|18||Low-Cost Integrated Spectrometer
|Project Members: Lukas Janavicius - janavic2, Drew Ingram - andrewi2
Problem: Optical spectrometers play a critical role in the characterization of chemicals and materials. However, budget digital spectrometers start at over $1000. This cost barrier effectively limits techniques like Raman spectroscopy to Universities and other research institutions. We aim to bring the cost of the device to under $100, as to lower the barrier of entry to spectroscopy techniques.
Solution: A spectrometer's cost lies in its expensive optics, dedicated high-frequency data collection hardware. Our proposed solution is to eliminate costly optical components by integrating an optical circuit in acrylic plastic and collecting the diffracted light using a linear CCD image sensor driven by a low-cost microcontroller [1, 2].
Solution Components: An integrated photonic circuit, data acquisition hardware, MCU software, and Host software.
Photonics: To spatially resolve the input light's spectrum, we propose fabricating an integrated photonic diffraction element . Specifically, we aim to make an elliptical Echelle grating, such that the output spectrum is tuned by changing the angle of the input waveguide . Although Echelle gratings can support broadband spectrometers, we aim to optimize our photonic circuit for the wavelengths of 400-800nm. Such wavelengths fit within our selected CCD's response curve, while also offering applications in spectrometry.
Data Acquisition: Our proposed solution requires three distinct electronic components; a TCD1103 linear CCD image sensor, a high-speed ADC to read the CCD data, and an ESP32 to acquire and send data to a host device. The Toshiba TCD1103 belongs to a family of devices shown to work with 8-bit MCU's, although this particular model can deliver faster data rates than demonstrated . To read the CCD data, an ADC and ESP must collect around 1 MSPs and stream the data to a host device.
MCU Software: To extract 1 MSPs, the ESP32 must communicate with the ADC over SPI operating at its maximum frequency, 40 MHz. The contents of the device's memory must be dumped to a host device over Wifi to avoid saturating the memory. We will accommodate these speeds by splitting Wifi and SPI communication across the ESP's two cores, with a message queue relaying the data between them.
Host Software: Raw data will stream directly to a PyQt application running on the User's PC. After correcting the TCD1103 data with the device's spectral response curve, and calibrating the positional data with a known source, a pyqtgraph presents the data to the user.
Criterion for Success and Challenges: We aim to resolve wavelengths in the 400-800nm range, we will assess the resolving power of our spectrometer by characterizing the emission spectra of an argon plasma. Perhaps the greatest challenge of our project is in the design of the photonic circuit; to minimize losses, we must simulate our structure, and characterize the lithography process of the circuit. However, we are unsure of which FTDT package to use in this situation. Fortunately, after our photonic design is verified, we can pattern and develop the circuit in under 10 minutes with only rubbing alcohol, ensuring we can tune the processing parameter rapidly.
 X. Ma, M. Li, and J. J. He, “CMOS-compatible integrated spectrometer based on echelle diffraction grating and MSM photodetector array,” IEEE Photonics J., vol. 5, no. 2, 2013.
 R. Cheng, C. L. Zou, X. Guo, S. Wang, X. Han, and H. X. Tang, “Broadband on-chip single-photon spectrometer,” Nat. Commun., vol. 10, no. 1, Dec. 2019.