Nickel-Functionalized Solid-State Titania Nanotube Array for the Electrochemical Detection of Breath-Based Biomarkers in Colorectal Cancer

Colorectal cancer (CRC) is the second most common cancer in the United States with approximately 137,000 new cases each year, 50,000 of which are fatal due to belated screenings [1]. The gold-standard detection method is colonoscopy, a direct visualization tool that facilitates the removal of minor polyps [2]. The benefits, however, come with severe drawbacks such as prohibitive cost, extreme invasiveness, lengthy preparation and high complicacy [2]. Recently, breath-based volatile organic compounds (VOCs), or biomarkers, that are specific to colorectal cancer patients have been found, detection of which may provide a diagnosis of colorectal cancer [3]-[6]. This study aimed to synthesize a titanium dioxide nanotube array (TNA) sensor capable of detecting four critical breath-based biomarkers diagnostic of colorectal cancer: cyclohexane, 1,3 dimethylbenzene, methylcyclohexane, and decanal [5]. The TNA was synthesized via standard anodization procedures and functionalized with electrodeposited nickel. XPS studies showed this nickel was Ni(OH)2 on the surface capable of oxidizing the four VOCs to facilitate their detection. The VOCs themselves were detected amperometrically. The sensor was not only able to detect all four VOCs, but the detection profile for each VOC was also distinct indicating unique interactions between Ni(OH)2 complexes and the VOCs. Reaction mechanisms have been proposed to explain features observed in peak currents. The sensor has demonstrated itself to be a potentially portable, cost-effective, and non-invasive diagnostic tool that resolves problems found in current CRC diagnostic methods.

NICKEL-FUNCTIONALIZED SOLID-STATE TITANIA NANOTUBE ARRAY FOR THE ELECTROCHEMICAL DETECTION OF BREATH-BASED BIOMARKERS IN COLORECTAL CANCER by Anurag Tripathy A Senior Honors Thesis Submitted to the Faculty of The University of Utah In Partial Fulfillment of the Requirements for the Honors Degree in Bachelor of Science In Biomedical Engineering Approved: _____________________ Swomitra K. Mohanty, PhD Thesis Faculty Supervisor _____________________________ David W. Grainger, PhD Chair, Department of Biomedical Engineering _______________________________ Kelly W. Broadhead, PhD Honors Faculty Advisor _____________________________ Sylvia D. Torti, PhD Dean, Honors College May 2019 Copyright © 2019 All Rights Reserved i ABSTRACT Colorectal cancer (CRC) is the second most common cancer in the United States with approximately 137,000 new cases each year, 50,000 of which are fatal due to belated screenings [1]. The gold-standard detection method is colonoscopy, a direct visualization tool that facilitates the removal of minor polyps [2]. The benefits, however, come with severe drawbacks such as prohibitive cost, extreme invasiveness, lengthy preparation and high complicacy [2]. Recently, breath-based volatile organic compounds (VOCs), or biomarkers, that are specific to colorectal cancer patients have been found, detection of which may provide a diagnosis of colorectal cancer [3]-[6]. This study aimed to synthesize a titanium dioxide nanotube array (TNA) sensor capable of detecting four critical breathbased biomarkers diagnostic of colorectal cancer: cyclohexane, 1,3 dimethylbenzene, methylcyclohexane, and decanal [5]. The TNA was synthesized via standard anodization procedures and functionalized with electrodeposited nickel. XPS studies showed this nickel was Ni(OH)2 on the surface capable of oxidizing the four VOCs to facilitate their detection. The VOCs themselves were detected amperometrically. The sensor was not only able to detect all four VOCs, but the detection profile for each VOC was also distinct indicating unique interactions between Ni(OH)2 complexes and the VOCs. Reaction mechanisms have been proposed to explain features observed in peak currents. The sensor has demonstrated itself to be a potentially portable, cost-effective, and non-invasive diagnostic tool that resolves problems found in current CRC diagnostic methods. ii TABLE OF CONTENTS ABSTRACT ii INTRODUCTION 1 BACKGROUND 4 METHODS 7 RESULTS 11 DISCUSSION 24 ACKNOWLEDGEMENTS 34 REFERENCES 35 1 INTRODUCTION Colorectal cancer (CRC) is listed by the Center for Disease Control and Prevention (CDC) as the second most common cancer in the United States with approximately 137,000 new cases diagnosed in 2013 [1]. The CDC has further substantiated these statistics with claims that CRC prevalence is expected to remain