Innovative Nanosensor for Disease Diagnosis - Accounts of Chemical

May 8, 2017 - ... protein cage, particle size distributions of metal nanoparticles, schematic illustration and corresponding camera image of sensor ar...
0 downloads 0 Views 10MB Size
Article pubs.acs.org/accounts

Innovative Nanosensor for Disease Diagnosis Sang-Joon Kim,†,§ Seon-Jin Choi,†,‡,§ Ji-Soo Jang,† Hee-Jin Cho,† and Il-Doo Kim*,† †

Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Republic of Korea ‡ Applied Science Research Institute, Korea Advanced Institute of Science and Technology (KAIST), 291 Daehak-ro, Yuseong-gu, Daejeon 34141, Republic of Korea S Supporting Information *

CONSPECTUS: As a futuristic diagnosis platform, breath analysis is gaining much attention because it is a noninvasive, simple, and low cost diagnostic method. Very promising clinical applications have been demonstrated for diagnostic purposes by correlation analysis between exhaled breath components and specific diseases. In addition, diverse breath molecules, which serve as biomarkers for specific diseases, are precisely identified by statistical pattern recognition studies. To further improve the accuracy of breath analysis as a diagnostic tool, breath sampling, biomarker sensing, and data analysis should be optimized. In particular, development of high performance breath sensors, which can detect biomarkers at the ppb-level in exhaled breath, is one of the most critical challenges. Due to the presence of numerous interfering gas species in exhaled breath, selective detection of specific biomarkers is also important. This Account focuses on chemiresistive type breath sensors with exceptionally high sensitivity and selectivity that were developed by combining hollow protein templated nanocatalysts with electrospun metal oxide nanostructures. Nanostructures with high surface areas are advantageous in achieving high sensitivity because the sensing signal is dominated by the surface reaction between the sensing layers and the target biomarkers. Furthermore, macroscale pores between one-dimensional (1D) nanostructures can facilitate fast gas diffusion into the sensing layers. To further enhance the selectivity, catalytic functionalization of the 1D metal oxide nanostructure is essential. However, the majority of conventional techniques for catalytic functionalization have failed to achieve a high degree of dispersion of nanoscale catalysts due to aggregation on the surface of the metal oxide, which severely deteriorates the sensing properties by lowering catalytic activity. This issue has led to extensive studies on monolithically dispersed nanoscale particles on metal oxides to maximize the catalytic performances. As a pioneering technique, a bioinspired templating route using apoferritin, that is, a hollow protein cage, has been proposed to obtain nanoscale (∼2 nm) catalyst particles with high dispersity. Nanocatalysts encapsulated by a protein shell were first used in chemiresistive type breath sensors for catalyst functionalization on 1D metal oxide structures. We discuss the robustness and versatility of the apoferrtin templating route for creating highly dispersive catalytic NPs including single components (Au, Pt, Pd, Rh, Ag, Ru, Cu, and La) and bimetallic catalysts (PtY and PtCo), as well as the core−shell structure of Au−Pd (Au-core@Pdshell). The use of these catalysts is essential to establish high performance sensors arrays for the pattern recognition of biomarkers. In addition, novel multicomponent catalysts provide unprecedented sensitivity and selectivity. With this in mind, we discuss diverse synthetic routes for nanocatalysts using apoferritin and the formation of various catalyst−1D metal oxide composite nanostructures. Furthermore, we discuss detection capability of a simulated biomarker gas using the breath sensor arrays and principal component analysis. Finally, future prospects with the portable breath analysis platform are presented by demonstrating the potential feasibility of real-time and on-site breath analysis using chemiresistive sensors.



INTRODUCTION Early diagnosis of diseases and monitoring of physical conditions are very important to reduce mortality and medical costs. For this reason, the development of portable and inexpensive diagnostic platforms is receiving much attention to achieve real-time and on-site diagnoses.1 To obtain a reliable diagnostic platform, specific biomarkers, which are related to a certain physical condition, should be detected with high sensitivity and selectivity. Such biomarkers can be generally found in the blood at relatively high concentrations. However, blood analysis inevitably involves pain because it is an invasive diagnostic method. As alternative methods, there have been © 2017 American Chemical Society

intensive studies for the development of noninvasive diagnostic platforms by detecting biomarkers, which originate from the blood and metabolic products such as breath, sweat, and urine.2,3 Human breath contains specific biomarkers because abnormal metabolism changes blood chemistry and volatile organic compounds (VOCs) as byproducts can be exhaled via the lungs by alveolar exchange from the blood.4 For this reason, breath analysis is a very attractive diagnostic method Received: January 23, 2017 Published: May 8, 2017 1587

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596

Article

Accounts of Chemical Research

Figure 1. Catalytic nanoparticles synthesized by the apoferritin protein cage. Schematic illustrations of (a) apoferritin, (b) inorganic cation infiltration and subsequent reduction to form apoferritin-encapsulated nanoparticles (Apo-NPs), and (c) 1D semiconductor metal oxide (SMO) nanofibers (NFs) functionalized by NPs. (d) Single component NPs such as Au, Pt, Pd, and Rh. (e) Multicomponent NPs such as PtY, PtCo, and Au(core)−Pd(shell) NP (Au@Pd).

considering that the analysis is totally painless, which is agreeable with subjects. These critical advantages have triggered extensive studies on breath analysis, and diverse biomarker species related to diabetes, lung cancer, and halitosis have been identified up to now.5,6 Generally, very low concentrations of biomarkers are found in the breath ranging from parts per billion (ppb) to parts per million (ppm).7 To detect such trace amounts of biomarker species, highly sensitive and selective detection techniques are essential for diagnostic purposes. Typically, gas chromatography−mass spectrometry (GC-MS) analysis is widely used to identify specific biomarkers with relatively high precision.8 However, GC-MS analysis frequently requires a long analysis time due to the preconcentration step to increase the detection capability and needs a professional operator to operate the equipment.9 Since the first observation of the gas sensing property in the early 1960s, semiconductor metal oxides (SMOs) have been developed for flammable gas detectors.10 In the first stage, ntype SMOs such as SnO2 were generally used to investigate predominant factors that affect the gas sensing performance.11

