Mesoporous WO3 Nanofibers with ... - ACS Publications

May 11, 2016 - medical diagnosis.5 Exhaled breath analysis for diagnosis of .... hydrogen (H2), hydrogen sulfide (H2S), toluene (C6H5CH3), carbon mono...
0 downloads 0 Views 6MB Size
Mesoporous WO3 Nanofibers with ProteinTemplated Nanoscale Catalysts for Detection of Trace Biomarkers in Exhaled Breath Sang-Joon Kim,† Seon-Jin Choi,†,‡ Ji-Soo Jang,† Nam-Hoon Kim,† Meggie Hakim,§ Harry L. Tuller,⊥ and Il-Doo Kim*,† †

Department of Materials Science and Engineering and ‡Applied Science Research Institute, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 305-701, Republic of Korea § Platforms Engineering Group, Intel GmbH, Munich 85622, Germany ⊥ Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States S Supporting Information *

ABSTRACT: Highly selective detection, rapid response ( 50) against specific target gases, particularly at the 1 ppm level, still remain considerable challenges in gas sensor applications. We propose a rational design and facile synthesis concept for achieving exceptionally sensitive and selective detection of trace target biomarkers in exhaled human breath using a protein nanocage templating route for sensitizing electrospun nanofibers (NFs). The mesoporous WO3 NFs, functionalized with well-dispersed nanoscale Pt, Pd, and Rh catalytic nanoparticles (NPs), exhibit excellent sensing performance, even at parts per billion level concentrations of gases in a humid atmosphere. Functionalized WO3 NFs with nanoscale catalysts are demonstrated to show great promise for the reliable diagnosis of diseases. KEYWORDS: exhaled breath sensor, diagnosis of diseases, WO3 nanofiber, bioinspired catalyst, electrospinning

C

simultaneous control over many key features, such as grain size, grain interconnectivity, pore size, catalyst size, and catalyst dispersivity within the sensing layer.13,14,21,22 Enhancing mesoscale (2−50 nm) porosity plays a critical role in obtaining high sensitivities in SMOs due to the facile penetration of gas molecules dominated by Knudsen diffusion.13,22 In addition, the grain size of the SMOs is a key factor in significantly affecting the performance of SMO-based sensors. Indeed, the sensitivity is dramatically promoted when grain size is smaller than twice the depletion layer thickness.13,14,21,22 Often, there is a trade-off between average grain size and porosity.22 In the case of catalytic functionalization, the uniform distribution of catalytic NPs, tightly immobilized on the SMO surface, is one of the greatest challenges. The ready agglomeration between catalytic NPs is well-known to lead to serious degradation in the sensing properties.19 To accomplish the monodispersed catalytic NPs with nanoscale size, encapsulating routes of the outer surface of catalytic NPs by using polar templating

hemiresistive sensors, based on semiconducting metal oxides (SMOs), continue to receive much attention given their extensive applications in the monitoring of pollution,1 toxic2 and flammable3 gases, food processing,4 and medical diagnosis.5 Exhaled breath analysis for diagnosis of diseases has attracted recent attention given its ability to operate in a noninvasive and rapid mode without the need for highly trained personnel.6 In this regard, the SMO sensor’s facile and rational design, low cost, and ability to detect a large number of gases is particularly noteworthy.7−9 Because the human breath is fully humidified and contains thousands of potentially interfering gas species,10−12 it becomes highly challenging to adopt SMO-based sensing structures exhibiting both high sensitivity and selectivity in the detection of trace sub-ppm level disease-linked (biomarker) gases. Given that the gas-sensing mechanism is strongly associated with surface reactions between target gases and surface-chemisorbed oxygen species on the sensing layers,13,14 the optimization of nanostructured SMO-sensing layers15−19 combined with effective catalytic functionalization20 would appear to be a logical approach toward improving sensing properties. Conventional synthesis techniques, however, do not allow for © 2016 American Chemical Society

