Open Access
Add to BibSonomy
Abstract
The detection of acetone in the gaseous form in exhaled breath using an integrated sensor can provide an effective tool for disease diagnostics as acetone is a marker for monitoring human metabolism. An on-chip acetone gas sensor based on the principle of Mach-Zehnder interferometer is proposed and demonstrated. The sensing arm of the device is activated with a composite film of polyethyleneimine and amido-graphene oxide as the gas-sensitive adsorption layer. The composite film demonstrates good selectivity to acetone gas, can be used repeatedly, and is stable in long-term use. Room temperature operation has been demonstrated for the sensor with high sensitivity under a 20 ppm acetone environment. The detection limit can reach 0.76 ppm, making it feasible to be used for the clinical diagnosis of diabetes and the prognosis of heart failure.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
According to the World Health Organization statistics, the global prevalence of diabetes is increasing annually, with a 5% increase in mortality rate during the past decade. The disease is more prevalent in a younger population [1]. Exhaled breath acetone (EBA) is widely regarded as the most reliable and valuable marker for diabetes detection [2]. Diabetic ketoacidosis exhibits a voluminous amount of exhaled acetone that can be used as a marker for monitoring the disease progression. According to the latest medical reports, the concentration of EBA from healthy people is about 0.3 ppm to 0.9 ppm [3]. However, the concentration of EBA from diabetic patients is generally above 1.8 ppm, and it is particularly important to realize the low limit detection of EBA. Also, according to the American College of Chest Physicians report, EBA can be used as a new biomarker of the severity of heart failure [4]. Patients with heart failure can be differentiated into chronic heart failure and acute decompensated heart failure according to the severity of the disease. The EBA concentration in patients with chronic heart failure is between 1.56 ppm and 5.03 ppm, while the EBA concentration in patients with acute decompensated heart failure is between 8.28 ppm and 34.96 ppm. Therefore, the detection limit of acetone and its quantification at the ppm level is essential for diagnosis and monitoring disease progression. The widely used acetone detection methods are gas chromatography, mass spectrometry, and gas mass spectrometry-chromatography in clinical practice. These methods have a high sensitivity and simple operation procedure. However, sample storage requirements, bulky equipment size, analysis speed, and operation cost are detrimental to its use in actual clinical diagnosis.
Miniaturized sensors based on semiconductors have proven to be highly sensitive. For example, a SnO2 semiconductor nanowire-based acetone sensor has good selectivity to acetone in various volatile organic compound (VOC) gas environments [5]. However, the semiconductor sensors have poor long-term stability. A hierarchical structure of ZnFe2O4 core-shell microsphere acetone sensor based on template-free synthesis was reported to be more stable [6]. The sensor exhibits a stable operation beyond 30 days of continuous measurements and is sensitive to 20 ppm of acetone. A solid electrolyte-type acetone sensor can detect acetone by measuring the electric potential change. For example, solid electrolyte materials sensors using yttrium stabilized zirconia (YSZ) and NASICON (Na+ fast ion conductor) were used for sensing with high sensitivity [7,8]. However, the operating temperature is higher. Sensors based on Fabry-Perot interferometer (FPI) have been well explored for various VOC sensing, such as dual-cavity Fabry-Perot interferometer (DFPI) sensors of graphene/PMMA composites [9] and PVA-coated FPI sensors [10] have been used for acetone sensing.
Five methods have been proposed in developing on-chip optical gas sensors, namely, refractive index sensors, Raman spectroscopy-based sensors, infrared absorption spectroscopy-based sensors, cavity ring-down spectroscopy-based sensors, and photothermal spectroscopy-based sensors [11]. For example, a mid-infrared gas sensor uses a silicon groove waveguide on silicon nitride to detect ammonia (NH3) molecules, and the limit of detection (LOD) of the designed sensor can be as low as 5 ppm [12]; a ZnO thin film is used in the refractive index gas sensor, and the ethanol vapor was detected with a LOD of less than 100 ppm [13]. Raman sensors based on silicon nitride (Si3N4) waveguides can be used for the detection of acetone from a broadband vapor-phase volatile organic compound (VOC) but is not conducive to detection in a clinical environment [14].
The integrated photonics sensor based on Mach-Zehnder interferometer (MZI) is a good solution to solve the above problems. The Mach-Zehnder interferometer is very sensitive to the small refractive index changes in the environment, making the sensor highly sensitive. It has been applied to pressure [15], gas [16], DNA [17], RNA [18], protein [19] and biomacromolecule [20,21] detection. Usually, a suitable gas-sensitive film must be chosen to achieve selective gas sensing [22]. The materials and manufacturing of the sensor must be compatible with the complementary metal-oxide-semiconductor (CMOS) process for mass production and manufacturing at a lower cost. In this work, a 3 cm long MZI acetone sensor was manufactured using a standard CMOS process. The sensor arm was coated with a composite film of polyethyleneimine (PEI) and amido-graphene oxide (GO-NH2) as the sensor's gas-sensitive film and adsorption layer. Compared with the pure PEI film, the sensitivity of the composite film was enhanced by 3.2 times, and LOD can be as low as 0.76 ppm. Experimental results show that the sensor performance can meet the requirements of clinical diabetes diagnosis and detection of the severity of heart failure. In addition, based on the standard CMOS manufacturing process, the sensor can be mass-produced at a low cost.