There are two important functions in gas sensing processes, receptor and transducer functions.12 The receptor function explains the importance of the surface reactions dominated by the chemisorbed oxygen species on the surface of the SMO, which can be formed as follows:13 O2(gas) + V(ads) ⇌ O2(ads)

(1)

O2(ads) + e− ⇌ O2−(ads)

(2)

O2−(ads) + e− ⇌ O2 2 −(ads)

(3)

O2 2 −(ads) + V(ads) ⇌ 2O−(ads)

(4)

2O−(ads) + e− ⇌ O−(ads) + O2 −(ads)

(5)

O−(ads) + O2 −(ads) + e− ⇌ 2O2 −(ads)

(6)

where V(ads) indicates the vacant adsorption site on the surface of SMO. On the other hand, the transducer function is related 1588

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596

Article

Accounts of Chemical Research

Figure 2. Porous nanofibrous structures functionalized by Apo-NPs. (a) SEM, (b) STEM, (c) cross sectional TEM, and (d) high-resolution TEM images of mesoporous WO3 NFs functionalized with Apo-Pt NPs. (e) Apo-Au NPs self-assembled on polystyrene (PS) colloid templates. (f) SEM and (g) TEM, and (h) EDS analysis of macroporous WO3 NFs functionalized with Apo-Au NPs. (i) SEM, (j) high-resolution SEM, (k) STEM, and (l) EDS analysis of mesoporous WO3 NFs functionalized with Apo-PtY NPs.

possible until recent years.18 In this Account, we describe an effective synthetic route for single and binary component catalysts using hollow protein cages, that is, apoferritin, and their uniform immobilization onto electrospun SMO nanofibers (NFs) and nanotubes, which possess a large surface-to-volume ratio and broad-scale pores. We also present various catalyst loaded nanofibrous SMO sensors with exceptionally high sensitivity and selectivity, even for very low concentrations of breath biomarker species. Finally, a classification of breath components is presented using a portable sensing platform that can be interconnected with a smartphone to demonstrate realtime and on-site monitoring of physical conditions.

to the transduction of the sensing signal induced by the chemical reactions into an output signal such as resistance. For this reason, the microstructure of the SMO sensing layer is very important to efficiently transduce the signal to the output. In particular, once the polycrystalline particles of n-type SMOs are connected with their neck contacts, the gas sensitivity dominantly depends on the particle size. 12 From the fundamental understanding of basic gas sensing mechanisms and structural design principles, a large surface area and open porosity are the most important factors in terms of the receptor function to accommodate chemisorbed oxygen molecules and to enhance surface reactions with chemical molecules. For this reason, recent studies have mainly focused on the synthesis of diverse nano-building-blocks, which can contribute to highly sensitive gas sensors.14,15 In addition to the microstructural effect, the functionalization by foreign additives (or catalysts) on the SMO surface is essential to improve the gas sensitivity and selectivity. Previous studies have shown that the functionalization of noble metallic particles such as Pt and Pd generates additional chemisorbed oxygen molecules, thereby modulating receptor function.12 To further enhance the catalytic effect of noble metal particles, very fine control of the catalyst size and distribution is critical.16,17 Moreover, for accurate detection of various biomarker species in breath, a new class of catalysts can be rationally designed by engineering of bimetallic catalysts to overcome current limitations of single-component catalysts. Through the recent advances in the synthesis of nanostructured SMOs and the functionalization of novel catalysts, significant progress has been achieved for highly sensitive and selective detection of breath components, which had not been



CATALYTIC NANOPARTICLES ENCAPSULATED BY APOFERRITIN Ferritin is a family of iron storage proteins consisting of 24 protein subunits, which self-assemble into dimers to form a dodecameric cage.19 The overall size of ferritin is approximately 12−13 nm in diameter with an inner cavity of 7−8 nm. A hollow protein cage, apoferritin, can be obtained by removing the inorganic component in the core of the ferritin, as illustrated in Figure 1a. Apoferritin is known as a promising building block for the synthesis of nanomaterials due to its unique structure of hollow proteins that can host diverse inorganic materials.19 Inorganic cations can infiltrate into the core through a 3-fold axis where three peptide subunits intersect with each other by an electrostatic field gradient.20 After an adequate reduction of inorganic cations using sodium borohydride (NaBH4), apoferritin-encapsulated inorganic nanoparticles (hereafter, Apo-NPs) can be achieved with a diameter that is less than the hollow cavity size of 8 nm (Figure 1589

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596

Article

Accounts of Chemical Research

Figure 3. Porous nanotubular structures functionalized by Apo-NPs. (a) SEM and (b) EDS analysis of thin-walled SnO2 NTs functionalized with Apo-Pt NPs. (c) SEM and (d) (b) EDS analysis of porous SnO2 NTs functionalized with Apo-Pt NPs. (e) SEM and (f) EDS analysis of 0D hollow SnO2 nanosphere-1D SnO2 NTs functionalized with Apo-Pt NPs. (g) SEM and (h) EDS analysis of macroporous WO3 NTs functionalized with Apo-Pd NPs.