Received: February 16, 2016 Accepted: May 11, 2016 Published: May 11, 2016 5891

DOI: 10.1021/acsnano.6b01196 ACS Nano 2016, 10, 5891−5899

Article

www.acsnano.org

Article

ACS Nano

Figure 1. (a) Schematic illustration of synthesis steps for achieving apoferritin-encapsulated catalytic NPs; TEM images of as-prepared apoferritin-encapsulated catalytic NPs (AF-NPs) for (b) AF-Pt NPs, (c) AF-Pd NPs, and (d) AF-Rh NPs. High-resolution TEM images of respective NPs in the insets of (b−d). Particle size distribution histograms of (e) AF-Pt, (f) AF-Pd, and (g) AF-Rh NPs.

NPs show very narrow size distributions of 1−3 nm (Figure 1e−g), with mean diameters for Pt-AF, Pd-AF, and Rh-AF NPs of 2.0 ± 0.58, 1.8 ± 0.71, and 2.0 ± 0.52 nm, respectively, as derived statistically by analysis of sample sizes of 50 NPs. In contrast, Pt (Figure S3a), Pd (Figure S3b), and Rh (Figure S3c) NPs, synthesized by the conventional polyol process (these NPs are referred to as Pt-polyol, Pd-polyol, and Rhpolyol, respectively) and serve as more typical reference NPs, showed large average size distributions of 6.6 ± 1.2, 7.7 ± 2.3, and 4.1 ± 0.4 nm, respectively. Furthermore, Pt-polyol, Pdpolyol, and Rh-polyol NPs were partially aggregated, mainly due to the strong attractive interactions between NPs experienced during the high-temperature synthesis (over 110 °C) process.33−35 In the case of AF-NPs, the protein shell prevents bulk aggregation between the catalyst NPs due to repulsion between positively charged surfaces, resulting in homogeneous monodispersion in water. The Pt-, Pd-, and RhAF NPs exhibited crystalline character (the inset of Figure 1b− d) even after reduction by NaBH4 at room temperature. Figure 2a shows a schematic illustration of the electrospinning process for producing mesoporous WO3 NFs functionalized with AF-NPs. The as-synthesized AF-NPs (i.e., Pt, Pd, and Rh) were dispersed in an electrospinning solution containing W precursor/poly(vinylpyrrolidone) (PVP) in water. The AF-NPs could be uniformly embedded in the interior and exterior of as-spun W precursor/PVP composite

components have been suggested for producing hollow cage structure.23−28 In this work, we employed a polar protein nanocage composed of hollow apoferritin, with an outer diameter of 12 nm and an inner cavity diameter of 8 nm, as a sacrificial encapsulator for embedding catalytic nanoparticles (NPs) in the SMO-sensing layer.29−32 These polar templates inhibit bulk agglomeration of catalytic NPs, resulting in an increase of dispersity in SMOs. Most importantly, mesopores are generated within the SMOs as a result of the decomposition of the apoferritin templates during calcination. Mesoporous WO3 nanofibers (NFs) functionalized with well-distributed nanoscale Pt, Pd, and Rh catalytic NPs were prepared by electrospinning and subsequent high-temperature calcination.

RESULTS AND DISCUSSION A schematic illustration of the synthesis of apoferritinencapsulated catalytic NPs (hereafter, AF-NPs) is shown in Figure 1a. AF-NPs are prepared by infiltration of metal ions within the hollow cavity of apoferritin, followed by chemical reduction. More specifically, apoferritin-encapsulated Pt, Pd, and Rh (hereafter, Pt-AF, Pd-AF, and Rh-AF, respectively) NPs were synthesized as catalysts for sensitization of WO3 NFs, with transmission electron microscopy (TEM) images of these respective as-synthesized NPs shown in Figure 1b−d. The AF5892