constant until 2020. While the current CRC mortality rate is 8.5% in developed countries like the US [2], it is imperative that CRC deaths be reduced given that the number of cases will not be decreasing any time soon. More sensitive and earlier CRC detection can improve patient prognosis and reduce CRC deaths. CRC early detection can focus on early-stage CRC diseases by identifying localized polyps that characterize early-stage disease [2]. Polyp detection is commonly performed by a variety of detection methods ranging from the gold-standard colonoscopy to stool tests, double-contrast barium enema, and more complex in situ imaging methods [2]. The American Cancer Society assessed several tests based on performance, complexity, and limitations and found that while the preparation and complexity of all tests, except colonoscopy, was intermediate-to-low, diagnosis confirmation still necessitated a colonoscopy [1], [3]. Colonoscopies allow direct visualization of the entire colon and even facilitate removal of some minor polyps [2], [3]. Though the most sensitive diagnostic method currently, colonoscopy comes with some severe limitations including a higher risk of complications from bowel tears and bleedings, exorbitant costs, required sedation, and general full bowel preparation [1], [3]. Due to the extreme invasiveness and the generally unpleasant experience of colonoscopies, a non-invasive, more precise and cost-effective diagnostic tool is paramount for the detecting precursor polyps [1]. Recent studies in 2 cellular metabolism have pointed to exhaled volatile organic compounds (VOCs) as possible diagnostic targets for CRC [4], [5]. VOCs are gaseous molecules (under standard conditions) found in a diverse array of biological samples including urine, blood, stool, and breath [5]-[9]. VOCs may be produced because of genomic and/or protein changes due to tumor growth, resulting in peroyxgenation of different components of cell membranes [5]. The resultant colonic VOCs are then carried by venous blood to the lung alveoli and exhaled by CRC patients [4], [5]. Thus, exhaled VOCs present themselves as an objective form that can be used as the basis for CRC diagnostics [4]-[7]. Exhaustive studies into the specificity of VOCs as diagnostic biomarkers have resulted in two critical conclusions: 1) the VOC profiles of no two diseases are the same [5], and 2) exhaled VOCs are significantly different in cancerous and non-cancerous patients [4]-[9]. Because of these properties, VOC detection presents a new domain of medical diagnostics because it would be non-invasive and potentially quite inexpensive. Using this understanding, the pilot study of Altomare et al. [5] recently identified fifteen VOCs in the air exhaled fromCRC patients, four of which have distinguishing capabilities (cancer vs. non-cancer patients) of greater than 92%. The four VOCs were: cyclohexane, 1,3-dimethylbenzene, methylcyclohexane and decanal. Detection of these four breath-based VOCs may potentially be able to discriminate CRC patients from normal, healthy patients. The objective of this study was to design a titanium dioxide (TiO2)-based nanotube array (TNA) sensor capable of detecting CRC-relevant VOCs under electrochemical settings. Metal oxides, such as TiO2, are used in gas sensing devices to facilitate surface absorption and desorption of the gases prompted by electron transfers on the oxide film 3 surfaces [10]. It is also favored for its robust mechanical properties along with high corrosion resistance, and reasonable cost [11], [12]. Gas sensitivities and detection are dependent upon the modifications of TiO2 with certain metallic depositions [11]. The copresence of the metallic constituent is expected to alter the band gap of the overall structure allowing for greater conductivity [10]-[12]. In this case, nickel functionalization can increase the sensitivity of the TiO2 substrate and perform the oxidation of the four target VOCs [13]-[16]. The effective binding of VOCs to nickel was assessed at certain optimal voltages determined by cyclic voltammetry studies. The TNA was synthesized via electrochemical anodization while its functionalization with nickel was achieved via cathodic electrodeposition. Sensor-based VOC detection was measured as a function of current vs. time. From the experiments, it was determined that the designed sensor was indeed capable of detecting the biomarkers of interest. The results of this study could be used to develop a new portable, cost-effective, and non-invasive diagnostic instrument capable of resolving the issues presented by current CRC diagnostic methods. 