additionally dissolving inorganic precursor(s) in a polymeric electrospinning solution.22 Subsequently, high-temperature calcination in air atmosphere results in the formation of SMO NFs by the decomposition of the polymeric scaffold and the oxidization of an inorganic precursor(s). Moreover, catalytic functionalization of NPs can be easily achieved on SMO NFs by additionally dispersing catalyst NPs within the composite electrospinning solution (Figure 1c). Figure 2 shows catalystloaded SMO composite NFs, which were synthesized by electrospinning an inorganic precursor/polymer solution together with the Apo-NPs and a subsequent calcination process. Highly porous WO3 NFs functionalized with apoferritinencapsulated Pt NPs (hereafter, Apo-Pt WO3 NFs) were obtained by electrospinning a W precursor [(NH4)6H2W12O40· xH2O]/polyvinylpyrrolidone (PVP) composite solution with Apo-Pt NPs followed by calcination.23 The 1D Apo-Pt WO3 NFs exhibited a nonwoven network with macropores between the NFs (Figure 2a). In addition, several mesopores were observed in the interior of the NFs by scanning transmission electron microscopy (STEM) (red circles in Figure 2b). The highly porous structure of the Apo-Pt WO3 NF was also confirmed by cross-sectional TEM analysis (Figure 2c). The unique mesopores of 23.3 ± 14.5 nm along the NFs were attributed to the decomposition of the protein shells during the calcination process. These macro- and mesopores can facilitate rapid and ready gas penetration into the sensing layers, which can remarkably enhance gas sensitivity by inducing an active surface reaction. Importantly, the catalytic Pt NPs were welldispersed along the WO3 NFs confirmed by energy dispersive X-ray spectroscopy (EDS) elemental mapping (Figure 2b), and individual Pt NPs were uniformly immobilized on the WO3 NFs without any aggregation, as shown in the high-resolution TEM image (Figure 2d). To manipulate the pore size and pore density in 1D SMO NFs, a colloid templating route has been suggested by dispersing colloidal beads in an electrospinning solution.24,25

1b). As examples, Figure 1d,e shows diverse catalytic NPs synthesized with apoferritin protein cages. Single compositional noble metallic NPs such as Au, Pt, Pd, Rh, Ag, and Ru, as well as non-noble metallic NPs such as Cu and La, were successfully synthesized (Figure 1d and Figure S1 in the Supporting Information). Protein cages are attractive templates because alloy particles can also be synthesized by infiltrating two different inorganic precursors with a controlled atomic ratio followed by a subsequent reduction process. For example, in this work, binary alloy catalysts such as PtY and PtCo were prepared with apoferritin (Figure 1e). Moreover, a unique core−shell structure of Au(core)−Pd(shell) (i.e., Au@Pd) was achieved by sequential infiltration and reduction steps of two different metallic precursors. Here, Au cations were first infiltrated into the protein cages and subsequently reduced to form metallic Au. Then, Pd cations were next infiltrated and subsequently reduced resulting in the formation of Au@Pd NPs. Importantly, all the Apo-NPs exhibited very small and uniform diameters ranging from 2 to 3 nm (Supporting Information, Figure S2). Single- and multicomponent NPs surrounded by a protein shell are the most suitable catalyst materials for effective sensitization of SMO sensors, thereby establishing sensing libraries composed of catalyst-decorated SMO layers for breath pattern recognition. In addition, Apo-NPs with a high dispersive nature due to the repulsive force between the positively charged apoferritin shells can lead to a fully activated catalytic property by an effective reaction with chemical molecules on the surface of the SMOs.21



ELECTROSPINNING FOR CATALYST-LOADED 1D NANOSTRUCTURES Electrospinning is a fascinating technique that forms 1D polymeric NFs by injecting a polymeric solution from a syringe nozzle under high electric fields between the nozzle and a ground collector.22 Not only polymeric NFs but also diverse inorganic/organic composite NFs can be synthesized by 1590

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596

Article

Accounts of Chemical Research

the homogeneous and omnidirectional gas emission to the outer side of the 1D nanostructure, Pt NPs were uniformly distributed on the shell of the SnO2 NTs (Figure 3b). Porous NTs with a gas permeable wall are highly desirable because direct gas penetration through the pores on the shell of the NTs enables more effective gas accessibility to the inner layers. For this purpose, sacrificial colloid templating route combined with ramping-rate controlled electrospinning was used to form discrete macropores on the wall of NTs.27 PS colloids were additionally dispersed in the electrospinning solution comprised of Sn precursor and PVP including the Apo-Pt NPs. After electrospinning and the subsequent calcination step, macroscale pores (>50 nm) were created on the surface of the SnO2 NTs by the thermal decomposition of the PS colloids (Figure 3c). In addition, numerous mesoscale pores (2−50 nm) were observed due to the thermal decomposition of the apoferritin shells. As shown in Figure 3d, Pt NPs were uniformly distributed on the porous SnO2 NTs during the Ostwald ripening behavior. Hierarchical NTs can be achieved by controlling the ambient conditions of the calcination, which results in a 0D−1D compositional structure. Specifically, the as-spun Sn precursor/ PVP composite NFs containing Apo-Pt NPs were first calcined at 450 °C in a reducing ambient condition (N2, 99.999%) to form carbon nanofibers (CNFs). During the heat-treatment, liquid Sn metal was precipitated from the surface of CNFs due to the low melting temperature of the Sn metal (232 °C), thereby leading to the formation of 0D Sn metal spheres on the CNFs. After additional heat-treatment in air, 1D SnO2 NTs combined with 0D hollow SnO2 nanospheres were achieved, which exhibits hierarchical structures with enhanced surface areas (Figure 3e). Interestingly, during the first heat-treatment process in a reducing ambient condition, the chelation between the Apo-Pt NPs and Sn metal induced catalyst migration with the liquid Sn. This leads to the uniform decoration of the Pt NPs on the hierarchical sensing matrix (Figure 3f). In addition, the coaxial (or two-fluid) electrospinning technique has been widely used for producing hollow NTs. The basic principle is the simultaneous electrospinning of two immiscible solutions through a concentrically arranged dualnozzle. In general, a thermally decomposable template such as a mineral oil is adequate as a core electrospinning solution, whereas a mixed solution that includes a metallic precursor and polymeric matrix can be adopted as a shell electrospinning solution. After electrospinning followed by the subsequent calcination, various hollow SMO NTs can be synthesized by the oxidization of the metallic precursor in the outer shell and the thermal decomposition of the core oil template as well as the shell polymeric scaffold in the outer shell. To obtain macroporous SMO NTs, sacrificial PS colloids with diameter of 200 nm can be introduced in the shell electrospinning solution. As a result, macroscale pores with an average pore diameter of 172 nm were generated on WO3 NTs.29 In addition, macroporous WO3 NTs functionalized by Pd NPs were successfully synthesized via coaxial electrospinning, which uses core (mineral oil) and shell (Apo-Pd NPs as well as PS colloids dispersed in W precursor/PVP) solutions (Figure 3g). The high dispersibility of the Apo-Pd NPs resulted in the uniform distribution of the Pd NPs on the wall of the macroporous WO3 NTs (Figure 3h).