DOI: 10.1021/acsnano.6b01196 ACS Nano 2016, 10, 5891−5899

Article

ACS Nano

Figure 2. (a) Schematic illustration of the electrospinning fabrication process for WO3 NFs functionalized by bioinspired catalytic NPs. (b) SEM image of calcined WO3 NFs functionalized by Pt-AF NPs (Pt-AF_WO3 NFs). STEM images of (c) Pt-AF_WO3 NFs with energydispersive X-ray spectroscopy element mapping of W (red), O (green), and Pt (yellow). (d) Cross-sectional Pt-AF_WO3 NFs. (e) Highresolution TEM image of cross-sectional Pt-Af_WO3 NFs (blue dotted box in (d)). (f) X-ray diffraction patterns of pristine WO3 NFs, Ptpolyol-loaded WO3 NFs, and Pt-AF_WO3 NFs.

mesopores (23.3 ± 14.5 nm) were created inside the WO3 NFs. The external open pore between the relatively large WO3 particles can be observed (red arrows in Figure 2d), which is good agreement with open pores (red circles in Figure 2c). The formation of mesopores originates from the thermal decomposition of the 8 nm apoferritin shell during calcination given that the apoferritin shells, composed of 24 organic subunits, are easily decomposed over 500 °C.36 During WO3 crystallite growth, tiny pores also migrate and become agglomerated, leading to the buildup of mesopores. This mesoporous structure can be confirmed by Brunauer−Emmett−Teller (BET) and pore distribution analyses. The surface area and corresponding pore distribution of the synthesized Pt-AF_WO3 and pristine_WO3 NFs were further determined by nitrogen adsorption−desorption. The results revealed that an increased BET surface area (13.042 m2/g) was presented with PtAF_WO3 compared with the surface area (6.155 m2/g) of the pristine_WO3 NFs (Figure S5). For Pt-AF_WO3 NFs, mesopores, which originated from decomposition of the apoferritin shell by calcination, range from 5 to 30 nm and are mainly distributed in larger quantity (Figure S6). These mesopores are essential for both increasing the surface area and facilitating gas diffusion (red arrows in Figure 2d).22,37 Grain size also has a considerable influence on gas-sensing performance. In general, the gas response increases in magnitude with decreasing grain size due to the enhanced

NFs without aggregation due to its polar nature. Figure 2b presents the morphologies and microstructural evolution of the calcined WO3 NFs functionalized with Pt-AF NPs (PtAF_WO3 NFs), with a weight fraction of 0.022 wt % with respect to WO3. During calcination at 600 °C for 1 h, the calcined NFs exhibited an approximately 50% shrinkage in diameter (400 nm) compared to the as-spun composite NFs with an average diameter of 765 ± 70 nm (see Figure S4). The Pt-AF_WO 3 NFs are randomly interconnected, resulting in macropores between the individual WO3 NFs (gray arrows in Figure 2b). This feature insures that the gas molecules can rapidly and readily diffuse throughout these macropores. In addition, scanning transmission electron microscope (STEM) examination in Figure 2c revealed that the calcined Pt-AF_WO3 NFs had many open pores, indicated by red circles (Figure 2c). Therefore, the gas molecules can easily penetrate into the interiors of WO3 NFs through these external openings. The distribution of catalytic Pt-AF was confirmed by the elemental mapping using energy-dispersive Xray spectroscopy (EDS), which revealed the well-distributed catalytic functionalization without aggregation in the WO3 NF. In order to clearly investigate the inner microstructure of individual Pt-AF_WO3 NFs, we obtained a high-resolution cross-sectional image of the Pt-AF_WO3 NF by cutting a single NF using a focused ion beam. As observed in the crosssectional STEM image (Figure 2d), a great number of 5893