2. Design and principles
The sensor is constructed on a silicon wafer with a size of 1 cm × 3 cm; a layer of 2 µm thick SiO2 is used as a buffer layer, the core layer of the waveguide is Si3N4 with 250 nm thick, and the upper cladding layer is SiO2 with 1 µm thick. A 50 µm wide and 18000 µm long area is created on the MZI sensing arm as the sensing area, which is filled with a composite film of polyethyleneimine (PEI) and amido-graphene oxide (GO-NH2). The layout of the sensor is shown in Fig. 1.
Fig. 1. The 3D structure diagram of the sensor and the cross-sectional view of the sensor arm and the reference arm. The material and size parameters are shown in the figure.
A laser beam is bifurcated into two pathways; one waveguided through the reference arm and the other through the sensing arm. A silicon dioxide coating covers the reference arm and is not exposed to the gas. Thereby the refractive index of the reference arm of the waveguide does not change with the gas environment. The sensor arm is covered with a gas-sensitive film. As the external gas environment changes, the gas-sensitive film on the sensor arm reacts with the gas, and the refractive index changes. The two beams of lasers are recombined in the Y-coupler through the reference and sensing arms, resulting in an interference signal. The output interference signal can be expressed as:
Here $\Delta {\varphi _0}$ is the initial phase difference between the two arms. This phase difference is caused by the difference in the refractive index of the two arms of the waveguide, including the imperfections introduced in the two arms due to the manufacturing process. $\varphi $ is the phase difference caused by changes in the gas environment; ${I_R}$ and ${I_S}$ are the light intensity of the reference arm and the sensing arm, respectively. The phase difference caused by the change in the gas environment is related to the effective refractive index difference between the two arms, and their relationship can be expressed as:
where L is the length of the sensing area, $\lambda $ is the wavelength of the incident light, $n_{eff}^s$ is the effective refractive index of the sensing arm, and $n_{eff}^r$ is the effective refractive index of the reference arm. For a highly-sensitive sensing application, a single-mode transmission must be maintained through the optical waveguide of the MZI along with a high surface sensitivity [23]. The effect of the waveguide design and the choice of the gas-sensitive film for achieving a highly-sensitive MZI sensor is critical and elucidated in the following sub-sections.2.1 Waveguide design
Due to the high refractive index, good transparency in the visible range, and low propagation loss, Si3N4 is selected as the core layer to fabricate the shallow rib waveguide of the MZI structure [24]. To obtain high sensitivity, the waveguide of the interferometer must be single-mode. Comsol Multiphysics was utilized to determine the geometric parameters of the waveguide and the single-mode limit condition based on the finite element method [25]. The operating wavelength was fixed at 633 nm, and a rib waveguide configuration was designed to support TM mode. The TM mode has more light field extending into the cladding layer than the TE mode. The functional relationship between the width W and the depth D of the waveguide is shown in Fig. 2(a) to excite the single-mode limitation.
Fig. 2. (a) The functional relationship between the width W and the depth D of the waveguide. (b) The relationship between the waveguide distance and the normalized power of the light output between the waveguides. (c) and (d) The mode field distribution of the waveguide rib width 2 µm, with the height of 10 nm and 15 nm, respectively, under TM mode.
The aperture angle of the Y-coupler in the MZI structure was also an essential part of the design. A larger aperture angle will cause more significant power loss. In comparison, a smaller aperture angle will reduce the distance between two parallel waveguides, and thus mode interference coupling will occur [26], leading to radiation losses. At the same time, if the aperture angle is designed too small, it is challenging to manufacture the waveguide. The length of the two arms of the Y-coupler was initially estimated, and the waveguide distance was obtained using the R-soft software. The mode coupling will occur between the waveguides when the waveguide spacing is less than 17 µm, as shown in Fig. 2(b). Considering the sensing area's width, we designed the Y-type coupler with an aperture angle of 2.52° and a waveguide spacing of 42 µm to ensure low propagation loss.
In MZI sensing applications, to obtain a highly sensitive surface, the thickness and refractive index of the waveguide core layer must be designed to allow a large part of the guided mode to pass through the external medium [27]. For Si3N4 waveguides, the refractive index contrast between the cladding and the core layers is high. We also simulated the mode field distribution of the cross-section of the waveguide with different rib heights under TM mode, as shown in Figs. 2(c) and 2(d). When the thickness of the core layer is hundreds of nanometers [23], higher surface sensitivity can be achieved, as shown in Fig. 2(c). 39% of the light extends into the cladding layer in this configuration.