Polystyrene (PS) colloids with diameters ranging from 100 to 500 nm can generate macroscale pores with 40−60% reduced sizes of the original PS colloids in the WO3 NFs due to the shrinkage of the polymeric colloids during calcination. To selectively decorate catalytic NPs on the pore sites, selfassembly of apoferritin-encapsulated Au NPs (Apo-Au NPs) anchored on PS colloid templates was achieved by electrostatic interactions between the positively charged apoferritin shell and the negatively charged PS colloid templates (Figure 2e).21 After electrospinning the W precursor/PVP matrix, which included the Apo-Au NP self-assembled PS colloids, followed by calcination, the Au-loaded porous WO3 NFs exhibited randomly distributed spherical pores as well as spherical humps on the surface of the NFs (Figure 2f). Multiple macropores were clearly observed by TEM (Figure 2g). In addition, EDS elemental mapping analysis was conducted to identify the Au components, which were selectively functionalized at the pore sites by transferring the Au NPs from the PS colloid surface to the pore sites of the WO3 NFs during the calcination process (Figure 2h). Designing bimetallic catalysts is very important for the creation of diverse sensor libraries. In this regard, multicompositional PtY NPs were functionalized on the WO3 NFs with electrospinning. Apoferritin-encapsulated PtY NPs (ApoPtY NPs) were uniformly dispersed in the composite electrospinning solution consisting of W precursor and PVP. After electrospinning followed by subsequent calcination, nonwoven PtY-loaded WO3 NFs (hereafter, Apo-PtY WO3 NFs) were obtained (Figure 2i). Cross-sectional scanning electron microscopy (SEM) (Figure 2j) and STEM analyses (Figure 2k) revealed the mesoporous structure of the Apo-PtY WO3 NFs. Moreover, a uniform distribution of the Pt and Y was confirmed by EDS elemental mapping analysis (Figure 2l). Hollow nanotubes (NTs) with thin-walled structures offer more effective signal transduction due to enhanced surface reactions of target biomarker species on both the inner and outer surfaces compared with dense NFs. In addition, the catalytic sensitization can be activated by functionalization of the Apo-NPs on thin-walled SMO NTs. To prove these advantages, 1D NTs were demonstrated by controlling the processing parameters, such as the use of a dual-nozzle, incorporation of sacrificial templates, ramping rate control, and two-step heat-treatment during electrospinning and calcination (Figure 3).26−29 Representative examples of 1D hollow nanotubular structures include (i) dense thin-walled NTs, (ii) porous NTs, and (iii) hierarchical NTs, which possess large active sites for increased gas reaction. By combining Apo-NP and ramping rate-controlled electrospinning, Pt functionalized SnO2 NTs were successfully synthesized with wall thicknesses ranging from 30 to 40 nm (Figure 3a).26 The synthesis mechanism of the Pt functionalized SnO2 NTs can be predominantly explained by the Ostwald ripening effect. During the calcination of as-spun composite NFs, that is, Sn precursor [SnCl2·2H2O]/PVP NFs including Apo-Pt NPs, the outer Sn precursor was preferentially oxidized to SnO2 compared to the core-side of the Sn precursor in the composite NFs, thereby forming the larger SnO2 grains at the exterior side of the NFs. Subsequently, smaller SnO2 grains are merged together to become larger SnO2 grains by the Ostwald ripening effect, which then are progressively transformed into SnO2 NTs. Simultaneously, Pt NPs automatically migrated to the wall of the SnO2 NTs through the H2O and CO2 gases emitted outward during the thermal decomposition of the PVP. Due to 1591

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596

Article

Accounts of Chemical Research

Figure 4. Sensing characteristics of WO3 NF-based sensing layers functionalized with Apo-NPs for simulated biomarker species. (a) Selective sensing property, (b) pattern recognition of multiple biomarker species using PCA, (c) pattern recognition of mixture biomarker species using PCA, (d) limit of detection property, (e) response time characteristics for concentrations ranging from 1−5 ppm, and (f) dynamic response transition characteristics toward acetone for concentrations ranging from 1−5 ppm.



GAS DETECTION CAPABILITY FOR BREATH ANALYSIS To understand the suitability of breath biomarker detection using chemiresistive sensors, important sensing parameters including selectivity, sensitivity, limit of detection (LOD), and response time were evaluated with the 1D nanostructures functionalized with the Apo-NPs (Figure 4). The conditions during the sensing characterization were maintained consistently under a highly humid atmosphere (90% RH) with gas exposure at concentrations ranging from 1−5 ppm with an operating temperature of 350 °C. Schematic illustrations of the sensor system and the sensor components are provided in the Supporting Information (Figure S3). The sensitivity of the sensing materials was evaluated by normalization of the resistance changes defined as a response, that is, Rair/Rgas, where Rair and Rgas are the resistances of the sensors in stabilized ambient air and ambient target gas, respectively. It should be noted that it is very difficult to diagnose specific disease through the detection of a single VOC because a number of biomarker VOCs are interrelated for breath analysis. For example, Saalberg et al. comprehensively reviewed lung cancer biomarkers reported from 1985 to 2015, which revealed numerous VOCs including isoprene, ethylbenzene, styrene, and toluene.30 Recently, Nakhleh et al. investigated 1404 subjects and identified 13 VOCs including acetone and toluene associated with diseases, which can clearly discriminate 17 diseases such as lung cancer.31 To diagnose halitosis, volatile sulfuric compounds (VSCs) such as hydrogen sulfide, dimethyl sulfide, and methyl mercaptan should be detected as biomarkers. In this Account, we focused on the detection of hydrogen sulfide (H2S), acetone (CH3COCH3), and toluene (C6H5CH3) as potential biomarkers of certain diseases in exhaled breath. All the sensors were exposed to three main target gases, hydrogen sulfide, acetone, and toluene. In addition, the response properties toward hydrogen (H2), ethanol (C2H5OH), carbon