DOI: 10.1021/acsnano.6b01196 ACS Nano 2016, 10, 5891−5899

Article

ACS Nano

Figure 3. Dynamic gas-sensing characteristics of pristine_WO3 NFs, WO3 NFs with polyol-NPs, and WO3 NFs with AF-NPs measured at 350 °C: (a) response toward acetone gas in concentration ranges of 1−5 ppm; (b) selective property of the Pt-AF_WO3 NFs toward 1 ppm acetone as well as the response toward nine interfering gases: hydrogen (H2), hydrogen sulfide (H2S), toluene (C6H5CH3), carbon monoxide (CO), and ethanol (C2H5OH), ammonia (NH3), pentane (CH3(CH2)3CH3), methane (CH4), and methyl mercaptan (CH3SH); (c) response as in (a) but to toluene gas in concentration ranges of 1−5 ppm; (d) selective property of the Pd-AF_WO3 NFs toward 1 ppm toluene as well as the response toward nine interfering gases; (e) response as in (a) but to hydrogen sulfide gas in concentration ranges of 1−5 ppm; (f) selective property of the Rh-AF_WO3 NFs to 1 ppm hydrogen sulfide as well as the response toward nine interfering gases.

modulation of the electron depletion layer, that is, the change in potential barrier height between grain boundaries upon exposure to target gases.21,22 Very interestingly, the PtAF_WO3 NFs exhibited a gradient in grain size distribution from larger in the exterior (56.1 ± 15.4 nm) to smaller in the interior (19.2 ± 7.9), as shown in Figure 2d. Since Pt-AF NPs (∼2 nm) immobilized in WO3 NFs interfered with the grain boundary migration, that is, crystallite growth of the WO321,38 during calcination, grain growth inhibition was achieved. More effective grain growth inhibition was observed in Pt-AF_WO3 NFs as compared to WO3 NFs functionalized by polyol-NPs (7 nm) given the higher numbers of AF-NPs with the smaller particle size (2 nm) for the same loading fraction. Pristine WO3 NFs and Pt-polyol-loaded WO3 NFs did not show the gradient microstructure in grain size distribution (Figure S7). We further investigated the average size of the polycrystalline WO3 particles comprising pristine WO3 NFs, WO3 Pt-AF_WO3

NFs, and Pt-polyol-loaded WO3 NFs by examining the peak broadening of the X-ray diffraction (XRD) peaks using the Scherrer formula, D = 0.9λ/β cos(2θ), where D is the average grain size, λ is the wavelength of the X-ray radiation (0.154 nm for Cu Kα), and β is the full width at half-maximum of the diffraction peak at 2θ (Figure 2f). The results revealed that the average sizes of the pristine WO3 NFs and reference Pt-polyolloaded WO3 NF particles calculated from the Scherrer formula based on the (002), (020), and (200) peaks were found to be 38.2 and 36.5 nm, respectively. In contrast, the mean grain size of the Pt-AF_WO3 NFs was 28 nm. This value is in good agreement with the direct observation by SEM (the inset of Figure 2f). Moreover, the uniform distribution of catalytic Pt NPs was observed in the cross-sectional high-resolution (HR) TEM image of Pt-AF_WO3 NF (Figure 2e). Similarly, WO3 NFs functionalized with Pd-AF NPs (PdAF_WO3 NFs) and Rh-AF NPs (Rh-AF_WO3 NFs) were 5894

DOI: 10.1021/acsnano.6b01196 ACS Nano 2016, 10, 5891−5899

Article

ACS Nano

Figure 4. Response of (a) Pt-AF_WO3 NFs toward acetone gas, (b) Pd-AF_WO3 NFs toward toluene, (c) Rh-AF_WO3 NFs toward hydrogen sulfide in the concentration range of 0.1−5 ppm at 350 °C, (d) linear approximation of the detection limit with Pt-AF_WO3, Pd-AF_WO3, and Rh-AF_WO3 NFs.