2.2 Selection of gas-sensitive film
Another critical factor in realizing sensor function is the choice of gas-sensitive film. The gas-sensitive film used as an optical sensor must have good light transmittance, and secondly, the cladding layer used as a waveguide must have a suitable refractive index. The optical transmittance of the branched polyethyleneimine (PEI) in the visible light band is 99%. Secondly, the refractive index of PEI is 1.62 higher than that of SiO2. If PEI is used as a cladding layer to cover the sensor arm, it can extend more light field into the cladding, the sensitivity of the sensor will be improved [28]. PEI and graphene oxide (GO) were first used as sensing films for humidity detection due to their wealthy hydrophilic groups [29,30]. However, as the humidity level increases, the response of the GO film will be affected by a hysteresis response. Therefore, the GO/PEI layered film humidity sensor is proposed [31]. Subsequently, PEI was used as an acetone sensor because the amino functional group can undergo a reversible nucleophilic addition reaction with the acetone molecule's polar carbon-heteroatom double bond (C = O) [29]. Also, due to the large specific surface area of GO, it is proposed that the large specific surface area of GO and the characteristics of the gas-sensitive film can be combined to improve the sensing performance. The graphene oxide is modified with amino groups (-NH2) to undergo a nucleophilic addition reaction with acetone, thereby greatly enhancing its efficiency to detect acetone. During the adsorption/desorption reaction between acetone gas and PEI + GO-NH2 film, the carbon atoms of the polar carbon-heteroatom double bond (C = O) on the acetone molecule are partially positively charged, making the carbon atoms as electrophilic center. Then, the negatively charged nucleophilic-NH2 in PEI/GO-NH2 film would bond with the electrophilic carbon atoms to form imines and generate water molecules. The above process will change the effective refractive index of the sensing arm of the sensor covered with PEI + GO-NH2 film, and the change in the phase difference between the two arms. Since the compounds produced by the nucleophilic addition reaction are usually unstable, during the desorption process of the acetone molecule, the carbon atoms of the polar carbon-nitrogen double bonds (C = N) of the produced compounds are partially positively charged, and the generated water molecules bond with the electrophilic carbon atoms to form a reversible reaction. So the covered PEI + GO-NH2 film sensor has good responsiveness and repeatability. The improved hummer method of freeze-drying amido-graphene oxide is used as it has a higher proportion of amino groups and is evenly dispersed in water. The amino-graphene oxide films on the fabricated MZI waveguides were subsequently covered with PEI film as the active gas-sensitive materials for detecting acetone.
3. Experimental setting
3.1 Sensor fabrication steps
The fabrication method of on-chip MZI was proposed by F. Prieto et al. [21]. A layer of 2 µm thick SiO2 (n = 1.46) was grown by thermal oxidation on a silicon substrate, and a layer of 250 nm thick Si3N4 (n = 2.02) was deposited on SiO2 by low-pressure chemical vapor deposition (LPCVD). The Si3N4 rib structure with a height of 10 nm is etched by reactive ion etching (RIE). Then, a 1 µm thick SiO2 (n = 1.46) protective layer is deposited by plasma-enhanced chemical vapor deposition (PECVD). Finally, an area with a width of 50 µm and a length of 18000 µm was etched using RIE above the sensor arm on the SiO2 protective layer. Figure 3 shows the SEM image and AFM image of the sensing arm in the sensing area.
Fig. 3. The image of the sensing arm. (a) The SEM image of one side of the sensing area and the position distribution of the sensing arm in the sensing area. (b) The AFM image of 10 nm high shallow rib waveguide and the interface between the waveguide and the sensing area.
The gas-sensitive region of the sensor was prepared separately using 2% (w/v) branched polyethyleneimine aqueous solution and 0.5 mg/ml aqueous dispersion of amido-graphene oxide. A uniform aqueous solution was obtained by diluting 50% (w/v) branched polyethyleneimine (Sigma-Aldrich, America) in deionized water and sonicating for 60 minutes [32]. The PEI solution was spin-coated on the manufactured MZI with a homogenizer at a low speed of 500 rpm for 5s and a high speed of 3000 rpm for 60s, and dried in a drying oven at 80°C for 48 hours. Improved Hummer method freeze-drying amido-graphene oxide powder (Kaina Carbon New Materials, China) added to water at a 0.5 mg/ml ratio and ultrasonicated for 60 minutes to obtain a uniform aqueous dispersion. For experimental testing, a uniform water dispersion was drip-coated on the dried PEI film and dried in a drying oven at 80°C for 48 hours. Figure 4 shows the SEM image of the surface of the pure PEI film sensor and the surface of the PEI + GO-NH2 composite film sensor.
Fig. 4. (a) The SEM image of the surface of the pure PEI film sensor. (b) The SEM image of the surface of the PEI + GO-NH2 composite film sensor and locally enlarged image. There are many folds on the surface of GO-NH2, which increases the specific area, provides more adsorption holes, and improves the adsorption efficiency.