monoxide (CO), ammonia (NH3), methane (CH4), pentane (C5H12), and methyl mercaptan (CH3SH) were investigated as interfering analytes (Figure 4a). For the pristine WO3 NFs, a relatively low sensitivity and selectivity were observed with a response of 7.6 for H2S and 5.7 for acetone at 1 ppm. In contrast, a significantly improved sensitivity and selectivity were achieved by the Apo-Rh WO3 NFs with a high response of 19.8 for H2S at 1 ppm while maintaining a low response (Rair/Rgas = 5.6) to acetone. In addition, the Apo-Pd WO3 NFs exhibited enhanced toluene response up to 13.8 at 1 ppm with minor responses (Rair/Rgas < 3.5) for other analytes including H2S and acetone. In the case of the Apo-Pt WO3 NFs, a superior response for acetone of 59.4 at 1 ppm was achieved. However, a low selectivity with a noticeable H2S response of 15.9 at 1 ppm was also observed with the Apo-Pt WO3 NFs. To provide further improved sensitivity and selectivity, we explored multicompositional catalysts, which are tightly immobilized on the WO3 NFs. Indeed, superior sensing performance was achieved using the Apo-PtY WO3 NFs with a high response of 65 for acetone at 1 ppm, and outstanding selectivity with minor responses (Rair/Rgas < 3.5) for other interfering analytes including H2S. However, a noticeable response (Rair/Rgas = 13.2) toward methyl mercaptan at 1 ppm was also observed. Principal component analysis (PCA) was performed to understand the discrimination capability of multiple biomarker species using WO3 NF-based sensors including pristine WO3 NFs, Apo-Rh WO3 NFs, Apo-Pd WO3 NF, Apo-Pt WO3 NF, and Apo-PtY WO3 NFs. The PCA result revealed that all six biomarker species were distinguished without overlapping each boundary by pattern recognition (Figure 4b). To investigate potential feasibility for discrimination of mixture gases, different sensing responses using 4-sensor arrays such as Apo-Pt WO3 NF, Apo-Pd WO3 NF, Apo-Rh WO3 NF, and Apo-PtY WO3 NF were evaluated toward two mixture gases selected from each 1 ppm of H2S, acetone, and toluene (Supporting Information, Figure S4). The result revealed that sensing 1592

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596

Article

Accounts of Chemical Research

Figure 5. Real breath analysis for the diagnosis of halitosis. (a) Experimental procedures for the breath analysis. (b) Response characteristic of four different sensor arrays for simulated halitosis breath (n = 10) and healthy human breath (n = 10). (c) Classification between simulated halitosis breath and healthy human breath by a PCA.

outstanding dispersibility on WO3 NFs. In addition, multicompositional NPs can further enhance the sensitivity and selectivity toward specific biomarker species. The exceptionally high sensitivity of the Apo-NPs was verified by a comparison with acetone sensors that were previously reported in the literature (Supporting Information Figure S5). The detailed specifications of the catalyst-loaded WO3 NFs sensors can be found in Table S1 as well as the conventional acetone sensors in Table S2. All comparison results emphasize the outstanding sensing performance of the Apo-NPs loaded SMO NFs and NTs.

responses to both pure gases and two mixed gases were also successfully classified without overlapping, which demonstrates potential capability of pattern recognition toward real exhaled breath including complex gas mixture (Figure 4c). In addition, all the WO3 NF-based sensors functionalized with the Apo-NPs exhibited very low LOD values comparable to a sub-ppb level (Figure 4d). In particular, the Apo-Rh WO3 NFs exhibited the lowest LOD down to 0.17 ppb for H2S. This concentration was far below the 100% odor recognition concentration (1 ppm) and the lowest concentration (95 ppb) of an odorant presenting an objectionable odor.32 In addition, Apo-Pd WO3 NF presented a LOD of 0.98 ppb for toluene. This LOD is low enough for diagnostic purposes considering that the concentration levels of exhaled VOCs of lung cancer patients appear in the range of 10−100 ppb.7 The response time, which is the time required to reach 90% of the maximum response, is very important to rapidly capture specific biomarkers in the breath. A very fast response time less than 50 s was achieved with all the catalyst decorated WO3 NFs at a concentration ranging from 1−5 ppm (Figure 4e). In the case of the Apo-Pd WO3 NFs, the response time was less than 10 s, which is favorable for patients using exhaled breath analysis. To verify the robustness of the catalytic property of the ApoNPs on the WO3 NFs, electrical transductions of the Apo-Pt WO3 NFs and Apo-PtY WO3 NFs were compared with reference samples such as pristine WO3 NFs and WO3 NFs functionalized with Pt NPs synthesized by a conventional polyol process (hereafter, Pt-polyol WO3 NFs) (Figure 4f). Although the Pt-polyol WO3 NFs exhibited improved sensitivity with a response of 67.3 at 5 ppm for acetone, approximately a 2.25-fold enhancement in the response (Rair/ Rgas = 151.5) was observed for the Apo-Pt WO3 NFs at 5 ppm compared with the response of Pt-polyol WO3 NFs at 5 ppm. Furthermore, a dramatically improved acetone response (Rair/ Rgas = 423.4 at 5 ppm) was achieved with the Apo-PtY WO3 NFs. This result indicates that the Apo-NPs have a superior catalytic performance due to their small particle size and