compared to Pt-polyol_WO3 NFs (Rair/Rgas = 2.6 at 1 ppm) and pristine_WO3 NFs (Rair/Rgas = 5.7 at 1 ppm), respectively. Figure 3b highlights the superior acetone (CH3COCH3) selectivity of the Pt-AF_WO3 NFs toward other interfering gases such as hydrogen (H2), hydrogen sulfide (H2S), toluene (C6H5CH3), carbon monoxide (CO), ethanol (C2H5OH), ammonia (NH3), pentane (CH3(CH2)3CH3), methane (CH4), and methyl mercaptan (CH3SH) at 1 ppm concentration. Highly selective sensing characteristics were clearly observed, confirming very high acetone response (Rair/Rgas = 62 at 1 ppm) while exhibiting weaker and/or negligible responses (Rair/Rgas< 15) to other gases. In the case of Pd-AF_WO3 NFs, they show the highest sensitivity and exceptional selectivity toward toluene gas. Similarly, the 0.06 wt % Pd-AF_WO3 NFs exhibited a toluene response (Rair/Rgas = 39 at 5 ppm) much higher than that of the WO3 NFs functionalized by 0.06 wt % Pd-polyol NPs (reference Pd-polyol_WO3 NFs) (Rair/Rgas = 5 at 5 ppm) (Figure 3c). Moreover, outstanding toluene selectivity was observed with minor cross-response toward other interfering gases (Figure 3d). In the case of 0.017 wt % Rh-AF_WO3 NFs, a noticeable improvement in hydrogen sulfide response (Rair/Rgas = 61 at 5 ppm) was observed as compared to that of the WO3 NFs functionalized by 0.017 wt % Rh-polyol NPs (reference Rh-polyol_WO3 NFs) (Rair/Rgas = 27 at 5 ppm) (Figure 3e). Interestingly, the Rh-AF_WO3 NFs led to high hydrogen sulfide selectivity in comparison to other interfering gases (Figure 3f). In this study, WO3 NFs functionalized by bioinspired catalysts clearly demonstrate significantly enhanced gas-sensing properties toward very low concentrations of acetone, toluene, and hydrogen sulfide in ambient conditions at 90% relative humidity when compared to previous reports (Tables S1−S3). At 350 °C, the Pt-AF-, Pd-AF-, and Rh-AF-loaded WO3 NF

synthesized with the same method used for the Pt-AF_WO3 NFs (Figure S8). The Pd- and Rh-AFs were found to be welldispersed in the WO3 NFs without noticeable aggregation (Figure S9). In addition, mesopores were generated within the granular WO3 NFs as a result of the decomposition of the apoferritin shell as well as effectively suppressing grain growth of WO3 crystallites during heat treatment (Figure S10). In order to investigate the catalytic effect of Pt-AF, Pd-AF, and Rh-AF NPs on the gas-sensing properties of the WO3-NFbased gas-sensing layers, we examined their response toward acetone (CH3COCH3), toluene (C6H5CH3), and hydrogen sulfide (H2S), known to be biomarkers for diabetes,39−43 lung cancer,44−46 and halitosis, respectively.47,48 The ambient air was maintained at a relative humidity of 90% to simulate human breath using a gas exposure test setup illustrated in Figure S2. In this study, optimized levels of Pt-AF (0.022 wt %), Pd-AF (0.06 wt %), and Rh-AF (0.017 wt %) were utilized (Figure S11) with a range of operating temperatures carefully investigated for all sensors (Figure S12). Figure 3a shows the acetone response of pristine WO3 NFs (pristine_WO3 NFs), WO3 NFs functionalized by 0.022 wt % Pt-polyol NPs (reference Pt-polyol_WO3NFs), and 0.022 wt % Pt-AF_WO3 NFs during cyclic exposures of acetone (1−5 ppm) with a 10 min on/off interval. The sensor response was defined as Rair/Rgas, where Rair and Rgas are the resistances of the sensor in air and target gas, respectively. The Pt-AF_WO3 NFs exhibited a markedly higher response upon exposure to acetone (Rair/Rgas = 153 at 5 ppm) compared to that of Pt-polyol_WO3 NFs (Rair/Rgas = 71 at 5 ppm) and pristine_WO3 NFs (Rair/Rgas = 18 at 5 ppm). At lower concentration of acetone (e.g., 1 ppm), the differences in gas responses were even more significant. Pt-AF_WO3 NFs (Rair/Rgas = 62 at 1 ppm) exhibited almost 24-fold and 11-fold higher acetone response 5895

