ZIF Nanocrystal-Based SAW Electronic Nose to Detect Diabetes in Human Breath

In the present work a novel, portable and innovative eNose composed of a surface acoustic wave (SAW) sensor array based ZIF-8, and ZIF-67 nanocrystals (pure and combined with gold nanoparticles) as sensitive layers has been tested as a non-invasive system to detect and differentiate disease markers, such as acetone, ethanol and ammonia, related with diagnosis and control of diabetes mellitus through exhaled breath. The sensors have been prepared by spin coating, achieving continuous and homogenous sensitive layers. Low concentrations (5 ppm, 10 ppm and 25ppm) of the marker analytes were measured, obtaining high sensitivities, good reproducibility, short time response and fast signal recovery.


Introduction
Nowadays one of the great challenges of science is the diagnosis of diseases in the least amount of time and using non-invasive techniques.This strategy intends to provide a higher quality of life for humans and reduce the mortality rate.Additionally, an early treatment of diseases and its complications has an important economic impact by helping to avoid or reduce treatment costs.The last portion of deeply exhaled breath, representing alveolar air, can be considered the headspace gas of blood.Exhaled breath, recognized mainly through the sense of smell, is a method used for a long time to disease diagnosis.This method was abandoned due to the emergence of new accurate and effective techniques, despite the fact that any of them were highly invasive.Over the last few decades, an important advance in the gas analysis technologies has re-launched the idea to diagnose diseases through exhaled breath.Various studies conducted by these analysis techniques, such as gas chromatography-mass spectrometry (GC-MS), proton transfer reaction-mass spectrometry (PTR-MS), selected ion flow tube-mass spectrometry (SIFT-MS), ion mobility and optical absorption [1][2][3][4][5], have shown a link between the chemical composition of the exhaled breath and certain diseases, chemical compounds present in the exhaled breath that change due to diseases are known as markers.The above conventional systems are accurate, but they are also bulky, expensive, and require highlyqualified operators, facts that increase the demand of low-cost systems with high sensitivity and low dimensionality based on solid-state chemical sensors, with different detection principles such as impedance [6], resistive [7,8], optical [9] and piezoelectric [10,11], surface acoustic wave (SAW) being one of the most sensitive devices among piezoelectric ones [12].At the nanoscale, materials and their advanced design features have led to a new generation of chemical sensors with enhanced sensitivity and response time [13][14][15].Due to its unique porous structure zeolites have been used to detect gases [16].However, in the last years, organic zeolites such as zeolitic imidazolate frameworks (ZIFs) have attracted major attention as gas detectors [17], because they offer two primary advantages over conventional zeolites.First, they have larger porous size (about 1.16 nm for ZIF-8 and ZIF-67) and usually smaller crystal size, resulting in higher surface area.Second, hydrophobic behaviour is more pronounced in many ZIFs [18,19].The World Health Organization (WHO) recognize diabetes mellitus, known as diabetes, as a serious and chronic disease that in 2012 caused 1.5 million deaths.A recent study from reports in 2015 of 111 countries, estimated that there were 415 million people with diabetes aged 20-79 years, 5 million of deaths attributable to diabetes, and the total global health expenditure due to diabetes was estimated on 673 billion US dollars representing a substantial clinical and public health burden [8].Moreover, the number of cases of diabetes among youths [20] and infancy [21] increased in recent years but recent incidence trends are lacking and only statistical data for some countries are available.The presence of ketones in the exhaled breath is a warning sign of ketosis that is related to fat catabolism either due to carbohydrate deprivation or to its lack of utilization in persons with diabetes.This condition is known as diabetic ketoacidosis and requires immediate treatment.One type of ketone, known as acetone, provides a non-invasive measure of ketosis through breath.The basal level of acetone in a healthy situation can be around 2 ppm [5,22].Adults following low-carbohydrate diets can have elevated levels of up to 40 ppm [23][24][25], and poorly controlled diabetes can cause ketoacidosis which can increase acetone concentration up to 1250 ppm [22,26].However, human exhaled air is a complex mixture of chemical compounds, making the detection and stage classification of a determinate disease through a unique marker difficult, and, consequently different disease markers need to be considered as indicators.Another marker related with blood glucose concentration and present in exhaled breath is ethanol.Though ethanol is not directly produced by any known mammalian cellular biochemical pathway, it may increase in exhaled gas mixtures because of alcoholic fermentation of an excessive overload of carbohydrate-rich food in conjunction with overgrowth of intestinal bacteria.Ethanol used in combination with exhaled acetone allowed the prediction of fluctuating plasma glucose concentrations in a multi-linear regression model [27][28][29][30], therefore it could be helpful for determining diabetes through breath.On the other hand, diabetes represents the cause of half of cases of renal failure cases.Kidney failure is related with ammonia levels higher than 3 ppm in exhaled breath [5,25,31,32].Consequently, a finger print using breath levels of acetone, ethanol and ammonia could be a non-invasive predictor of diabetes, its control, and a proxy of damage of the disease.In the present work, a SAW eNose, based on ZIF nanocrystals as sensitive layers, has been tested to acetone, ethanol, and ammonia as a potential non-invasive system to the diabetes diagnosis and conrol.

