UVG Photodiode Operating Principles: Quantum Efficiency, Radiation Hardness & Stability

UVG photodiodes are UV-enhanced silicon detectors built by Opto Diode, a division of ITW, in Camarillo, California. Opto Diode specifies the series for high-sensitivity detection of deep ultraviolet (DUV) and extreme ultraviolet (EUV) radiation, with responsivity across the 190 nm to 400 nm range. What separates them from ordinary UV-enhanced silicon is a pair of construction choices: silicon with no surface dead region, and a radiation-hard oxynitride entrance window. Together those give 100% carrier collection efficiency — which is what lets the above-unity internal quantum efficiency that ultraviolet photons produce in silicon reach the external circuit — and enough surface stability to be used open-faced without a fused-silica cover. This page explains the physics behind that and maps it onto the five current models.

Source note: this article is adapted from the IRD technical note Silicon Photodiodes – UVG Series (PDF), preserved online. Opto Diode acquired IRD in 2011, so the note is a first-party account — but a historic one. Its measurements stand; the program status it describes is late-1990s to early-2000s. Where the note and the current datasheets differ, the datasheets govern, and this page follows them.

How a UVG photodiode converts ultraviolet light into current

When the diode is exposed to photons of energy greater than 1.12 eV — a wavelength shorter than 1100 nm — electron-hole pairs are created in the silicon. The p-n junction electric field separates those photogenerated carriers, and a current proportional to the number of pairs created flows through the external circuit. That much is true of any silicon photodiode.

Ultraviolet photons behave differently. Below a wavelength of roughly 350 nm, a single photon creates more than one electron-hole pair, a result established by Korde and Geist (Applied Optics, Vol. 26, 5284–5290, 1987). The external circuit therefore sees more electrons than there were incident photons, and the internal quantum efficiency of the device rises above unity.

Why collection efficiency reaches 100%

Gain in the silicon is worth nothing if the carriers never reach the contacts. The first of the two properties that define the UVG series is the absence of a surface dead region: no photogenerated carrier recombination occurs in the doped n-type region, and none occurs at the silicon–silicon dioxide interface. Because absorption depths for the majority of UV photons are less than one micrometer in silicon, every carrier is generated within reach of the junction field, and the result is complete collection by the external circuit.

The evidence is the curve itself: the note points to the flat region from 310 nm to 640 nm as the signature of 100% collection efficiency.

Complete collection also makes the device calculable, which is what interests metrologists. Because collection efficiency is 100%, the external quantum efficiency of a UVG photodiode across roughly 160 nm to 600 nm is simply the internal quantum efficiency multiplied by one minus the reflectance — and the reflectance can be measured or calculated. The note publishes measured reflectance from 150 nm to 1100 nm for UVG devices with a 70 nm oxynitride front window. Separately, 1 cm² UVG-100 diodes were used to determine the quantum yield of silicon (electron-hole pairs generated per absorbed photon) over the 254 nm to 160 nm region for the first time (Canfield et al., Metrologia, Vol. 35, 329–334, 1998) — the measurement that makes trap detectors for absolute flux determination possible in deep UV lithography and in photorefractive and phototherapeutic keratectomy.

The oxynitride entrance window, and what radiation hardness means here

The second defining property is the radiation-hard, junction-passivating oxynitride protective entrance window. Silicon photodiodes have long been known to lose responsivity or linearity under intense UV flux, and the note traces that instability to the inferior quality of the silicon–silicon dioxide interface. Oxynitrides resist ionizing radiation and impurity diffusion better than pure oxides; the working explanation is that breaking an Si–N bond takes far more energy than breaking an Si–H or Si–OH bond, so fewer interface states are created after UV exposure.

The measured consequences:

• Less than 2% responsivity degradation after megajoules/cm² of 254 nm exposure and after tens of kilojoules/cm² of 193 nm exposure. The 254 nm test used a 20 mW/cm² low-pressure mercury lamp; the 193 nm test used a Lambda Physik ArF excimer laser at a 100 Hz pulse repetition rate and an energy density of 200 mJ/cm² (3.9 W at 100 Hz).
• No change in ultraviolet-visible quantum efficiency after exposure to 100% relative humidity for several weeks.
• Less than 1% change in 254 nm responsivity under a three-part accelerated test: one week at 100% relative humidity at room temperature, two weeks under 20 mW/cm² of 254 nm radiation, and four weeks of baking at 100 °C.

Diodes fabricated by incorporating nitrogen into the passivating oxide were reported at 1 G-rad (SiO₂) hardness by Korde, Cable and Canfield (IEEE Transactions on Nuclear Sciences, Vol. 40, no. 6, 1655–1659, 1993), a figure the note puts at about 10,000 times the hardness of the p-on-n photodiodes then in common use. Two further long-term degradation mechanisms — latent recombination centers formed by metallic impurities such as silver, and moisture penetration over time — are addressed by fabricating in an extremely clean environment and by the nitrogen at the interface, which makes it insensitive to impurity penetration.

The practical payoff is that the fused-silica protective window normally fitted to a UV photodiode becomes unnecessary. That open-face configuration is a real advantage wherever fused-silica window interference effects are problematic.

Space heritage and calibration use

n-on-p diodes are more radiation-hard than the more common p-on-n devices (Korde et al., IEEE Transactions on Nuclear Sciences, Vol. 36, 2169–2175, 1989), which makes the UVG construction better suited to space missions than conventional silicon photodiodes. UVG photodiodes with 4 mm and 5 mm diameter active areas were used in the Multi-Angle Imaging SpectroRadiometer (MISR), launched in December 1999 as part of NASA’s Earth Observing System (Jorquera et al., Proc. IGARSS ’94). The note also records that the devices were slated for an Argentine satellite.

