Absorbing light, making energy

An exclusive interview with a test engineer

By Martin Rowe, Senior Technical Editor -- Test & Measurement World, 11/1/2010 12:00:00 AM

Greg Dudgeon was the first full-time test engineer hired by Sionyx, a start-up company located in Beverly, MA, that manufactures light-absorbing silicon devices. Using a process called shallow-junction photonics, also called "black silicon," the company makes products that are orders of magnitude more sensitive to light than devices made using other processes. Dudgeon has spent most of his test-engineering career with start-up companies. His previous job was at Luminus Devices. Senior technical editor Martin Rowe spoke with Dudgeon by telephone.

Every other month, we interview an electronics engineer who has test, measurement, or inspection responsibilities. To participate in a future column, contact Martin Rowe. To read past "Test Voices" columns, go to www.tmworld.com/testvoices.

Q: What's involved in testing light-sensitive devices?
A: As you might expect, light-sensitive silicon devices are the opposite of LEDs. Because Sionyx is a start-up company, we're still working to improve our manufacturing processes.

We shine a light of known power on the DUT (device under test) and measure its output current as a function of light power, wavelength, and bias voltage. When testing at the wafer level, we use wafer probes to reach the test points. A wafer may have several dozen wafer probes on it at any time because we often need to connect to many bond pads. The devices on a wafer are designed for different purposes, and their responses depend on light intensity and wavelength. We use SMUs (source-measure units) to provide bias voltage and to measure output current. Test stations use switching cards to switch the SMUs to the wafer probes. Data-acquisition cards let us control the system. Analog voltages from the cards control shutters. The switches let us connect an SMU to a set of wafer probes, and they let us illuminate different light sources.

Q: What kind of light sources do you use?
A: We perform tests with a monochrometer that sweeps across a preset range of wavelengths. Instead of the monochrometer, we may use several LEDs or lasers, each designed to produce light at a different wavelength. A monochrometer is a single source, but if we use LEDs, we need the switches to power the LEDs at the proper time.

Q: How do you determine the power of the light?
A: The DUT is in a black box that prevents ambient light from affecting our measurements. Each day, we calibrate the system using a light source. Under LabView software control, the system adjusts light intensity and measures current from a reference sensor that has a known calibration curve. The software converts the measured current into light power. Because the system is automated, a calibration takes just a few minutes to run.

Q: What do you do with test data?
A: I wrote the test software to store the complete data set in a database. We can then pull data into a spreadsheet or statistics software for analysis. I use JMP statistical software rather than Excel to produce 2-D and 3-D plots and histograms. Plots let me spot trends that indicate how a device responds to different light power, wavelength, and temperature. Designers often want to pull data into Excel or Matlab for their own analysis.

Q: How do you work with designers to develop tests?
A: I use my background in physics and fiber optics to suggest tests to the designers. For example, I may recommend testing using both pulsed and continuous light. Using pulsed light lets you see how long a device takes to respond. This kind of testing is important for sensors, because of possible "image lag." That is, if the response is too slow, residual signal from a sensor's response will add to that of the next sample, producing an error.

Editor's note: The last paragraph has been modified from what appeared in print, because it contained an incorrect phrase.