3.2 Characterization of sensor
The performance of the sensor was characterized as shown in Fig. 5. The sensor chip was placed in a 4 ml micro-glass air chamber with Peltier element-based temperature stabilizers at the bottom to attain a precise control within 1mk temperature. The micro-glass air chamber was then connected to the mass flow controller to control the flow rate precisely. The 633 nm HeNe laser uses a polarization controller to incident TM polarized light [33] onto a 20 × objective lens. The light was directly coupled to the sensor input through the objective lens. The choice of a low magnification objective can reduce the influence of thermal noise and mechanical noise on input coupling when temperature changes. The sensor's output was connected to a 40 × objective lens, and the 40 × objective lenses were connected to a photodetector. The photodetector was connected to a computer to monitor the light output intensity in real-time. A certain concentration of acetone gas and nitrogen mixed gas was prepared by the static gas distribution method. The acetone concentration can be controlled as long as the mixed gas and nitrogen are mixed in proportion. The mixed gas and nitrogen were connected to the mass flow controller to maintain the total flow rate at 20 ml/min. It should be noted that since both PEI and GO-NH2 contain hydrophilic amine groups, the influence of humidity on the results must be eliminated, and anhydrous copper sulfate must be used for drying before the gas is introduced.
Fig. 5. The experimental configuration. The 1 cm × 3 cm sensor is placed in a micro-glass gas chamber, and the end surface of the sensor has been polished.
4. Results and discussion
The most intuitive manifestation of the change in acetone gas concentration is the output intensity of the photodetector. As shown in Eq. (1), the phase difference $\Delta \varphi $ between the two arms is one $\pi $ cycle of the period, and the output intensity will vary from minimum to maximum (or maximum to minimum). The initial phase difference between the two arms is caused by the difference in the refractive index of the clad layer over the two arms and any defect introduced during the manufacturing process. The MZI sensor has a high sensitivity to temperature (about 6.72 rad/$\mathrm{\circ{C}}$). At a stabilized temperature, one can optimally determine the minimum and maximum output intensity. The difference between the maximum output intensity phase difference $\Delta {\varphi _1}$ and the minimum output intensity phase difference $\Delta {\varphi _2}$ is $\pi $ approximately. The output intensity of the reference arm and the sensing arm can be expressed by the maximum output intensity and the minimum output intensity [16]:
According to Eq. (1), one can obtain:
The maximum output intensity and the minimum output intensity can be measured and $\Delta {\varphi _0}$ can be determined by adjusting the temperature. Therefore, the phase difference $\Delta \varphi (t )$ between the two arms can be calculated by measuring the output intensity.
4.1 Sensor performance
The output intensity of the sensor is adjusted to the lowest by controlling the temperature stabilizer. The acetone gas with a concentration of 400 ppm is prepared by the static gas distribution method. The nitrogen and mixed gas with a concentration of 400 ppm acetone are alternately introduced into the air chamber every five minutes, and the flow rate is maintained at 20 ml/min. The photodetector detects noticeable and repeatable output intensity changes, as shown in Fig. 6. The nitrogen and 400 ppm acetone are alternately introduced, as shown in Fig. 6(a). As a result, the output intensity of the sensor changes over time, as shown in Fig. 6(b).
Fig. 6. (a) Varying input acetone concentration. (b) The resulting output intensity change using the MZI sensor.
Figure 6(b) shows that in the 50s after the introduction of acetone gas, the output intensity of the photodetector changes significantly. The output power of the photodetector drifts to different amplitudes at every instance the acetone gas is introduced. However, the overall change in the trend is identical, proving the stability of the sensor due to its repeatability. The sensor's response time can be considered by measuring the photodetector output power from the moment the acetone gas is introduced to the moment the output power response decreases significantly. From Fig. 6(b), it can be concluded that the estimated response time is about 50s. Since the air chamber has a certain volume, the response time should be slightly less than 50s. In the duration of 21 min to 24 min, the output intensity drops from the peak value to the level without the test object. This processing time is about 120s, and the response recovery time of the sensor can be estimated to be 120s. The small peaks in Fig. 6(b) can be explained as the fluctuation of the sensor chip pressure due to the instantaneous flow velocity of the gas flow, which causes the output intensity signal to be disturbed.
We set the acetone concentration gradient to show the variation range of acetone gas concentration between the maximum and minimum output intensity. We set a denser acetone concentration gradient near the maximum output intensity to determine this range more accurately. After acetone gas is introduced, we choose the output intensity from 120s to 300s to calculate the average value and use it as an indicator to detect the change in acetone gas concentration. In this process, the sensor covered with a pure PEI film and the sensor covered with a composite film of PEI + GO-NH2 are tested, as shown in Fig. 7(a). The pure PEI film was made of a 2% (w/v) PEI aqueous solution. The phase difference of the two films under different acetone gas concentrations is calculated by Eq. (5), as shown in Fig. 7(b).