4 BACKGROUND Electrochemical detection of VOCs for diagnostic purposes is an exciting, ongoing field of research with applications in a variety of medical diseases. The idea is a significant departure from the veritable pantheon of previous studies that have identified VOCs of such diseases using derivatives of Gas Chromatography-Mass Spectrometry (GC-MS) technology, which, despite its high accuracy and specificity, suffers from the disadvantage of low clinical translatability. In the study of Peng et al. [6], organically functionalized gold nanoparticles (GNPs) on a cross-reactive nanosensor array were analyzed for their ability to distinguish between breath VOCs of healthy patients and patients afflicted with colorectal, lung, prostate, and breast cancers. Though the broad nature of the cross-reactive GNP sensor successfully helped identify cancer odor prints, it demonstrated poor selectivity and specificity in discriminating between cancer patients because it failed to detect the constituent VOCs of the different cancers. Sensitivity, specificity, and selectivity are critical parameters for a successful diagnostic tool intended for use in point-of-care settings making the quest for a novel, non-invasive screening system, specific to CRC, increasingly important [1], [3], [17]. In this study, the synthesized sensor’s design highlights the capabilities of functionalized metal oxides on inert, but stable, substrates for electrochemical vapor sensing. Recent studies by Zonta et al. [18] and Malagù et al. [19] have successfully demonstrated the application of various metal oxide films in detecting gas phase CRC VOCs such as 1-iodononane, decanal and benzene, detectable in patients’ stool samples. In the Zonta et al. study [18], the oxide film consisted of a mix of titanium (Ti), tantalum (Ta), and vanadium (V) and exhibited strong detection of benzene. In contrast, the oxide 5 film mix of titanium and tin (Sn) demonstrated a better detection capability for 1iodononane and decanal. The success of these sensors in detecting CRC VOCs must be contextualized with the fact that they require temperatures of up to 650oC for operation [18], [19]. Elevated temperatures are necessary for converting atmospheric (O2), adsorbed onto the surface of these sensors, into the superoxide anion (O2-) which then interacts with the gases. While this reaction mechanism is an efficient method of detecting the VOCs, it possesses two major drawbacks: 1) a lack of specificity due to the wide number of gases the O2- anion can react with; and 2) high temperatures require considerably more power and elaborate reactor/sensor designs making the application of these sensors in point-ofcare settings highly inefficient [6], [10], [11]. Indeed, their sensor design lacked truly specific “recognition factors” for absolute specificity since the Ti/Ta/V oxide exhibited good detection of methane gas as well [18]. In contrast to this, electrochemical, metallicbased sensing, recent studies have shown the potential of polymeric electronic nose (enose) technology for non-invasive screening of CRC patients via detection of VOCs in fecal gas [20]. Though this technology has demonstrated favorable advantages like high sensitivity and wide applicability, limitations such as loss of sensitivity due to water vapor or oversaturation by one target compound, relatively low shelf-lives, sensor drift, and general hardware issues prevent their widespread adoption for human disease diagnostics [21]. The chosen sensor design in this study consisted of a TiO2 substrate with electrodeposited nickel on the surface. Previous studies have demonstrated the feasibility of such a strategy in the detection of breath-based VOCs of tuberculosis, and the highly favorable morphology and versatility that these design elements confer on the sensors [11], 6 [12]. The decision of using nickel for the detection of the CRC breath VOCs was made due to its successful application in various industrial catalytical systems in previous studies [13]-[16] which demonstrated high catalytical activity in environments consisting of these compounds. For example, the oxidation of cyclohexane and other hydrocarbons has been of great industrial attention for producing compounds like adipic acid, vital in the synthesis of Nylon-6 and Nylon-66 [24], [25]. In the study of Alshammari et al. [25], direct catalytical conversion of cyclohexane to adipic acid was performed with nanogold catalysts while copper nanoparticles on a chromium (III) oxide (Cr2O3) substrate, synthesized