EXHALED BREATH ANALYSIS: DISEASE DIAGNOSIS AND MOBILE APPLICATION Correlation studies that analyze biomarkers in the breath have been widely performed to diagnosis diseases.30,33 For the accurate detection of biomarkers, the use of multiple sensor arrays with characteristic sensitivity and selectivity toward specific biomarkers is essential. To investigate the feasibility of breath analysis, an experiment setup was systemically established by capturing real breath in Tedlar bags and injecting the captured breath into the sensor arrays, which were composed of WO3 NFs functionalized by Pt, Pd, Rh, and PtY (Figure 5a). For the breath analysis, exhaled breath was collected from healthy people (n = 10). In addition, simulated halitosis breath (n = 10) was synthesized by injecting H2S and maintaining a concentration of 1 ppm with real exhaled breath, which is a similar condition to the breath of a halitosis patient.34 The collected breath molecules were injected into the 4-sensor arrays, and the signal of each sensor was evaluated by measuring the response (Rgas/Rair). As shown in Figure 5b, the box plot based on the responses of the 4 sensors exhibited 2-fold higher response values for the simulated halitosis breath compared with healthy human breath due to the high sensitivity of the sensor arrays to H2S (Figure 5b). In addition, we performed a PCA to discriminate healthy human breath from halitosis breath. As a result, the simulated halitosis breath and 1593

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596

Article

Accounts of Chemical Research

Figure 6. Demonstration of mobile breath analysis using a portable sensing module. (a) Low-power sensing of simulated biomarker using MEMS sensor arrays and (b) pattern recognition of H2S, acetone, and toluene using PCA.

but also multicompositional NPs were synthesized to identify their novel catalytic activity. As promising sensing layers, SMO NFs with a bimodal porosity were prepared by the electrospinning technique combined with sacrificial templating route, which can readily facilitate gas penetration into the sensing layers. By exploiting the 1D nanostructures functionalized with the Apo-NPs, an unprecedented high sensitivity and selectivity were accomplished with a sub-ppb level of detection limit for biomarker species. These features are mainly attributed to the uniform functionalization of nanoscale (2−3 nm) catalysts with a high dispersive nature on the surface of the 1D SMO structures. Pattern recognition analysis with multiple sensor arrays offers enhanced identification of individual biomarker molecules discriminating against other interfering analytes. To further develop a reliable breath analyzer for diagnostic purposes, several issues should be addressed. First, since a number of VOCs are related simultaneously for a specific disease, multiple VOCs should be accurately detected and statistically analyzed using sensor arrays with high sensitivity and selectivity. Second, the concentration levels of VOCs vary from one patient to another. For this reason, a reliable sensing property of chemical sensors should be achieved even under the fluctuation of VOC levels. Lastly, humidity affects the sensing response, which may induce distortion of VOC responses. To obtain humidity insensitive SMO sensing layers, adequate catalytic functionalization should be performed. In addition to these limitations, correlation studies between biomarkers and diseases should be carefully attempted given the observation that the concentration of a certain VOC does not change as a disease evolves. A new class of sensor elements can be further invented by the discovery of multicompositional catalysts providing high catalytic performance on the SMO sensing layers. Innovation in sensing materials and platform will eventually fulfill a futuristic diagnostic platform providing a noninvasive and simple diagnostic method.

the breath of healthy people were clearly classified into two separate clusters without overlap (Figure 5c). To demonstrate the mobile healthcare application, the SMO NFs sensitized by the Apo-NPs, that is, Pt, PtCo, and PtY, were integrated with a portable diagnostic platform. Low power consumption is an essential requirement for portable sensor devices. To satisfy this demand, four different sensing layers composed of the pristine WO3 NFs, Apo-Pt WO3 NFs, ApoPtCo WO3 NFs, and Apo-PtY WO3 NFs were mounted on the microelectromechanical system (MEMS) heating substrates with an overall dimension of 1.4 mm × 1.4 mm (Supporting Information, Figure S6). The four MEMS sensors were then placed on a portable sensing module to establish 2 × 2 sensor arrays (yellow dotted box in Figure 6a). The gas sensing characteristics were measured using the sensor module integrated with a smartphone, which presents resistance transition curves on a smartphone screen (red dotted box in Figure 6a). Simulated breath analytes such as H2S, acetone, and toluene were injected at concentrations ranging from 1−5 ppm with 90% RH. The operating temperature of the MEMS heating substrates was maintained at 300 °C by applying voltages ranging from 2.4−3.5 V. For pattern recognition of the three different biomarker analytes, PCA was performed based on the characteristic response properties of the sensor arrays (Supporting Information, Figure S7). The result revealed that all the simulated biomarkers were clearly clustered in separate regions by PCA (Figure 6b). The demonstration of a portable breath analysis platform indicates the potential feasibility of SMO-based NFs, particularly sensitized by different Apo-NPs as promising breath biomarker detectors with high selectivity for fast monitoring of physical conditions and early diagnosis of diseases.



CONCLUSION AND PERSPECTIVES As evident in this Account, a futuristic diagnostic platform was introduced with diverse synthetic approaches and biomarker sensing properties. As a unique synthetic technique, bioinspired protein cages, apoferritin, were employed as robust encapsulators of diverse metallic NPs. Not only single component NPs 1594

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596

Article

Accounts of Chemical Research



ASSOCIATED CONTENT

at MIT. Prof. Kim’s current research emphasizes controlled processing and characterization of functional nanofibers via electrospinning for practical applications in exhaled breath sensors and energy storage devices. He has published over 177 articles and holds 180 patents. Prof. Kim is a Deputy Editor of the Journal of Electroceramics (Springer).