Synthesis of ZIF-8 and ZIF-67
ZIF-8 and ZIF-67 samples were synthesized using the aqueous method reported elsewhere [33].For ZIF-8 synthesis, a solution of 1.17 g of zinc chloride dissolved in 8 mL DI water was added into a solution of 2-methylimidazole (2MeIM) (22.70 g) dissolved in 80 mL DI water, to yield a molar ratio of 2-methylimidazole to zinc of 70:1.The mixture was stirred at room temperature for 5 min.The product was collected by centrifugation (24000 rpm, 10min), washed in DI water three times and dried at 65°C for 24h in an oven.ZIF-67 was synthesized identically to the ZIF-8 material as described above, now replacing zinc chloride with the equivalent quantity of cobalt chloride hexahydrate.A scheme of the synthesis paths is described in Fig. 1 [34,35].

Synthesis of gold nanoparticles
Gold nanoparticles (AuNP) were prepared according to the procedure described in reference [36].An aqueous solution of HAuCl4 (0.001M, 40 mL) was placed into a 250 mL round bottom flask.The solution was heated to 90 °C followed by the addition of sodium citrate aqueous solution (38.8mM, 2 mL) into it while stirring for 15min.After cooling down to room temperature, the solution was centrifuged three times with ethanol and three times with DI water, finally the precipitate was redispersed in DI water, resulting a water solution with AuNP of ~5 nm.

ZIFs nanocrystal characterization
The synthetized samples were kept at room conditions and characterized using: Fourier transform-infrared spectroscopy (FTIR), X-ray diffraction (XRD), scanning electron microscope (SEM) and energy-dispersive X-ray (EDS).FTIR spectra were recorded using a Thermo Nicolet NEXUS 670 FTIR spectrometer.The sample was diluted into KBr pellets in a 1:100 weight ratio (sample to KBr).The scanning range was 400-4000 cm −1 and the resolution was 4 cm −1 .XRD powder patterns were recorded with CuKα radiation in a D8 advance diffractometer from Bruker.The morphological features were examined by SEM.The SEM and EDS analysis were performed on a JEOL JMS-7600F.

Love-wave sensor
Love-wave (LW) are a specific type of SAW sensors based on shear horizontal (SH) waves guided by a layer with a lower propagation velocity than the piezoelectric substrate.The energy of the wave is confined in the guiding layer and any perturbation in it affects the acoustic wave velocity.The LW sensors used in the present work were designed with a delay line (DL) configuration.This device is based on a micro-electromechanical system composed of a piezoelectric material (ST-Quartz) with facing input/output aluminium interdigital transducers (IDT) on its surface, working at a 28 μm wavelength (λ), with a separation between its IDTs of 2100 μm (Fig. 2a).The SH waves were guided by a 3.1 μm thick layer of SiO2, on which sensitive layers were deposited.An oscillator circuit consisting of a DL, with an amplification stage and a coupler were used for measuring the changes in the velocity of the waves by means of the resonant frequency (Fig. 2b).

ZIF as sensitive layers
The eNose was based on a SAW sensor array with different sensitive layers to achieve a specific fingerprint for analytes of interest.The sensitive material samples were obtained mixing each main solution with a volumetric proportion of 75% (solution-1) and 25% (solution-2) (Table 1).Spin coating technique was used to deposit a thin layer of sensitive material.The process consisted of putting sample drops directly on the SiO2 guiding layer, and then the assemblage is rotated at a speed of 3000 rpm during one minute (Fig. 3).A suitable thickness for each sensor is obtained after depositing four times in a multilayer configuration achieving an optimal sensitivity for sensors.After ZIFs deposition process a thermal treatment at 180 °C with 50 ml/min nitrogen flow in a tubular oven was applied during 4 hours for extracting the excess of the 2MeIM organic ligand in the sensitive layer.

Experimental setup
The sensor array was tested to acetone, ethanol, and ammonia gas analytes diluted in synthetic dry air to obtain the required concentrations: 5 ppm, 10 ppm, and 25 ppm.The gas sample generator (Fig. 4) consists of two mass flow controllers, which were utilized for obtaining desirable concentrations at a constant flow of 100 ml/min.For the detection, a sensing system composed of a measuring instrument (eNose) manufactured to be operated with a SAW sensor array was used.Finally, sensor responses were acquired by a micro-frequency counter, and the information was transmitted wirelessly.A custom application was developed to display and store the sensor data in real time.