by Sarkar et al. [26], performed the same conversion at room temperature. The same oxidation of cyclohexane has been reported with hybridized substrates like tungsten trioxide/vanadium pentoxide (WO3/V2O5), with H2O2 as a generic oxidant, and in photocatalytic systems that utilized TiO2 functionalized with nickel, iron, and gold [15], [24]. Another CRC breath VOC, 1,3-dimethylbenzene, was demonstrated to be catalytically oxidized with nickel impregnated carbon fiber in the study of Gaur et al. [19] Yolcular et al.’s Ni/Al2O3 catalysts [21] exhibited dehydrogenation activity with methylcyclohexane (another CRC breath VOC) for hydrogen production. Evidence provided in these studies further supported the work of Hosseini et al. [22] which demonstrated nickel’s potential to be a highly active oxidant element capable of interacting with the target VOCs. Nickel was additionally chosen for its high compatibility with TiO2 substrates and the good sensing capabilities the composite system provides as exhibited in the study of Li et al. [27] whose sensors functioned well in hydrogen environments of various concentrations, across a wide temperature range (25oC – 200oC). 7 METHODS A. TiO2 Nanotube Array (TNA) Synthesis The TiO2 nanotubes were grown on titanium foil as per the protocol set in the studies of Bhattacharyya et al. [11]. These foils (GI grade 0.004”x4”x12” ESPI, 99.99%) were first cut into 1.5x1.5cm coupons. Each coupon was then mechanically polished with emery paper. One side of the coupons was taped with Kapton tape to ensure that the titania tubes grow only on one side. The coupons were then degreased via ultrasonication in a 50vol% acetone and ethanol solution for 20 min in a Branson 5510 ultrasonicator. The coupons were then anodized in an ethylene glycol (Fischer Scientific Grade), ammonium fluoride (Fischer Scientific Grade), and DI water solution with the following recipe (all by % weight): Ethylene Glycol-96.5%; Ammonium Fluoride-0.5%; DI H2O-3%. The anodization of the coupons was carried out at a constant 30V (Agilent E3647A Power Supply) for 60 min in a Teflon beaker on a stir plate (60 rpm). A Pt foil was used as the counter electrode and held ~2.0 cm away for all the coupons. Post anodization, the coupons were again ultrasonicated in DI H2O for ~5 sec and then cleaned with isopropanol. The coupons were finally left to dry in an oven (VWR) at 120oC overnight. The anodized coupons were then annealed in oxygen at 500oC for 2 hours with a ramp rate of 3oC/min and were then placed in desiccant until functionalization to prevent moisture damage. B. Electroplating of Nickel on TNA Nickel was electroplated on annealed TiO2 nanotubes in a 0.5M aqueous NiCl2 solution [22]. The pH of the solution was adjusted to 8.0 with 30% ammonia, ammonium hydroxide solution (Fisher Chemicals). The deposition was carried out at room temperature, under 8 cathodic polarization at 22.5 mA, a current density of 10 mA/cm2 for 1 min [22]. A Pt foil was used as an anode. The anode-to-cathode distance was ~2.0 cm. The Ni-deposited TiO2 nanotubes were then rinsed with DI water and dried overnight in an oven (VWR) at 110oC and were then placed in a desiccant to prevent moisture damage. C. Characterization of the sensor a) Scanning Electron Microscopy (SEM) The morphological features of the synthesized TNA, with and without nickel deposition, were examined under a Hitachi S-4800 scanning electron microscope (SEM) with an Energy Dispersive X-ray Spectroscopic (EDS;Oxford make) attachment; the latter was used in conjunction with the SEM images to approximate element distributions on the sample. EDS analysis was carried out at 20kV accelerating voltage and high probe current. AZtecEnergy acquisition and EDS analysis software were synchronized with the X-Max detector for mapping and spectral analysis. b) X-ray photoelectron spectroscopy (XPS) The surface composition of the nickel-deposited TiO2 structures was analyzed by XPS (Kratos Axis, Ulta DLD model, Utah Core facility). c) X-ray powder diffraction (XRD) X-ray diffraction (XRD) was carried out using a Rigaku Miniflex XRD (CuKa = 1.54059 Å) from 2θ = 20 to 80o with a step size of 0.015o and dwell time of 1o/min. The diffraction patterns were analyzed using Rigaku PDXL2 analysis software and indexed with standard JCPDS cards. 