What Went Wrong in a Colloidal Quantum Dot SWIR Sensor Test
The Test Looked Simple, but the Sensor Was Not
A colloidal quantum dot short-wave infrared sensor converts infrared photons into an electrical signal. The quantum dots absorb light, create mobile charge carriers, and pass those carriers through a photoconductive layer or photodiode structure. Depending on the design, the output may be measured as photocurrent, voltage, or a change in resistance.
In one failed bench test, the sensor appeared to work at first. A strong infrared source produced a visible signal, but the reading drifted, response time changed from one trial to the next, and the calculated sensitivity looked far better than the image quality suggested. The problem was not one dramatic component failure. Several small testing errors had been combined into one misleading result.
Responsivity Was Confused with Useful Sensitivity
The first mistake was treating responsivity as the complete measure of performance. Responsivity describes electrical output per unit of incident optical power, commonly expressed in amperes per watt. A large value can result from photoconductive gain, where trapped charges allow many carriers to circulate before recombination. That gain can increase signal, but it can also slow the detector and amplify noise.
The test also used total source power instead of the power actually reaching the active pixel. Reflections from the optics, aperture losses, and the detector window reduced the delivered signal. Meanwhile, the calculation assumed uniform illumination across the entire device. A calibrated detector and a defined illuminated area are necessary if the result is meant to be compared across samples.
- Measure dark current before illumination and after the device has stabilized.
- Record optical power at the detector plane, not only at the source.
- Separate rise time, fall time, responsivity, noise, and detectivity.
- Repeat measurements after changing bias and illumination history.
Traps Turned a Fast Test into a Memory Test
Colloidal quantum dots have surfaces rich in chemical bonds and defects. Ligand exchange improves electrical transport, but incomplete exchange, oxidation, or uneven film formation can leave trap states behind. These traps capture carriers and release them slowly. The result is persistent photocurrent, hysteresis, a long decay tail, or a response that depends on what the sensor saw several minutes earlier.
That behavior explained the inconsistent readings. The sensor was illuminated continuously, then measured as though every trial began from the same electrical state. It did not. A five-second exposure after darkness produced a different output from a five-second exposure following a bright calibration pulse. The test sequence had effectively changed the device before each measurement.
A reliable protocol should include a defined dark-rest period, a fixed illumination duration, and a recorded recovery interval. If the response does not return close to its baseline, the test should report persistence rather than quietly treating the remaining current as signal.
Noise and Bias Hid the Real Failure
The test used a higher bias voltage to increase photocurrent, but that choice raised dark current and electrical noise. Dark current is not merely an offset to subtract in software. It contributes shot noise, while imperfect contacts and surface defects can add low-frequency noise. At low modulation frequencies, flicker noise may dominate even when the average signal looks impressive.
The wiring introduced another problem. Long unshielded leads picked up interference, and the measurement instrument shared a ground path with the light-source controller. The resulting ripple became especially visible when the sensor output was processed with a narrow lock-in bandwidth. The software produced a clean number, but the number represented both optical response and electrical contamination.
Before repeating the test, the setup should be checked with the sensor shielded from light. Compare open-circuit voltage, dark current, and noise spectra at several bias levels. A useful result should remain stable when cable routing, modulation frequency, and averaging time are changed within reasonable limits.
A More Honest Retest
For the retest, the device was allowed to rest in darkness, then measured at zero bias and at two modest reverse-bias levels. The optical beam was chopped, its power was measured at the sample plane, and the active area was mapped rather than estimated from the package outline. Each wavelength was tested in random order to prevent gradual heating or conditioning from tracking with the spectrum.
The revised data showed a lower but credible responsivity, a measurable recovery tail, and a noise floor that increased at low frequency. Those results were more useful than the original headline figure because they revealed the operating limits. The sensor could detect SWIR illumination, but it was not ready for high-speed imaging without better passivation, tighter process control, and improved readout shielding. Documentation practices for a small technical operation can also be organized using guidance from the U.S. Small Business Administration, especially when test records need to remain consistent across teams.
What the Failed Test Actually Taught
A failed colloidal quantum dot SWIR test often points to measurement discipline rather than a dead sensor. Persistent charge, incorrect optical-power estimates, excessive bias, and unexamined noise can each distort the result. how colloidal quantum dot SWIR sensors work The practical question is not whether the device produces a signal, but whether that signal remains repeatable after its optical, electrical, and thermal history are controlled.
The sensor is ready for serious evaluation only when its dark baseline, response timing, noise behavior, and wavelength response can be reproduced under a documented sequence. That standard may produce less spectacular numbers, but it produces data that can guide the next fabrication step.