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.accounts.7b00047. Catalytic NPs synthesized with the apoferritin protein cage, particle size distributions of metal nanoparticles, schematic illustration and corresponding camera image of sensor architecture, schematic illustration of sensor measurement system, gas response properties of various NF constructs toward pure gases or gas mixtures, response characteristics of various catalyst-SMO composites for detection of acetone molecule under a humidified atmosphere, images of portable sensing module integrated with MEMS sensors, characteristic sensing properties of WO3 NF-based sensors functionalized with apoferritin-templated nanocatalysts, detailed sensing specifications of diverse catalyst-loaded SMO nanostructures, and sensing performances of conventional acetone sensors (PDF)





REFERENCES

(1) Khan, Y.; Ostfeld, A. E.; Lochner, C. M.; Pierre, A.; Arias, A. C. Monitoring of Vital Signs with Flexible and Wearable Medical Devices. Adv. Mater. 2016, 28, 4373−4395. (2) Konvalina, G.; Haick, H. Sensors for Breath Testing: From Nanomaterials to Comprehensive Disease Detection. Acc. Chem. Res. 2014, 47, 66−76. (3) Gao, W.; Emaminejad, S.; Nyein, H. Y. Y.; Challa, S.; Chen, K. V.; Peck, A.; Fahad, H. M.; Ota, H.; Shiraki, H.; Kiriya, D.; Lien, D. H.; Brooks, G. A.; Davis, R. W.; Javey, A. Fully integrated wearable sensor arrays for multiplexed in situ perspiration analysis. Nature 2016, 529, 509−514. (4) Hakim, M.; Broza, Y. Y.; Barash, O.; Peled, N.; Phillips, M.; Amann, A.; Haick, H. Volatile Organic Compounds of Lung Cancer and Possible Biochemical Pathways. Chem. Rev. 2012, 112, 5949− 5966. (5) Broza, Y. Y.; Haick, H. Nanomaterial-based sensors for detection of disease by volatile organic compounds. Nanomedicine 2013, 8, 785− 806. (6) Turner, A. P. F.; Magan, N. Electronic noses and disease diagnostics. Nat. Rev. Microbiol. 2004, 2, 161−166. (7) Peng, G.; Tisch, U.; Adams, O.; Hakim, M.; Shehada, N.; Broza, Y. Y.; Billan, S.; Abdah-Bortnyak, R.; Kuten, A.; Haick, H. Diagnosing lung cancer in exhaled breath using gold nanoparticles. Nat. Nanotechnol. 2009, 4, 669−673. (8) Sun, X. H.; Shao, K.; Wang, T. Detection of volatile organic compounds (VOCs) from exhaled breath as noninvasive methods for cancer diagnosis. Anal. Bioanal. Chem. 2016, 408, 2759−2780. (9) Righettoni, M.; Ragnoni, A.; Guntner, A. T.; Loccioni, C.; Pratsinis, S. E.; Risby, T. H. Monitoring breath markers under controlled conditions. J. Breath Res. 2015, 9, 047101. (10) Seiyama, T.; Kato, A.; Fujiishi, K.; Nagatani, M. A New Detector for Gaseous Components Using Semiconductive Thin Films. Anal. Chem. 1962, 34, 1502−1503. (11) Yamazoe, N.; Sakai, G.; Shimanoe, K. Oxide semiconductor gas sensors. Catal. Surv. Asia 2003, 7, 63−75. (12) Yamazoe, N. New Approaches for Improving Semiconductor Gas Sensors. Sens. Actuators, B 1991, 5, 7−19. (13) Kim, I. D.; Rothschild, A.; Tuller, H. L. Advances and new directions in gas-sensing devices. Acta Mater. 2013, 61, 974−1000. (14) Koo, W. T.; Choi, S. J.; Kim, S. J.; Jang, J. S.; Tuller, H. L.; Kim, I. D. Heterogeneous Sensitization of Metal-Organic Framework Driven Metal@Metal Oxide Complex Catalysts on an Oxide Nanofiber Scaffold Toward Superior Gas Sensors. J. Am. Chem. Soc. 2016, 138, 13431−13437. (15) Choi, S. J.; Lee, I.; Jang, B. H.; Youn, D. Y.; Ryu, W. H.; Park, C. O.; Kim, I. D. Selective Diagnosis of Diabetes Using Pt-Functionalized WO3 Hemitube Networks As a Sensing Layer of Acetone in Exhaled Breath. Anal. Chem. 2013, 85, 1792−1796. (16) Ma, N.; Suematsu, K.; Yuasa, M.; Shimanoe, K. Pd Size Effect on the Gas Sensing Properties of Pd-Loaded SnO2 in Humid Atmosphere. ACS Appl. Mater. Interfaces 2015, 7, 15618−15625. (17) Hwang, I. S.; Choi, J. K.; Woo, H. S.; Kim, S. J.; Jung, S. Y.; Seong, T. Y.; Kim, I. D.; Lee, J. H. Facile Control of C2H5OH Sensing Characteristics by Decorating Discrete Ag Nanoclusters on SnO2 Nanowire Networks. ACS Appl. Mater. Interfaces 2011, 3, 3140−3145. (18) Righettoni, M.; Amann, A.; Pratsinis, S. E. Breath analysis by nanostructured metal oxides as chemo-resistive gas sensors. Mater. Today 2015, 18, 163−171.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Il-Doo Kim: 0000-0002-9970-2218 Author Contributions §

S.-J. Kim and S.-J. Choi contributed equally.

Funding

This work was supported by Wearable Platform Materials Technology Center (WMC) funded by the National Research Foundation of Korea (NRF) Grant of the Korean Government (MSIP) (No. 2016R1A5A1009926). This work was also supported by the Ministry of Science, ICT & Future Planning as Biomedical Treatment Technology Development Project (2015M3A9D7067418). This work was also supported by the National Research Foundation (NRF) of Korea grant funded by the Ministry of Science, ICT and Future Planning (NRF2016R1E1A2A02945984). Notes

The authors declare no competing financial interest. Biographies Sang-Joon Kim received Ph.D. degree in the department of materials science and engineering from Korea Advanced Institute of Science and Technology (KAIST) in 2017. Seon-Jin Choi received Ph.D. degree in the department of materials science and engineering from KAIST in 2016. Currently, Dr. Choi is a postdoctoral research fellow in the applied science research institute at KAIST. Ji-Soo Jang received M.S. degree in the department of materials science and engineering from KAIST in 2016, where he is currently pursuing Ph.D. degree. Hee-Jin Cho received M.S. degree in the department of materials science and engineering from KAIST in 2015, where she is currently working on her Ph.D. degree. Il-Doo Kim received his Ph.D. degree (2002) from KAIST. From 2003 to 2005, he was a postdoctoral fellow with Prof. Harry L. Tuller 1595