9 D. Volatile Organic Compound Detection The detection of the four VOCs was performed in a specialized sensing chamber made from a 50mL centrifuge tube with two alligator clip extensions, under ambient conditions. Figure 1 below shows the experimental setup used for sensing. One side of each alligator clip was insulated with Kapton tape such that one connection was made to the Ni-deposited TNA side while the other was made to the unanodized face of the coupon. A GAMRY Reference 600 two-electrode based potentiostat was used during these experiments. VOC vapor, in concentrations of 0.1, 1.0 and 10.0 mM was introduced into the sensing chamber using nitrogen as the carrier gas at a flow rate of 200sccm (standard cubic centimeters per minute). From our cyclic voltammetry (CV) studies (Figure 9 below shows a representative CV plot) bias voltages of 1.0V, 1.40V, 1.45V, and 1.55V were found to be optimal for the detection of cyclohexane, 1,3 dimethylbenzene, decanal and methylcyclohexane, respectively. For the first ~130 seconds, only nitrogen gas was flowed to establish the baseline current. VOC vapor was then introduced, and VOC detection by the sensor was allowed to proceed until the maximum current was observed. VOC flow was then disconnected, and the sensor was exposed to nitrogen gas again. Several trials for each VOC at each concentration were conducted. The data presented here show the calculated average peak currents recorded. 10 Figure 1: Experimental Setup. Schematic arrangement of the nickel-deposited TNA for the detection of cyclohexane (and other VOCs) under ambient conditions. 11 RESULTS A. TNA Surface Characteristics Scanning electron micrographs showed the arrangement of the TNA to be ordered and regular-sized (Fig. 3). A top side view indicated that the tubes had a diameter in the range 55-60 nm and a wall thickness of approximately 14 nm. Side views of the nanotube array (Fig. 3), suggested the length of the nanotubes to be in the range of 1.21.5 µm. The TNA film indicated the presence of anatase titania which upon annealing in oxygen gas exhibits a greater resistivity, a property vital for such VOC electrochemical detection [6]. Scanning electron micrographs of Ni-deposited TNA showed deposition of nickel as globules on the TNA surface (Fig. 4). The diameter of the globules was in the range 0.4-0.6 µm. Fig. 4 also shows the homogenous distribution of the nickel globules on the TNA surface. Lateral cross-sections of these globules revealed that individual particulates, albeit smaller in diameter, of nickel fit within the openings of the tubes. This supports the presumption that nickel electroplating favored the congregation of the particulates into globules rather than individual particles on the surface. EDS mapping of the SEM images revealed the chemical composition of the nickel globules to consist of 47.0 wt % titanium, 26.8 wt % nickel and 26.2 wt % oxygen respectively (Fig. 4b-c and Fig. 5). No discernable nickel was present on the inner surface of the tubes. 12 Figure 3: SEM images of the unfunctionalized TNA: (a) Tubular view of the TNA; (b) Top side view of the TNA; (c) Lateral cross-section of the TNA at a scratched surface Figure 4: SEM-EDS images of the nickel-deposited TNA: (a) SEM image of Ni deposited TNA; (b) EDS image of Ti, O, and Ni on the surface (c)EDS showing nickel deposition exclusively on the surface. Note that dense zones of red particulates in (c) correspond to the position of the globules observable in (b). 13 Figure 5: Spectra of relative mass percentages of titanium, nickel, and oxygen on the Nideposited TNA surface. B. XPS Characterization The chemical nature of the electrodeposited nickel on the surface of TNA was observed to consist of Ni(OH)2. A typical photoelectron spectrum, over the entire cross-section of the sample, showed the presence of a multitude of peaks, the prominent ones being nickel and oxygen (Fig. 6). Oxygen’s O1s peak is observable at 532eV (Fig. 6). In the case of nickel (Fig. 7), two prominent peaks of interest were observed: Ni 2p1/2 at 872eV, and Ni 2p as well as Ni 2p3/2 both at 855eV. Wu et al. [23] have confirmed that these energy values correspond to Ni(OH)2. A gap of ~17eV between the nickel peaks indicated that the Ni 2p peak has significant split spin-orbit components. A small, almost imperceptible gap exists between the 2p and 2p3/2 peaks, but due to spin coupling 14 and overlay of similar energy orbitals on the nickel atom, they appeared as one peak. Towards the lower end of the XPS spectrum (0-100eV, Fig. 6) Ni3s and Ni3p peaks were observed. However, due to their low energy values and minor peak heights, these were assumed to be not participating during the progression of the reaction. Figure 6: Complete wide-scan XPS spectrum of the nickel-deposited TNA showing the energy states of the constituent nickel and oxygen atoms on the surface. Significant Ni 2p1/2, Ni2p, and O1s peaks are visible indicating the formation of Ni(OH)2 on the NiTNA surface. 