Photodiode: Photovoltaic and Photoconductive Modes

A photodiode is a p-n (or p-i-n) junction optimized to convert incident photons into electrical current, and it can be biased in two fundamentally different ways that trade off speed, noise, and linearity. Choosing photovoltaic (unbiased) or photoconductive (reverse-biased) operation is one of the first decisions in any optical receiver, power meter, or safety interlock design — and getting it wrong shows up as excess dark current, bandwidth that's an order of magnitude too low, or noise that swamps a weak signal.

Junction Physics Recap

When light with photon energy hν greater than the semiconductor bandgap Eg strikes the depletion region (and the diffusion-length-limited regions near it), it generates electron-hole pairs. The junction's built-in field sweeps these carriers apart before they recombine, producing a photocurrent Iph that is, to first order, proportional to incident optical power:

Iph = R · Popt

where R (A/W) is the responsivity, related to quantum efficiency η by R = η·q/(hν) = η·λ/1.24 µm (λ in µm, R in A/W). This photocurrent is a current source in parallel with the normal diode junction — that single fact is what makes both operating modes describable with one equation:

I = Isat·(exp(V/Vt) − 1) − Iph

Isat is the reverse saturation (dark) current, Vt = kT/q ≈ 25.85 mV at 300 K, and Iph is defined positive flowing from cathode to anode inside the device (i.e., it subtracts from the diode's forward current in the sign convention above). Everything about photovoltaic vs. photoconductive behavior comes from where you sit on this I-V curve.

Photovoltaic Mode (V = 0, Unbiased)

With no external bias, set V = 0:

I = Isat·(e^0 − 1) − Iph = −Iph

The diode delivers current Iph at essentially zero applied voltage — this is exactly how a solar cell operates, just at lower light levels. If the diode is open-circuited instead of short-circuited, it develops an open-circuit voltage:

Voc = Vt · ln(1 + Iph/Isat)

Key characteristics:

• No dark current (ideally) — since V = 0, the diode's forward current term is zero, so the only current is the photocurrent itself. This makes photovoltaic mode the low-noise choice for very low light levels.
• Junction capacitance is high — with no reverse bias, the depletion region is at its narrowest, so Cj = Cj0/√(1 + Vr/Vbi) (where Vr = −V ≥ 0 is the reverse-bias magnitude, Vr = 0 in photovoltaic mode) is maximized. This capacitance forms an RC low-pass with the load/transimpedance input, limiting bandwidth.
• Noise is dominated by shot noise of the photocurrent plus the amplifier's input noise — there's no bias-dependent leakage to add.
• Response is highly linear only over a limited range near the origin; large Iph starts to forward-bias the junction, causing the I-V relationship to bend logarithmically rather than staying linear (this is what makes solar cells nonlinear power sources, why they need MPPT).

Photoconductive Mode (V < 0, Reverse Biased)

Applying a reverse bias V = −Vr (Vr > 0):

I = Isat·(e^(−Vr/Vt) − 1) − Iph ≈ −Isat − Iph  (for Vr >> Vt)

The exponential term collapses toward −1, leaving a nearly constant dark current −Isat added to the photocurrent. This is the regime used in high-speed photodiodes, avalanche photodiodes (in their linear region), and most fiber-optic receivers.

Key characteristics:

• Wider depletion region → lower junction capacitance → higher bandwidth. For a typical PIN photodiode, reverse bias can drop Cj from tens of pF (photovoltaic) to a few pF, pushing the RC-limited bandwidth from hundreds of kHz into the hundreds of MHz.
• Faster carrier transit time — the stronger internal field accelerates carriers to (or near) their saturation velocity, reducing transit-time-limited rise time, which often matters more than RC time constant at GHz-class bandwidths.
• Dark current is present and bias-dependent, contributing shot noise in = √(2·q·Idark·B) that did not exist in photovoltaic mode. Dark current also roughly doubles every 8–10°C for silicon, making thermal stability a real design concern.
• Better linearity over a wide dynamic range — because the operating point stays clamped near V = −Vr regardless of Iph (as long as the load doesn't pull the junction toward forward bias), the current-to-power relationship stays linear over several decades of optical power.

Worked Example: Bandwidth Comparison

Consider a silicon PIN photodiode with Cj = 20 pF at V = 0 (photovoltaic) and Cj = 2 pF at V = −10 V (photoconductive), feeding a transimpedance amplifier with feedback resistor Rf = 50 kΩ and negligible amplifier bandwidth limitation (ideal op-amp assumption for this first pass).

Photovoltaic mode:

f_-3dB = 1/(2π·Rf·Cj) = 1/(2π · 50k · 20p) = 1/(6.283 × 10^-6) ≈ 159 kHz

Photoconductive mode:

f_-3dB = 1/(2π · 50k · 2p) = 1/(6.283 × 10^-7) ≈ 1.59 MHz

Check: capacitance dropped by 10×, and since f_-3dB ∝ 1/C, bandwidth should rise by 10× — 159 kHz → 1.59 MHz confirms this. In practice the transimpedance amplifier's own gain-bandwidth product and the feedback capacitance Cf needed for stability (Cf ≈ √(Cj/(2π·Rf·GBW))) will cap the achievable bandwidth well below this ideal number, but the 10:1 relative improvement from reverse bias is the real, dominant effect.

Choosing a Mode: Practical Guidance

Design implications:

• Transimpedance amplifier (TIA) front ends almost always reverse-bias the photodiode when bandwidth matters, biasing the cathode positive relative to the TIA's inverting input held near ground/virtual ground.
• Temperature compensation becomes necessary in photoconductive designs if dark current is a meaningful fraction of signal current — either by measuring/subtracting a dark reference or by choosing a photodiode with low Isat (larger bandgap material, smaller active area).
• Reverse breakdown voltage (VBR) sets the upper limit on bias; exceeding it (deliberately, in APDs) trades noise for gain via avalanche multiplication — a different regime from ordinary photoconductive operation.
• Area vs. speed tradeoff: larger photodiode area collects more light (higher Iph for a given irradiance) but also raises Cj, which is why high-speed reverse-biased devices are often physically small, pushing designers toward tighter optical coupling (lenses, fiber alignment).

Key Takeaways

• Photovoltaic mode (V = 0) gives zero dark current and the lowest noise floor but suffers from high junction capacitance and nonlinearity at high photocurrent — the natural choice for solar cells and precision low-light sensing.
• Photoconductive mode (reverse bias) narrows the depletion region, cutting Cj by 5–10× or more and speeding up carrier transit, at the cost of a bias-dependent dark current that adds shot noise.
• Bandwidth in a transimpedance front end scales as 1/(Rf·Cj), so the capacitance reduction from reverse bias directly and predictably multiplies achievable bandwidth.
• Dark current in photoconductive mode roughly doubles every 8–10°C in silicon, making thermal management and possibly dark-current compensation necessary in precision designs.
• Mode choice is a system-level tradeoff: noise and simplicity favor photovoltaic; speed, dynamic range, and linearity favor photoconductive.