wavelength.

Dark Count Rate and Noise Mechanisms

Dark counts arise from thermally generated carriers and band-to-band tunneling in the high-field depletion region. Thermal generation follows Shockley-Hall-Read statistics with an exponential temperature dependence: DCR doubles approximately every 8°C due to the E_g/2kT term in the generation rate equation where E_g = 1.12 eV for silicon. A device exhibiting 100 kHz/mm² dark count rate at 20°C will produce 1.6 MHz/mm² at 40°C, overwhelming weak signals unless cooled. Medical PET scanners often operate SiPMs at −20°C to achieve sub-10 kHz/mm² dark count rates, though this adds system complexity and power consumption.

Afterpulsing represents a correlated noise mechanism where carriers become trapped in defect states during an avalanche and subsequently release after the dead time, triggering a secondary false count. Afterpulsing probability increases with overvoltage because larger avalanches generate more carriers available for trapping. Typical afterpulsing probabilities range from 0.5% at 2 V overvoltage to 5-10% at 8 V overvoltage with time constants of 10-500 ns depending on trap energy levels. Applications requiring coincidence detection over nanosecond windows must carefully model afterpulsing correlations to avoid systematic biases.

Multi-Industry Applications

Medical Imaging (PET Scanners): Time-of-flight positron emission tomography achieves 200-300 ps coincidence time resolution by coupling fast scintillators like LSO or LYSO to SiPM arrays. PDE directly determines the number of detected scintillation photons per 511 keV gamma interaction, with higher PDE improving both energy resolution and timing precision. Modern whole-body PET scanners employ 24,000-32,000 individual SiPM channels operating at PDE values of 35-45% at the 420 nm LSO emission peak. The dark count rate must remain below 100 kHz per channel to maintain acceptable random coincidence rates, necessitating temperature stabilization within ±1°C and careful selection of low-noise devices.

Automotive LiDAR: Direct time-of-flight ranging at 905 nm requires SiPM arrays capable of detecting return pulses containing as few as 50-100 photons after atmospheric scattering and optical losses. Fill factor becomes critical in this application because imaging LiDAR systems employ microlens arrays to focus light onto active regions, and geometric efficiency directly translates to increased range capability. A SiPM with 65% fill factor compared to 50% provides 30% greater detection probability, extending maximum range by 13% for weak returns. Operating at 3-4 V overvoltage balances PDE (typically 20-25% at 905 nm) against sunlight-induced background counts which can reach MHz rates in bright conditions.

Quantum Key Distribution: Single-photon quantum communication systems at 1550 nm telecom wavelength require InGaAs/InP SPADs rather than silicon SiPMs due to wavelength limitations, but silicon-based systems at 780-850 nm quantum memory wavelengths employ SiPMs as single-photon detectors. Afterpulsing becomes the dominant noise source in this application because quantum states must be distinguished with near-unit fidelity. Gating the detector only during expected photon arrival times with 1-10 ns windows reduces dark count probability to 10^-4-10^-3 per gate while afterpulsing from previous detections corrupts measurements. Manufacturers specify afterpulsing characteristics at multiple delay times, and system designers model the time-dependent afterpulsing probability to optimize gate timing.

High-Energy Physics Calorimetry: Particle detectors in collider experiments employ SiPM readout of plastic scintillator tiles to measure electromagnetic shower energies. Radiation damage progressively increases dark count rate and degrades PDE over years of operation at integrated neutron fluences exceeding 10¹² n/cm². Careful device selection prioritizes radiation-hard epitaxial layer thicknesses and doping profiles that minimize defect introduction. Initial PDE values of 40-45% may degrade to 25-30% after 10 years at the Large Hadron Collider's high-luminosity interaction points, requiring periodic recalibration and eventual replacement of front-end modules.

Comprehensive Worked Example: PET Detector Module Design

Scenario: Design the SiPM photodetector array for a clinical PET scanner detector module. Each module consists of a 4×4 array of LYSO scintillator crystals (4 mm × 4 mm × 20 mm) coupled to SiPMs. Specifications require 12% energy resolution at 511 keV and 250 ps coincidence time resolution at 20°C ambient temperature.

Given Parameters:

• LYSO light yield: 32,000 photons per 511 keV gamma interaction
• LYSO emission peak: 420 nm
• Optical coupling efficiency (crystal to SiPM): 75%
• SiPM active area per crystal: 4 mm × 4 mm = 16 mm²
• Available SiPM specifications: QE = 78% at 420 nm, fill factor = 68%, dark count rate = 85 kHz/mm² at 20°C
• Scintillation decay time: 40 ns (primary component)

Step 1: Determine required PDE for energy resolution target.