Structural and morphological characterization of ZIFs
The FTIR spectra of KBr diluted ZIF-8 and ZIF-67 samples show the characteristics adsorption bands of 2meImidazole ring vibrations, reported for this structures in previous works [37] (Fig. 5a).The XRD patterns of both ZIFs samples evidence the formation of a largely crystalline structure with long-range order (Fig. 5b) The position and relative intensity of the diffraction maxima are in correspondence with the literature statements for ZIF-8 and ZIF-67 frameworks [34,37,38].SEM images exhibited that the nanocrystals were configured like a continuous layer with very small particles (Fig. 6a).This fact is important because nanostructured layers had two advantages: first, the surface area of reaction with the gaseous environment is higher; second, the wave is propagated as in a continuous layer with very low scattering due to its wavelength (28 μm), which is much higher than the diameter of the nanocrystals.The micrographs also showed hexagonal shaped nanocrystals of 50 nm approx.for the ZIF-8 samples (Fig. 6b) and 200 nm approx.for the ZIF-67 samples (Fig. 6c).

Electrical characterization of the Love wave sensors
The LW sensors were characterized before and after depositing the ZIF nanocrystals sensitive layers.In the array, a reference device without sensitive layer was used (Fig. 7a) to compensate sensor responses for undesirable changes in temperature and pressure.The sensors were characterized by means of the automatic network analyzer (ANA Wiltron 360B) and the S21 parameter was used to measure insertion loss transmission.The frequency response of each sensor exhibited a frequency shift of the minimum insertion loss occasioned by the mass loading of the sensitive material (Fig. 7b  and c).On the other hand, increases of insertion losses (Table 2) were a consequence of the scattering due to the propagation of the SAW wave in the sensitive material.

Gas characterization
The SAW eNose based on a LW sensor array with ZIFs nanocrystals was tested to acetone, ethanol and ammonia markers related to diabetes mellitus disease.The sensors were exposed during two minutes to each analyte at concentrations of 5 ppm, 10 ppm and 25 ppm, and then the array was purged with dry synthetic air for 10 minutes.The LW sensors showed a notable and fast response, e.g., for 10 ppms of acetone a frequency shift of 275 Hz and a 90, around 30 s, with a complete recovery achieved after 10 min (Fig. 8a), 90 being defined as the time taken to reach the 90% of the frequency shift.The response of the sensor array to different concentrations of acetone (Fig. 8b), ethanol (Fig. 8c), and ammonia (Fig. 8d) showed a high frequency shift for the different sensitive layers tested, obtaining best sensitivities for S2 (ZIF67_Au) and higher responses for higher concentrations.Therefore, the eNose could be used to make a diagnosis in a few seconds and be repeated or carry out a new diagnosis after ten minutes.The reference sensor helped to compensate in the sensor array the pressure and temperature changes due to external factors, measuring only variations related to the interaction of the analytes with the sensitive layers.The noise level for LW sensors is 10 Hz/min, thus the minimum detectable measurement, three times signal-to-noise ratio, is a frequency shift of 30 Hz, obtaining the lowest detection limits for the sensor based on ZIF-8/AuNP, at 0.7 ppm, 0.3 ppm and 1 ppm for acetone, ethanol and ammonia, respectively.Therefore, sensors with nanocrystalline ZIF as sensitive materials allowed for the detection and discrimination of acetone, ammonia, and ethanol, as shown in the radial surface analysis to 10 ppm of the three markers (Fig. 9), therefore, the SAW eNose presented could be used to relate determined fingerprints with diabetes diseases.

Conclusions
A SAW eNose based on Love wave sensors combined with ZIF-8, ZIF-67, ZIF-8/AuNP and ZIF-67/AuNP as sensitive layers was tested to three breath markers of diabetes mellitus: acetone, ethanol and ammonia at concentrations of 5 ppm, 10 ppm and 25 ppm.These diabetes markers were detected with fast response time, around 30 s, and a complete recovery time, below 10 min.In addition, sensor array response showed as each marker has its own fingerprint.In conclusion, tests carried out in this work shows a properly performance of the SAW/ZIF eNose to be proved in future works as a prototype for a noninvasive system contributing to the diagnosis and control of diabetes mellitus in real cases.

Figure 1 .
Figure 1.Representative synthesis and crystal structures of ZIF-8 and ZIF-67 used as the sensitive layers of gas sensors.

Figure 2 .
Figure 2. a) Scheme representation of LW sensor and layer composition.b) Oscillator circuit used to read the resonant frequency.

Figure 3 .
Figure 3. Process sequence of sensitive layer deposition on LW devices by means spin coating technique.

Figure 4 .
Figure 4. Diagram of instrumentation and experimental setup for the eNose characterization.

Figure 7 .
Figure 7. Spectral response of the LW sensors (a) without (b) and (c) with sensitive layers.

Figure 9 .
Figure 9. Radial representation of the sensor array responses to 10 ppm of acetone, ethanol and ammonia.

Table 1 .
Different composition of the sensitive layers used in the SAW sensors included in the eNose..

Table 2 .
Insertion loss and frequency shift with respect to the reference sensor, 165 MHz and 18.3 dB.