Fig. 7. (a) The output power of 2% (w/v) PEI sensor and 2% (w/v) PEI + GO-NH2 sensor under different acetone gas concentrations. (b) The phase difference of the two films under different acetone gas concentrations.
Figure 7(b) shows that the PEI + GO-NH2 composite film at the same concentration has a higher response intensity than the pure PEI film. It proves that the composite film improves the performance of the sensor. The output power of the composite film reaches the maximum when the acetone concentration is 860 ppm. The change of the phase difference of the two films can be linearly fitted with the change of the acetone gas concentration. The sensitivity of the sensor can be defined as:
where c is the concentration of acetone gas, $\Delta \varphi $ is the change in phase difference at the concentration of c ppm acetone. It is estimated that the sensitivity of the pure PEI film sensor is 1.67×10−3 rad/ppm and the sensitivity of the PEI + GO-NH2 composite film sensor is 3.65×10−3 rad/ppm. The addition of GO-NH2 increases the sensitivity of the sensor by 1.2 times.
4.2 Improvement of sensor sensitivity by PEI content
In the clinical diagnosis of diabetes, the detection limit of EBA is required to be lower than a ppm level. According to the estimation of the sensor detection limit by Hans et al. [34]:
where $\Delta {\varphi _{pn}}$ is the phase noise of the system, and S is the sensitivity of the sensor. At a concentration of 100 ppm acetone, the output intensities were measured ten times and averaged at stable intervals (95% confidence interval) [18], and the phase noise of the system $\Delta {\varphi _{pn}}$ can be estimated to be 2.21×10−3 rad. According to Eq. (7), it can be estimated that the detection limit of 2% (w/v) PEI + GO-NH2 composite film is 1.82 ppm.The sensor's sensitivity was further improved by changing the content of PEI in the gas-sensitive film. The output intensity of the sensor is modulated to the lowest level by the temperature stabilizer, and the composite film sensors with PEI content of 1% (w/v), 2% (w/v), 5% (w/v), 8% (w/v), and 10% (w/v) are tested. The phase difference change is calculated in the 100 ppm acetone gas environment, as shown in Fig. 8. It shows that the 8% (w/v) PEI content has the best sensitivity enhancement performance. However, the increase of the PEI content also increases the response and recovery time of the sensor (Table 1). In comparison, the sensitivity of the composite film with 5% (w/v) PEI content also has a bigger improvement and has a faster response time and recovery time.
Fig. 8. In the 100 ppm acetone gas environment, the phase difference change of different PEI content sensors.
Table 1. The response time and recovery time of different PEI content sensors.
For clinical detection, the performance of the sensor needs to be tested in a low-concentration acetone environment. The initial phase difference can be modulated to $\pi $/2 by controlling the temperature stabilizer, yielding the highest sensitivity at the given time. In the concentration range of 0-20 ppm, the sensor output intensity for the 5% (w/v) pure PEI film and 5% (w/v) PEI + GO-NH2 composite film were respectively tested. We take the average value of 120s to 300s after each acetone gas is introduced as the output intensity. The result is shown in Fig. 9(a). The phase difference in the 0-20 ppm acetone environment is linearly fitted, as shown in Fig. 9(b). The result of the linear fitting is a good estimate of sensor performance. It can be estimated that the sensitivity of 5% (w/v) pure PEI film is 2.05×10−3 rad/ppm, and the sensitivity of 5% (w/v) PEI + GO-NH2 composite film is 8.72×10−3 rad/ppm. The addition of GO-NH2 increases the sensitivity of the sensor by 3.2 times. It is estimated that the sensor detection limit of 5% (w/v) PEI + GO-NH2 composite film is 0.76 ppm. In response to clinical diabetes diagnosis and detection of the severity of heart failure, the sensor's performance has met the requirements, which proves the potential of the sensor in clinical application.
Fig. 9. In the 0-20 ppm acetone environment, (a) the sensor output power of the 5% (w/v) pure PEI film and 5% (w/v) PEI + GO-NH2 composite film, (b) the phase difference of the 5% (w/v) pure PEI film and 5% (w/v) PEI + GO-NH2 composite film.
PEI and GO are also used for humidity detection, so we tested the humidity sensing performance of the 5% (w/v) pure PEI film sensor and 5% (w/v) PEI + GO-NH2 composite film sensor. Different levels of humidity environment are produced by the ratio of saturated NaCl solution (75%RH) and dry air. The test results are shown in Fig. 10. The phase difference generated by the humidity level is shown in Fig. 10(b), so the sensitivity of the 5% (w/v) pure PEI film sensor to humidity is 3.62×10−3 rad/%RH, and the sensitivity of the 5% (w/v) PEI + GO-NH2 composite film sensor is 2.77×10−2 rad/%RH.