DOI: 10.1021/acsnano.6b01196 ACS Nano 2016, 10, 5891−5899

Article

ACS Nano

Figure 5. (a) Response and recovery time of Pt-AF_WO3, Pd-AF_WO3, and Rh-AF_WO3 NFs in the gas concentration range of 0.1−5 ppm. (b) Average response and recovery time in 1−5 ppm ranges.

oxygen in ambient air. The Pd-AF and Rh-AF, which decorate the WO3 NFs, can be oxidized to p-type PdO and Rh2O3 during high-temperature calcination. By forming a p−n junction, an electron (surface) depletion layer is formed at the surface of WO3 NFs, thereby inducing a higher baseline resistance (Figure S15b,c). A thinning in the depletion layer is induced upon exposure of Pd-AF_WO3 and Rh-AF_WO3 to toluene (C6H5CH3) and hydrogen sulfide (H2S) by electrons being donated back to the conduction band, resulting in large resistivity changes. In this manner, the surface depletion depths are effectively modulated by facile electron exchange between the catalytic electronic sensitizers, that is, between PdO and Rh2O3, and the WO3 NFs. In addition, the reactions C6H5CH3 (gas) + 2O− → C6H5CH3O− + H2O + e− for toluene sensing6 and H2S + 3O− → SO2 + H2O + 3e− for hydrogen sulfide sensing7 can be further accelerated by the electronic sensitizers, which contribute to the improved selective properties toward toluene and hydrogen sulfide. AF-NPs, being very small (∼2 nm) and well-distributed within the NF-sensing layer, lead to a maximized catalytic effect distributed across the whole surface of the sensing layers, leading, in turn, to significant variations in sensing layer resistivity. In addition, mesopores and gradient grain size distribution within the WO3 microstructure facilitate gas diffusion and effectively activate surface reactions over an extended area. To investigate the utility of pattern recognition, sensor arrays were assembled utilizing Pt-AF_WO3 NFs, Pd-AF_WO3 NFs, and Rh-AF_WO3 NFs, and principal component analysis (PCA)49,50 was carried out to identify and discriminate diverse biomarkers (Figure 6). The PCA result revealed that all 10 different biomarker species with various concentrations (1−5 ppm) were clearly classified into 10 clusters without overlap, confirming the ability of the sensors to distinguish the 10 biomarkers. These results confirm the excellent capabilities of protein nanocage-templated sensor arrays in achieving accurate and rapid detection and identification of trace target biomarker gases.

sensors showed very stable response and recovery characteristics down to 100 ppb of acetone, toluene, and hydrogen sulfide, respectively, as demonstrated by the response transition with respect to time (Figure 4a−c). In addition, the theoretical detection limit37 of the three sensors (Pt-AF_WO3 NFs, PdAF_WO3 NFs, and Rh-AF_WO3 NFs) was calculated by an approximated linear plot as a function of acetone, toluene, and hydrogen sulfide concentration, considering the noise level of the base resistances (Figure S13). Based on the linear extrapolation detection limits of Pt-AF_WO3, Pd-AF_WO3, and Rh-AF_WO3 NFs, all sensors are estimated to be below 1 ppb (Figure 4d). The analysis of exhaled breath gases requires not only high sensitivity and selectivity in fully humid air but also rapid reaction and recovery times. Figure 5a shows the response and recovery times for Pt-AF_WO3 NF, Pd-AF_WO3 NF, and RhAF_WO3 NF sensors exposed to acetone, toluene, and hydrogen sulfide, respectively, in the gas concentration range of 0.1−5 ppm. It should be noted that the longer response times with decreasing gas concentrations for all the sensors were observed due to the diffusion-limited kinetics.32 Each sensor showed fast response/recovery characteristics when exposed to acetone (response