Study in silico and experimental of a double-pass Mach–Zehnder interferometer sensor for acetone detection as a biomarker of diabetes mellitus

Acetone detection through breath analysis has gained relevance, as it is a biomarker of diabetes mellitus, with optical fiber sensors being a promising alternative for constructing compact, inexpensive, and rapid measurement devices. Furthermore, the design of an in silico optical fiber sensor is essential, since it provides information for optimizing and validating experimental results. This work focuses on in silico and experimental study of an intrinsic double-pass Mach–Zehnder interferometer (MZI) in optical fiber for acetone detection. The device was based on two long-period fiber gratings (LPFG) recorded in cascade with a sensing film of polydimethylsiloxane (PDMS) deposited over the MZI, and a silver thin film deposited at the tip of the optical fiber as a reflective film. A model based on coupled-mode theory (CMT) and the transfer-matrix method (TMM) was employed to simulate the reflection spectrum of this sensor when exposed to acetone, where the PDMS sensing film and acetone were treated as a mixture. Sensor responses were studied using multivariate analysis techniques, such as principal component analysis (PCA) and projection to latent structures (PLS), to analyze the response behavior in function of acetone concentration. Experimental sensors were constructed and characterized with the aim of comparing the results to those of the in silico sensor. A qualitative analysis was performed, revealing multiple similarities in the experimental and simulated results, especially in the spectra profiles and, PCA and PLS curve shapes. A minimum experimental limit of detection (LOD) of 1.46 ppm was obtained for a PLS regression.