Fig. 10. (a) The output power of 5% (w/v) PEI sensor and 5% (w/v) PEI + GO-NH2 sensor under different relative humidity levels. (b) The phase difference of 5% (w/v) PEI sensor and 5% (w/v) PEI + GO-NH2 sensor under different relative humidity levels.
4.3 Cross-sensitivity and stability
In order to estimate the selectivity of the sensing platform for the detection of acetone gas in the presence of other human exhalations, the sensitivity of the sensor to relevant organic volatiles was tested. Considering the median of the measurement, ammonia (833 ppb), acetone (477 ppb), methanol (461 ppb), ethanol (112 ppb), and acetaldehyde (22 ppb) [35] are tested separately. Figure 11 shows a comparison chart of the output intensity changes of the above five gases at the concentration of 100 ppm. Comparatively, acetaldehyde has a higher response, which will cause the most significant interference in acetone sensing. It can be explained by the fact that acetaldehyde also contains C = O, which will undergo a nucleophilic addition reaction with the amine group (-NH2) of PEI and GO-NH2. Still, the molecular mass of acetaldehyde is smaller than that of acetone, so the sensor has a higher selectivity to acetone [31]. In addition, the concentration of acetaldehyde gas exhaled by healthy humans is less than one-tenth of the concentration of acetone, which will have a smaller impact in practical applications. The inset in Fig. 11 shows that the sensor tested 20 ppm acetone gas 20 times in 36 days, and the overall output power from the sensor was nearly uniform. The output intensity of the sensor stabilized after 36 days, which proves the good stability of the sensor.
Fig. 11. At the concentration of 100 ppm, the cross-sensitivity of the 5% (w/v) PEI + GO-NH2 composite film sensor to other organic volatiles in the exhaled gas of the human body. The inset is the stability of the 5% (w/v) PEI + GO-NH2 composite film sensor within 36 days under the same conditions (at the concentration of 20 ppm and room temperature 25°C).
In practical applications, the gas exhaled by the patient is a mixed gas containing nitrogen, oxygen, carbon dioxide, ammonia, acetone, ethanol, acetaldehyde, etc., so it is necessary to test the gas mixture. We mixed 20 ppm ammonia, acetone, ethanol and acetaldehyde with nitrogen (80%), oxygen (16%), and carbon dioxide (4%) respectively. The test results are shown in Fig. 12, which shows that the performance of the sensor in the mixed gas test is close to the single acetone gas test.
5. Conclusions
In summary, an on-chip acetone gas sensor based on the MZI waveguide structure was fabricated to detect an extremely low level of acetone. The key to achieving high sensor sensitivity is the single-mode transmission and high surface sensitivity of the waveguide. The structural parameters of the waveguide are determined by simulation. In the TM mode, 39% of the light enters the cladding. On the sensing arm of the MZI, we cover the composite film of PEI + GO-NH2, and the amine groups (-NH2) in PEI and GO-NH2 can undergo a reversible nucleophilic addition reaction with the polar carbon-heteroatom double bond (C = O) of the acetone molecule. This composite film provides selectivity and repeatability for the sensor. By changing the concentration of the aqueous solution of the PEI film, the sensor's sensitivity is increased to 8.72×10−3 rad/ppm, which is 3.2 times higher than that of the pure PEI film. The sensor has high sensitivity in an acetone environment below 20 ppm at room temperature. The detection limit can reach 0.76 ppm, the estimated response time is 44s, and the recovery time is 106s. The results prove that the sensor performance can meet the requirements of clinical diabetes diagnosis and detection of the severity of heart failure. Compared with other semiconductor and solid electrolyte sensors, our sensors have higher sensitivity and stability and can work at room temperature. In addition, based on the standard CMOS manufacturing process, the sensor can be mass-produced and low cost. Multiple MZI sensors can also be combined on one chip to form a compact and stable device.
Finally, we consider integrating the entire testing equipment to achieve clinical applications in the next step. The patient's exhaled gas is collected by a gas collection device, which can dry its exhaled gas and send it to the sensor for testing. The sensor maintains a constant temperature during the test. The complete test results can be obtained within five minutes. The integration of the equipment will provide a new practical solution for diagnosing diabetes and detecting the severity of heart failure.