Energy resolution in scintillation detectors is limited by Poisson statistics of photoelectron generation. The relative energy resolution scales as 1/√N_pe where N_pe is the average number of detected photoelectrons. For 12% resolution (FWHM), we require approximately N_pe ≈ (2.35 / 0.12)² ≈ 384 photoelectrons per 511 keV event.

Photons reaching SiPM surface: 32,000 × 0.75 = 24,000 photons

Required PDE: N_pe / photons = 384 / 24,000 = 0.016 = 1.6%

This seems exceptionally low, but actual PET detectors achieve far better statistics. Recalculating for professional-grade specifications: N_pe = 24,000 × PDE. For 12% energy resolution, N_pe should be approximately 1500-2000 photoelectrons, requiring PDE = 1500/24,000 = 6.25% minimum, though practical systems target 8-10% to account for variations.

Corrected approach: Standard PET energy resolution formula: ΔE/E = k/√N_pe where k ≈ 2.35 for statistical limit. For ΔE/E = 0.12, we need N_pe = (2.35/0.12)² = 384 detected photoelectrons minimum. Given 24,000 incident photons, PDE = 384/24,000 = 1.6%. However, real systems target 3000-5000 photoelectrons for robust 10-12% resolution accounting for non-statistical broadening. Using N_pe = 4000: PDE_required = 4000/24,000 = 16.7%.

Step 2: Calculate achievable PDE with available device.

We need to determine the avalanche probability P_avalanche required to achieve PDE = 16.7% given QE = 78% and FF = 68%.

PDE = QE × FF × P_avalanche

0.167 = 0.78 × 0.68 × P_avalanche

P_avalanche = 0.167 / (0.78 × 0.68) = 0.167 / 0.5304 = 0.315 = 31.5%

Step 3: Determine required overvoltage.

Using avalanche probability model with characteristic voltage V_char = 2.2 V (typical for this SiPM type):

P_avalanche = 1 − exp(−V_ov / V_char)

0.315 = 1 − exp(−V_ov / 2.2)

exp(−V_ov / 2.2) = 0.685

−V_ov / 2.2 = ln(0.685) = −0.378

V_ov = 0.378 × 2.2 = 0.83 V

Step 4: Evaluate dark count impact on timing resolution.

Total dark count rate per SiPM: 85 kHz/mm² × 16 mm² = 1.36 MHz per channel

Scintillation pulse integral (photon arrival time window): Primary 40 ns component contains ~63% of light (1 − e^−40/40). For timing, we integrate over first 10 ns where approximately 22% of photons arrive.

Signal photons in timing window: 24,000 × 0.167 × 0.22 = 882 detected photoelectrons in first 10 ns

Dark counts in 10 ns window: 1.36 MHz × 10 ns = 0.0136 counts

Dark count contribution is negligible (1.5% of signal), confirming acceptable noise performance for timing resolution.

Step 5: Verify timing resolution capability.

Coincidence timing resolution (CTR) scales approximately as CTR ≈ τ_decay / √N_pe,fast where τ_decay is the scintillation rise time (~0.1 ns for LYSO) and N_pe,fast represents photoelectrons in the rising edge.

Using more precise formula: CTR = (τ_rise / √N_pe,10ns) × FWHM_factor where FWHM_factor ≈ 2.35

CTR = (0.1 ns / √882) × 2.35 = (0.1 / 29.7) × 2.35 = 0.0079 ns = 7.9 ps (single photon limit)

This represents the statistical limit. Real CTR including jitter, transit time spread, and electronics timing: CTR_total = √(CTR_stat² + CTR_SiPM² + CTR_electronics²)

With SiPM jitter ~100 ps SPTR (single photon time resolution) and 100 ps electronics contribution:

CTR_total = √(7.9² + 100² + 100²) = √(62 + 10,000 + 10,000) = √20,062 = 142 ps

This significantly exceeds the 250 ps specification, confirming the design is viable. In practice, 200-220 ps is achievable with optimized readout electronics and timestamp extraction algorithms.

Conclusion: The proposed SiPM configuration requires only 0.83 V overvoltage, resulting in low noise operation with dark count contribution under 2%. The PDE of 16.7% provides 882 detected photoelectrons in the timing window, enabling 142 ps statistical timing resolution. The design meets both energy resolution and timing specifications with margin for manufacturing variations and long-term performance degradation.

Frequently Asked Questions