Photodiode array

A photodiode array (PDA) is a set of photodiodes fabricated side by side on a single chip, either in a line (a linear array) or in a two-dimensional grid, with each element producing its own photocurrent. The array thus records a spatial profile in one measurement: a spectrum dispersed by a grating, a spot position, or signals from parallel fibers. Silicon linear arrays, the most common kind, typically have 128 to 2048 or more pixels at pitches from about 7 to 50 µm and respond from roughly 200 to 1100 nm. InGaAs linear arrays, built on the same material as a single InGaAs photodetector, typically have 256 to 1024 pixels and cover about 900 to 1700 nm.

Readout architectures

Arrays with few elements are usually individually wired: each diode has its own bond pad and its own amplifier, often a transimpedance amplifier, so all channels are read continuously and at full bandwidth. Parallel optical receivers work this way; 4- and 12-channel PIN photodiode arrays at 250 µm pitch match the fiber spacing of MT-ferrule ribbon connectors, and a 4 × 25 Gb/s receiver of this kind carries 100 Gb/s.

Arrays with hundreds or thousands of pixels are self-scanned. Each photodiode charges or discharges its own junction or an integrating capacitor during an integration time, and an on-chip shift register then connects the pixels one after another to a common video line, where a charge amplifier and an analog-to-digital converter read them. The pixel therefore reports accumulated charge, and the integration time sets the sensitivity in the same way an exposure time does for a camera.

Signal, well capacity and dark current

The photocurrent per pixel follows from the responsivity, $R = \eta\lambda/1.23984$ A/W with $\lambda$ in micrometres. For a silicon pixel with a quantum efficiency of 80% at 550 nm:

$$R = \frac{0.80 \times 0.55}{1.23984} = 0.355\ \text{A/W}.$$

A pixel receiving 1 nW then carries 0.355 nA. Over a 10 ms integration it collects

$$Q = 0.355\ \text{nA} \times 10\ \text{ms} = 3.55\ \text{pC},$$

or about 2.2 × 10⁷ electrons. If the saturation charge of the pixel is 10 pC (6.2 × 10⁷ electrons, a representative value for large-pixel self-scanned arrays), the same illumination saturates it in about 28 ms. Near full well the shot-noise-limited signal-to-noise ratio is the square root of the electron count, about 7900.

Dark current integrates along with the signal. A dark current of 1 pA per pixel adds about 62,000 electrons in 10 ms, 0.1% of a 10 pC well; at longer integration times, or at higher temperature, where dark current roughly doubles every several kelvin, it fills a noticeable fraction of the well and adds its own shot noise. InGaAs arrays have much higher dark current than silicon because of the smaller bandgap, so they are commonly cooled with a thermoelectric cooler and operated at shorter integration times.

Crosstalk

Carriers generated deep in the substrate, or light scattered in the package, can reach a neighbouring pixel. Crosstalk to adjacent pixels is typically specified at the level of a percent or less, and it matters most where a strong spectral line sits next to a weak one.

Where arrays are used

The diode-array detector (DAD or PDA detector) in UV-Vis spectrophotometers and HPLC systems places the array after a diffraction grating, so the complete absorbance spectrum is captured at each time point instead of scanning a monochromator. Compact fiber spectrometers and wavelength monitors use the same arrangement: 1024 pixels at 25 µm pitch span 25.6 mm, and a grating that spreads 400 nm across that length gives about 0.39 nm per pixel. Position sensing uses a four-element array, the quadrant photodiode, discussed below.

Comparison with CCD and CMOS image sensors

A CCD or CMOS image sensor is also an array of photosensitive sites, and the boundary is partly a matter of convention (see CCD vs CMOS image sensors). The term photodiode array usually refers to devices with large pixels, often 25 µm by 0.5 to 2.5 mm tall in spectroscopic arrays, large well capacity, and simple readout, or to individually wired arrays with continuous output. Image sensors have small square pixels, full wells typically of 10⁴ to 10⁵ electrons, and low read noise, which suit low-light imaging.

The wider family of detectors, from single photodiodes to cameras and photomultipliers, is compared in Types of photodetectors.

Common questions

What is a photodiode array detector in HPLC?

It is a UV-Vis absorbance detector in which light passes through the flow cell and is then dispersed onto a linear photodiode array, so the detector records the full spectrum, typically about 190 to 800 nm, at every point in the chromatogram. The spectra support peak identification and purity checks.

How does a quadrant photodiode measure position?

A quadrant photodiode divides one circular detector into four elements A, B, C, D, numbered counterclockwise from the upper right. The spot position is taken from normalized differences such as $x \propto (A + D - B - C)/(A + B + C + D)$. If the right half collects 0.55 of the power and the left half 0.45, the normalized difference is 0.10; the conversion to distance depends on the spot size and is calibrated.

References: S. M. Sze and K. K. Ng, Physics of Semiconductor Devices 3rd ed. (Wiley, 2007); Saleh & Teich, Fundamentals of Photonics 3rd ed. (Wiley, 2019); G. H. Rieke, Detection of Light: from the Ultraviolet to the Submillimeter 2nd ed. (Cambridge University Press, 2003).