Funding
National Key Research and Development Program of China (2019YFB2203400); 111 Project (B20030); National Natural Science Foundation of China (62005037); Chengdu University of Information Technology (KYTZ202180).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. G. Roglic, “WHO Global report on diabetes: A summary,” Int. J. Noncommunicable Dis. 1(1), 3 (2016). [CrossRef]
2. V. Ruzsányi and M. P. Kalapos, “Breath acetone as a potential marker in clinical practice,” J. Breath Res. 11(2), 024002 (2017). [CrossRef]
3. N. Teshima, J. Li, K. Toda, and P. K. Dasgupta, “Determination of acetone in breath,” Anal. Chim. Acta 535, 181–199 (2005). [CrossRef]
4. B. Marcondes, G. Fabiana, I. Gutz, G. L. Batista, P. Saldiva, S. Ayub-Ferreira, I. S. Victor, and S. Mangini, “Exhaled acetone as a new biomarker of heart failure severity,” J. Chest 142(2), 457–466 (2012). [CrossRef]
5. D. Chen, J. Xu, Z. Xie, and G. Shen, “Nanowires assembled SnO2 nanopolyhedrons with enhanced gas sensing properties,” ACS Appl. Mater. Interfaces 3(6), 2112–2117 (2011). [CrossRef]
6. X. Zhou, B. Wang, H. Sun, C. Wang, P. Sun, X. Li, X. Hu, and G. Lu, “Template-free synthesis of hierarchical ZnFe 2 O 4 yolk–shell microspheres for high-sensitivity acetone sensors,” Nanoscale 8(10), 5446–5453 (2016). [CrossRef]
7. M. Mori, Y. Sadaoka, K. Shinichi, K. Masahito, and T. Kojima, “Development of ethanol and toluene sensing devices with a planar-type structure based on YSZ and modified Pt electrodes,” Sens. Actuators, B 187, 509–513 (2013). [CrossRef]
8. H. Zhang, C. Yin, Y. Guan, X. Cheng, X. Liang, and G. Lu, “NASICON-based acetone sensor using three-dimensional three-phase boundary and Cr-based spinel oxide sensing electrode,” Solid State Ionics 262, 283–287 (2014). [CrossRef]
9. D. Pawar, R. Kanawade, A. Kumar, Ch. N. Rao, P. Cao, S. Gaware, D. Late, S. N. Kale, S. T. Navale, W. J. Liu, D. L. Zhu, Y. M. Lu, and R. K. Sinha, “High-performance dual cavity-interferometric volatile gas sensor utilizing Graphene/PMMA nanocomposite,” Sens. Actuators, B 312, 127921 (2020). [CrossRef]
10. A. Kumar, D. Pawar, D. J. Late, and R. Kanawade, “PVA-coated miniaturized flexible fiber optic sensor for acetone detection: a prospective study for non-invasive diabetes diagnosis,” J. Mater. Sci.: Mater. Electron. 33(5), 2509–2517 (2022). [CrossRef]
11. J. Wu, G. Yue, W. Chen, Z. Xing, J. Wang, W. R. Wong, Z. Cheng, S. Y. Set, G. S. Murugan, X. Wang, and T. Liu, “On-Chip Optical Gas Sensors Based on Group-IV Materials,” ACS Photonics 7(11), 2923–2940 (2020). [CrossRef]
12. B. Kumari, A. Barh, R. K. Varshney, and B.P. Pal, “Silicon-on-nitride slot waveguide: A promising platform as mid-IR trace gas sensor,” Sens. Actuators, B 236, 759–764 (2016). [CrossRef]
13. N. A. Yebo, P. Lommens, Z. Hens, and R. Baets, “An integrated optic ethanol vapor sensor based on a silicon-on-insulator microring resonator coated with a porous ZnO film,” Opt. Express 18(11), 11859–11866 (2010). [CrossRef]
14. H. Zhao, B. Baumgartner, A. Raza, A. Skirtach, B. Lendl, and R. Baets, “Multiplex volatile organic compound Raman sensing with nanophotonic slot waveguides functionalized with a mesoporous enrichment layer,” Opt. Lett. 45(2), 447–450 (2020). [CrossRef]
15. C. Wagner, J. Frankenberger, and P. Deimel, “Optical pressure sensor based on a Mach-Zehnder interferometer integrated with a lateral a-Si: H pin photodiode,” IEEE Photon. Technol. Lett. 5(10), 1257–1259 (1993). [CrossRef]
16. F. T. Dullo, Firehun S. Lindecrantz, J. Jágerská, J. Hansen, H. Engqvist, S. Magnus, A. Stian, and O. G. Hellesø, “Sensitive on-chip methane detection with a cryptophane-A cladded Mach-Zehnder interferometer,” Opt. Express 23(24), 31564–31573 (2015). [CrossRef]
17. S. M. Grist, S. A. Schmidt, J. Flueckiger, V. Donzella, W. Shi, S. Fard, T. Sahba, J. T. Kirk, D. M. Ratner, K. Cheung, and L. Chrostowski, “Silicon photonic micro-disk resonators for label-free biosensing,” Opt. Express 21(7), 7994–8006 (2013). [CrossRef]