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596

Article

Accounts of Chemical Research (19) Jutz, G.; van Rijn, P.; Miranda, B. S.; Boker, A. Ferritin: A Versatile Building Block for Bionanotechnology. Chem. Rev. 2015, 115, 1653−1701. (20) Crichton, R. R.; Declercq, J. P. X-ray structures of ferritins and related proteins. Biochim. Biophys. Acta, Gen. Subj. 2010, 1800, 706− 718. (21) Choi, S. J.; Kim, S. J.; Cho, H. J.; Jang, J. S.; Lin, Y. M.; Tuller, H. L.; Rutledge, G. C.; Kim, I. D. WO3 Nanofiber-Based Biomarker Detectors Enabled by Protein-Encapsulated Catalyst Self-Assembled on Polystyrene Colloid Templates. Small 2016, 12, 911−920. (22) Jiang, S.; Lv, L. P.; Landfester, K.; Crespy, D. Nanocontainers in and onto Nanofibers. Acc. Chem. Res. 2016, 49, 816−823. (23) Kim, S. J.; Choi, S. J.; Jang, J. S.; Kim, N. H.; Hakim, M.; Tuller, H. L.; Kim, I. D. Mesoporous WO3 Nanofibers with ProteinTemplated Nanoscale Catalysts for Detection of Trace Biomarkers in Exhaled Breath. ACS Nano 2016, 10, 5891−5899. (24) Choi, S. J.; Kim, S. J.; Koo, W. T.; Cho, H. J.; Kim, I. D. Catalyst-loaded porous WO3 nanofibers using catalyst-decorated polystyrene colloid templates for detection of biomarker molecules. Chem. Commun. 2015, 51, 2609−2612. (25) Choi, S. J.; Choi, C.; Kim, S. J.; Cho, H. J.; Hakim, M.; Jeon, S.; Kim, I. D. Highly Efficient Electronic Sensitization of Non-oxidized Graphene Flakes on Controlled Pore-loaded WO3 Nanofibers for Selective Detection of H2S Molecules. Sci. Rep. 2015, 5, 8067. (26) Jang, J. S.; Kim, S. J.; Choi, S. J.; Kim, N. H.; Hakim, M.; Rothschild, A.; Kim, I. D. Thin-walled SnO2 nanotubes functionalized with Pt and Au catalysts via the protein templating route and their selective detection of acetone and hydrogen sulfide molecules. Nanoscale 2015, 7, 16417−16426. (27) Jang, J. S.; Choi, S. J.; Kim, S. J.; Hakim, M.; Kim, I. D. Rational Design of Highly Porous SnO2 Nanotubes Functionalized with Biomimetic Nanocatalysts for Direct Observation of Simulated Diabetes. Adv. Funct. Mater. 2016, 26, 4740−4748. (28) Jang, J. S.; Yu, S.; Choi, S. J.; Kim, S. J.; Koo, W. T.; Kim, I. D. Metal Chelation Assisted In Situ Migration and Functionalization of Catalysts on Peapod-Like Hollow SnO2 toward a Superior Chemical Sensor. Small 2016, 12, 5989−5997. (29) Choi, S. J.; Chattopadhyay, S.; Kim, J. J.; Kim, S. J.; Tuller, H. L.; Rutledge, G. C.; Kim, I. D. Coaxial electrospinning of WO3 nanotubes functionalized with bio-inspired Pd catalysts and their superior hydrogen sensing performance. Nanoscale 2016, 8, 9159−9166. (30) Saalberg, Y.; Wolff, M. VOC breath biomarkers in lung cancer. Clin. Chim. Acta 2016, 459, 5−9. (31) Nakhleh, M. K.; Amal, H.; Jeries, R.; Broza, Y. Y.; Aboud, M.; Gharra, A.; Ivgi, H.; Khatib, S.; Badarneh, S.; Har-Shai, L.; GlassMarmor, L.; Lejbkowicz, I.; Miller, A.; Badarny, S.; Winer, R.; Finberg, J.; Cohen-Kaminsky, S.; Perros, F.; Montani, D.; Girerd, B.; Garcia, G.; Simonneau, G.; Nakhoul, F.; Baram, S.; Salim, R.; Hakim, M.; Gruber, M.; Ronen, O.; Marshak, T.; Doweck, I.; Nativ, O.; Bahouth, Z.; Shi, D. Y.; Zhang, W.; Hua, Q. L.; Pan, Y. Y.; Tao, L.; Liu, H.; Karban, A.; Koifman, E.; Rainis, T.; Skapars, R.; Sivins, A.; Ancans, G.; LiepnieceKarele, I.; Kikuste, I.; Lasina, I.; Tolmanis, I.; Johnson, D.; Millstone, S. Z.; Fulton, J.; Wells, J. W.; Wilf, L. H.; Humbert, M.; Leja, M.; Peled, N.; Haick, H. Diagnosis and Classification of 17 Diseases from 1404 Subjects via Pattern Analysis of Exhaled Molecules. ACS Nano 2017, 11, 112−125. (32) Tangerman, A.; Winkel, E. G. Extra-oral halitosis: an overview. J. Breath Res. 2010, 4, 017003. (33) Righettoni, M.; Schmid, A.; Amann, A.; Pratsinis, S. E. Correlations between blood glucose and breath components from portable gas sensors and PTR-TOF-MS. J. Breath Res. 2013, 7, 037110. (34) Deng, C. H.; Zhang, J.; Yu, X. F.; Zhang, W.; Zhang, X. M. Determination of acetone in human breath by gas chromatographymass spectrometry and solid-phase microextraction with on-fiber derivatization. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2004, 810, 269−275.

1596

DOI: 10.1021/acs.accounts.7b00047 Acc. Chem. Res. 2017, 50, 1587−1596