Statistical Analysis to Identify Volatile Organic Compounds (VOCs) in Breath from Hypoglycemia and Development of Cross-Selective Nanosensors for Its Detection

Ali Daneshkhah, Purdue University


Hypothesis: Trained canines can identify irregular health conditions via their sense of smell. This study hypothesizes that human breath during hypoglycemia (blood glucose < 70 mg/dl) is different than other conditions (blood glucose > 70 mg/dl) and statistical analysis on gas chromatography and mass spectroscopy (GC/MS) signal of these breath samples can detect hypoglycemia. Such analysis of the outcome could support the development of resistor based sensors for detecting volatile organic compounds (VOCs) linked to hypoglycemia. Method: GC/MS signals of breath samples from young type 1 diabetic patients (age 7 to 21 years old) are investigated to detect hypoglycemia. A statistical model is developed to distinguish hypoglycemia from the other conditions. The data analyzed is used to design and develop sensors to detect VOCs similar to those biomarkers found in human breath. Conducting polymer composites are selected to detect alcohols and ketones while monolayer protected gold nanoparticle based sensors are utilized to target hydrocarbons. Super hydrophobic poly(vinylidene fluoride-hexafluoropropylene) (PVDF-HFP) conducting composites are investigated for detecting ethanol, acetone, isoprene while rejecting water vapor. Fabrication of a porous thin film PVDF-HFP/carbon black (CB) sensors are proposed via spin casting method for targeting alcohols and ketones. Facile methods to enhance the sensitivity and selectivity of this thin film on one single substrate is developed. Addition of a polyethylene oxide (PEO) layer, infra-red (IR)-treatment, addition of DEC (diethyl carbonate), and addition of DEC/PEO are used to alter the sensitivity and selectivity of the in response to acetone (ketone) and ethanol (alcohol). An array of PVDF-HFP/CB before and after addition of the PEO layer is tested with dodecane (hydrocarbon), 1-octanol (alcohol), and valerophenone (aromatic ketone). Principal component analysis (PCA) based on 12 features extracted from the sensors is used to detect both type and concentration of each VOCs. Dodecanethiol coated gold nanoparticle (GNP) sensors have been developed to target dodecane. Two post treatment methods were explored to increase the sensor sensitivity: 1) addition of PEO layer and 2) Development of PVDF-HFP layer over dodecanethiol coated gold nanoparticles (GNPs). Effect of humidity in the carrier gas is investigated by testing the sensors at different humidity levels including relative humidities (R.Hs) of 0%, 45% and 80%. Results: The statistical analysis shows it is possible to detect hypoglycemia condition from other conditions with sensitivity of 91 (95% confidence interval (CI) [57.1, 94.7]), and specificity of 84 (95% CI [73, 92.7]). The proposed thin PVDF-HFP/CB sensor could target ketones (acetone) and alcohols (ethanol). The described sensor treatments altered PVDF-HFP/CB sensor selectivity to ethanol from 0.32 to 1.74. Addition of a PEO layer enhanced the sensitivity of the PVDF-HFP/CB sensor up to 320%, 180 %, and 32% to dodecane, valerophenone, and 1-octanol, respectively. Dodecanethiol coated gold nanoparticle (GNP) based sensors are more selective to dodecane than PVDF-HFP/CB. Addition of PEO layer over the dodecanethiol coated GNP film enhanced the sensor’s relative response up to 61% to dodecane. Development of a PVDF-HFP layer over the film enhanced the relative response of the dodecanethiol coated GNP sensor up to 150%, 193 % and 46% to dodecane, valerophenone, and 1-octanol, respectively. The modified sensors also demonstrated higher sensitivity at higher humidity levels. Conclusion: This study shows how breath can be a reliable source for detection of hypoglycemia. The proposed PVDF-HFP/CB sensor detects low concentration of the 1-octanol and valerophenone. Addition of a PEO layer to the film further enhances the sensitivity and selectivity of the sensors. Dodecane protected GNP could be used for better detection of dodecane. The sensor’s sensitivity to the targeting VOCs can be further enhanced by post treatment methods. Addition of a PEO layer on dodecane coated GNP enhances the sensitivity of the sensor to dodecane. Developing a PVDF-HFP film over dodecanethiol coated GNPs film further enhances the sensor’s sensitivity and selectivity to valerophenone. The sensors’ fabrication, testing methods, and results are presented and discussed herein.




Jung, Purdue University.

Subject Area

Electrical engineering|Nanotechnology

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server