18. Q. Liu, Y. Shin, J. S. Kee, K. Kyoung, S. R. M. Rafei, A. P. Perera, X. Tu, G.Q. Lo, E. Ricci, and M. Colombel, “Mach–Zehnder interferometer (MZI) point-of-care system for rapid multiplexed detection of microRNAs in human urine specimens,” Biosens. Bioelectron. 71, 365–372 (2015). [CrossRef]
19. Q. Liu, X. Tu, S. Kee, K. Kyoung, J. Shin, H. Yong, Y. Kyungsup, G. Q. Lo, and M. K. Park, “Highly sensitive Mach–Zehnder interferometer biosensor based on silicon nitride slot waveguide,” Sens. Actuators, B 188, 681–688 (2013). [CrossRef]
20. P. Ramirez-Priego, D. Martens, A. Elamin, P. Soetaert, W. Van Roy, R. Vos, B. Anton, B. Birgit, H. Ronny, and M. Singh, “Label-free and real-time detection of tuberculosis in human urine samples using a nanophotonic point-of-care platform,” ACS Sens. 3(10), 2079–2086 (2018). [CrossRef]
21. Z. Li, J. Zou, H. Zhu, B. T. Nguyen, S. Thanh, Y. Liu, Y. Patricia, R. Bailey, J. Zhou, H. Wang, and Z. Yang, “Biotoxoid photonic sensors with temperature insensitivity using a cascade of ring resonator and Mach–Zehnder interferometer,” ACS Sens. 5(8), 2448–2456 (2020). [CrossRef]
22. A. L. Siarkowski, L. F. Hernandez, N. Morimoto, and B. V. Borges, “Sensing based on Mach-Zehnder interferometer and hydrophobic thin films used on volatile organic compounds detection,” Opt. Eng. 51(5), 054401 (2012). [CrossRef]
23. F. Prieto, B. Sepúlveda, A. Calle, A. Llobera, C. Dominguez, A. Abad, and A. Montoya, “An integrated optical interferometric nanodevice based on silicon technology for biosensor applications,” Nanotechnology 14(8), 907–912 (2003). [CrossRef]
24. N. Daldosso, M. Melchiorri, F. Riboli, F. Sbrana, L. Pavesi, G. Pucker, C. Kompocholis, M. Crivellari, P. Bellutti, and A. Lui, “Fabrication and optical characterization of thin two-dimensional Si3N4 waveguides,” Mater. Sci. Semicond. Process. 7(4-6), 453–458 (2004). [CrossRef]
25. F. T. Dullo, J.-C. Tinguely, S. A. Solbø, and O. G. Hellesø, “Single-mode limit and bending losses for shallow rib Si3N4 waveguides,” IEEE Photonics J. 7(1), 1–11 (2015). [CrossRef]
26. W. Zheng, C. Wang, W. L. Yang, and C. B. Wu, “Effect of wall thickness on Broadband Design of a Ka-Band TE21-Mode coupler,” J. Electron. Sci. Technol. 19(4), 100137 (2021). [CrossRef]
27. K. Tiefenthaler and W. Lukosz, “Sensitivity of grating couplers as integrated-optical chemical sensors,” J. Opt. Soc. Am. B 6(2), 209–220 (1989). [CrossRef]
28. E. F. Schipper, A. M. Brugman, C. Domínguez, L. M. Lechuga, R. Kooyman, and J. Greve, “The realization of an integrated Mach-Zehnder waveguide immunosensor in silicon technology,” Sens. Actuators, B 40(2-3), 147–153 (1997). [CrossRef]
29. B. Chachulski, J. Gebicki, G. Jasinski, P. Jasinski, and A. Nowakowski, “Properties of a polyethyleneimine-based sensor for measuring medium and high relative humidity,” Meas. Sci. Technol. 17(1), 12–16 (2006). [CrossRef]
30. Y. Y. Chen, X. Guo, H. Hui, and Z. Wu, “Graphene oxide thin film coated quartz crystal microbalance for humidity detection,” Appl. Surf. Sci. 257(17), 7778–7782 (2011). [CrossRef]
31. Z. Yuan, H. Tai, Z. Ye, C. Liu, G. Xie, X. Du, and Y. Jiang, “Novel highly sensitive QCM humidity sensor with low hysteresis based on graphene oxide (GO)/poly (ethyleneimine) layered film,” Sens. Actuators, B 234, 145–154 (2016). [CrossRef]
32. Q. Zhao, Z. He, Y. Jiang, Z. Yuan, H. Wu, C. Su, and H. Tai, “Enhanced acetone-sensing properties of PEI thin film by GO-NH2 functional groups modification at room temperature,” Front. Mater. 5, 82 (2019). [CrossRef]
33. E. Ashalley, C. Ma, Y. Zhu, H. Xu, P. Yu, and Z. M. Wang, “Recent progress in chiral absorptive metamaterials,” J. Electron. Sci. Technol. 19(3), 100098 (2021). [CrossRef]
34. H. P. Loock and P. D. Wentzell, “Detection limits of chemical sensors: Applications and misapplications,” Sens. Actuators, B 173, 157–163 (2012). [CrossRef]
35. D. Smith, C. Turner, and P. Španěl, “Volatile metabolites in the exhaled breath of healthy volunteers: their levels and distributions,” J. Breath Res. 1(1), 014004 (2007). [CrossRef]
