It’s always interesting to see how various technical disciplines can be merged to create new functions and instruments, and a recent university-based breathalyzer project is a good example. The effort, by a team at Laval University in Canada, uses the screen glass of a standard smartphone, a photodiode, and data-reduction/analysis software to assess a user’s blood alcohol content (BAC) by simply breathing on that glass and waiting a few seconds (Figure 1).
Figure 1 The image shows photograph of the prototype (a), scheme of the photodiode installation on the side of the glass screen (b), the electronics (c), and in (d), screenshots of the application software steps performed before displaying the BAC result shown in (a). Source: MDPI
The project leverages an optical characteristic of the glass; edge-on, it can act as an optical waveguide. The overall concept is simple and elegant. It uses the planar waveguide formed by the dense anti-scratch stress layer in multimedia device screens made of toughened glass such as Corning Gorilla and AGC Dragontrail glasses.
In operation, the user simply breathes onto the lit phone screen, and moisture microdroplets condense and form on the screen (Figure 2). The edges of the droplets initiate a strong oblique reflection (total internal reflection) which can be coupled into the glass screen (just as in an optical fiber). A photodiode at the edge of the screen—incorrectly labeled “CCD camera” in the drawing, perhaps that was the company’s original plan—captures this light and subsequent reduction in light due to evaporation. By analyzing the rate of light decrease due to evaporation, the associated data-analysis software can provide a reading of the alcohol concentration in that exhaled breath.
Figure 2 The image explains optical breathalyzer principle, showing evaporation of the microdroplets when a person breathes on a glass screen (a). The inset is a zoom on the microdroplets (b). The light from the smartphone display is not coupled to the planar waveguide on the surface of the glass screen (c). Finally, with water droplets on the glass screen, the light from smartphone display is coupled to the planar waveguide due to the strong oblique reflections at the edge of the droplets and is guided to the side of the screen (d). Source: MDPI
To capture the light at the edge of the screen, researchers used an optical epoxy with matching index of refraction to attach the photodiode. The output of the photodiode goes to a commercially available, galvanically isolated analog front-end (AFE) designed for photodiode and similar sensors which deliver a digitized, USB-compatible output. This AFE looks interesting; it’s a 45 × 20 mm Yocto-milliVolt-Rx module from Swiss-based Yoctopuce. It comes with a 400-page manual, covering several interface and application-software scenarios (Figure 3).
Figure 3 The analog front-end and digital interface between the photodiode and USB port are implemented by the galvanically-isolated Yocto-milliVolt-Rx module. Source: Yoctopuce
As with so many other real-world sensor-related projects, the challenge with this one is calibration because there are numerous corrections and nonlinearities that must be factored into the algorithms. For example, ambient temperature and humidity affect the rate of evaporation; furthermore, the spread of these microdroplets is not uniform across the screen and this needs a correction factor as well.
For a BAC-sensing situation, you need lots of volunteers to collect a meaningful number of samples—easier said than done, strange as it sounds—and you need to use them under both lab and field conditions. You also need a qualified standard for comparison; Laval University team used a costly commercial breathalyzer.
The team first tested the prototype on a volunteer in a laboratory. Then, to identify any parameters that might affect the measurements, researchers tested the prototype in real-world environments, taking 140 measurements from 36 volunteers over the course of three gatherings where alcohol was served. Analysis of the 140 evaporation curves over time allowed the team to identify several parameters and conditions affecting the measurements as well as ways to improve the accuracy of the device.
Details of the work, including results of the many tests, are in the paper “Smartphone Screen Integrated Optical Breathalyzer” published in the journal MDPI Sensors. The researchers further plan to use the prototype to acquire a large dataset of BAC optical signatures to train a deep neural network and then use machine learning to obtain accurate BAC measurements. Using this approach, the researchers hope to develop an ecological, inexpensive, and discreet breathalyzer for smartphones and smartwatches.
Have you leveraged electro-optical principles and available modules to create a unique electro-optical instrument? How did you calibrate and compare its performance to a credible level or standard?
- When Your Sensors Mislead You
- Hot car-seat deaths and sensor solutions
- Sensors Without Wires, But Not Wireless
- When sensors are easy, but sensing is hard
- Super-Sensitive Magnetic Sensor Exploits Diamond Defect
- Sensor, signal conditioning define the real performance envelope
- Sophisticated Sensors, Extreme Conditioning, Advanced Algorithms Yield Amazing Geolocation Results