On the metrology side, the note reports uniformity data for a 10 mm × 10 mm IRD photodiode used by NIST as a transfer standard in the 5 nm to 254 nm region, and records that national laboratories including NIST and PTB (Germany) were, at the time of writing, evaluating UVG photodiodes for use as transfer standards. That is heritage, not a current accreditation.

Linearity at high photocurrent

Linearity decides whether a detector can be used for absolute measurement, because non-linearity errors are difficult to correct after the fact. Using the standard ac-dc method at 430 nm with no reverse bias, a widely used p-on-n photodiode of equivalent area showed a noticeable decrease in responsivity at photocurrents greater than 500 µA. The UVG-20 showed only a 0.02% decrease in responsivity at a photocurrent of 3 mA, six times further up the current scale. The note attributes the difference in linear range to the difference in series resistance between the two constructions.

The failure mode in the other direction is supralinearity: responsivity that rises with flux. In a p-on-n device with 80% internal quantum efficiency at 430 nm, the missing 20% is photogenerated carrier recombination in the front region. Increasing flux fills trap centers, minority carrier lifetime rises, recombination falls, and measured responsivity climbs. UVG series photodiodes have 100% internal quantum efficiency at 430 nm — no front-region recombination to suppress — so they show no supralinearity. At 980 nm the same comparison put noticeable supralinearity in the p-on-n above 20 µA of photocurrent, with none observed in the UVG-20, and at high irradiance the p-on-n lost responsivity far more rapidly.

Applying a reverse bias extends the linear range of UVG photodiodes further when measuring UV radiation, which is why the reverse breakdown voltages listed below matter even in nominally photovoltaic applications — they are absolute limits measured at I_R = 1 µA, not recommended operating biases.

Uniformity and temperature behavior

Response uniformity over a 10 mm × 10 mm active area was within ±0.5% when scanned with a 254 nm photon beam of 1 mm diameter. For comparison, the note reproduces ±2% uniformity for the UV-enhanced diode NIST was then using as a transfer standard for the 200 nm to 400 nm range (Shaw et al., Rev. Sci. Instrum., Vol. 72, 2242–2247, 2001). Uniformity of that order is what gives better reproducibility and a lower overall measurement uncertainty.

Two temperature coefficients are worth designing around:

• Responsivity increases by about 0.045% per °C at 254 nm. Because the device already has 100% internal collection efficiency, the note attributes this to an increase in quantum yield at higher temperatures and, in part, to the change in surface reflectance — and notes it is smaller than the figures reported by other manufacturers of UV-enhanced photodiodes.
• Shunt resistance halves for every 7.5 °C rise in temperature. A shunt-resistance number is therefore only meaningful alongside the temperature it was measured at, which matters when you are budgeting offset and noise in photovoltaic mode.

Handling, windows and packaging

Windowless UVG devices are supplied with protective epoxy on their wire bonds, and the note warns that this constrains the thermal envelope: it advises operating near room temperature, neither heating the device for outgassing nor cooling it to reduce noise. The current datasheets rate all five models for -20 °C to 80 °C operation; treat the note’s room-temperature guidance as the conservative handling limit for an epoxied windowless part. The reason is mechanical — the difference in thermal coefficient of expansion between the wire and the epoxy would stress the bonds. If necessary, UVG series diodes can be provided without epoxy on the wire bonds, or with a UV-transmitting window such as fused silica or magnesium fluoride.

The current models differ in exactly this respect. The UVG12 ships with no cap at all, for maximum responsivity, with a sacrificial cap taped on for shipping only. The UVG5S has a welded cap with a UV glass window and the UVG20S an epoxy-bonded UV quartz window; the UVG100, like the UVG12, ships under a temporary protective cover plate that is removed before use; the UVG20C is supplied in a TO-8 package.

Where the response ends

The internal quantum efficiency of a UVG series diode drops rapidly after 700 nm, owing to the limited silicon thickness. That is a consequence of optimizing for the ultraviolet rather than a defect, but it does make the UVG series the wrong choice for near-infrared work. Historically, IRD also supplied p-on-n variants (UVG-PN100, UVG-PN20 and similar) that showed over 97% internal quantum efficiency at 950 nm; those are legacy IRD parts and are not part of the current five-model UVG line. For other bands, start from how to choose a photodetector.

The current UVG models

Opto Diode lists three headline features for the series — high UV responsivity, low dark current, high quantum efficiency — and three application areas: semiconductor photolithography, aerospace and environmental sensing, and UV spectroscopy. Five models are in the current catalog.

The feature line is the wavelength range each datasheet calls out on its own front page; the series specification remains 190–400 nm, and the UVG20C is the one part featured for electron detection rather than for a wavelength band, though the category page lists it, like the others, for 190–400 nm.

Each figure carries the min / typ / max designation the datasheet assigns it; consult the datasheet for the full test conditions. Complete listings and ordering information are on the UVG photodiode range page, and active area, window and package can all be built to order — see custom photodiodes and detectors. For the shorter-wavelength family built on the same silicon technology, see AXUV photodiode operating principles.

Frequently asked questions

Adapted from the IRD technical note “Silicon Photodiodes – UVG Series”. Reviewed and updated July 2026.