15 Figure 7: Hi-res XPS spectrum showing only prominent nickel peaks from the nickel globules on the surface of the Ni-TNA. Peaks of Ni2p1/2 at 872eV, and Ni2p, as well as Ni2p3/2 both at 855eV, are evidence of the formation of Ni(OH)2 on the surface. C. XRD Characterization XRD analysis of the Ni-TNA surface was performed to determine the crystallinity of the atoms present. The XRD profile of the surface (Fig. 8) shows that the TNA component of the sensor consisted of both anatase and rutile TiO2, the former being the dominant crystalline form. The nickel globules on the surface were found to exist in two separate crystalline forms as well: α-Ni(OH)2 and β-Ni(OH)2, the latter of which was the dominant crystalline form of nickel [23]. An important observation in the spectrum is the lack of metallic nickel on the surface indicating that only the two hydroxide crystalline forms of nickel are found on the sensor. 16 Figure 8: XRD profile of the nickel-deposited TNA surface showing the crystalline phases of both titanium and nickel. D. Detection of the VOCs with the sensor Detection of the four VOCs was recorded as a change in current over time with a constant applied potential. Figure 10 shows a representative detection profile of 10 mM cyclohexane at 1.0V. Nitrogen gas was exclusively flowed for ~130 sec followed by the introduction of the VOC vapor; the current response of the sensor to nitrogen gas was ~20-35 nA. After a 15 sec delay, the sensor picks up the VOC, and the current then proceeds to rise with noise appearing around ~700 s. The noise persists, and calculation of the peak current was done by the root mean square of the noise (after 800s because that is where the noise starts to equilibrate) giving a peak current of around ~210 µA. The VOC flow was then discontinued leading to a sharp decrease in the observed current. Current profiles for all VOCs at all three concentrations were recorded. Figures 17 11-14, below, provide a summary of all the peak currents that were measured for the 4 VOCs at the different concentrations. The figures also show the limit of detection of the sensor for each of the VOCs. The detection limit was calculated with the following equation: 𝐷𝑒𝑡𝑒𝑐𝑡𝑖𝑜𝑛 𝑙𝑖𝑚𝑖𝑡 (𝐷𝐿) = 3 𝑥 𝑆𝑡𝑑.𝐷𝑒𝑣 𝑜𝑓 𝑠𝑖𝑔𝑛𝑎𝑙 𝑎𝑡 𝑙𝑜𝑤 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑆𝑙𝑜𝑝𝑒 𝑜𝑓 𝐶𝑎𝑙𝑖𝑏𝑟𝑎𝑡𝑖𝑜𝑛 𝐿𝑖𝑛𝑒 (1) Here Std. Dev refers to the standard deviation, and the slope of the calibration line is calculated from the slope of the regression line fit to the data at all the concentrations. Five sensors were used for each of the concentrations of all four VOCs (i.e., 60 sensors in total were used). It should be noted that for the objective of this study, a greater emphasis has been placed on the presence of significant positive slope indicating detection of the VOC at its specific bias voltage, rather than the exact value of the current peak. 18 Figure 9: Sample plot of cyclic voltammetry (CV) run of a 50/50 (by vol.) of 0.5M NiCl 2 (aq.) and 10mM cyclohexane in ethanol. Electrode setup consisted of a TiO2 working electrode, Pt counter electrode and Ag(AgCl) reference electrode. Current onset at 1V is indicative of a reaction between the sensor and the biomarker. 19 Figure 10: Detection of 10mM cyclohexane VOC at 1.0V bias voltage over a period of 1000s. 20 Figure 11: Chart depicting average peak currents recorded across five sensors (each bar) during the detection of cyclohexane. The error bars indicate the spread of peak current recorded for the same VOC concentration by the different sensors. Currents were also measured by sensors five sensors each for ethanol and ambient air as controls and were found to be in the range of 2.5-5μA and 25-45nA, respectively. 21 Figure 12: Chart depicting average peak currents recorded across five sensors (each bar) during the detection of 1,3 dimethylbenzene. The error bars indicate the spread of peak current recorded for the same VOC concentration by the different sensors. 22 Figure 13: Chart depicting average peak currents recorded across five sensors (each bar) during the detection of methylcyclohexane. The error bars indicate the spread of peak current recorded for the same VOC concentration by the different sensors. 23 Figure 14: Chart depicting average peak currents recorded across five sensors (each bar) during the detection of decanal. The error bars indicate the spread of peak current recorded for the same VOC concentration by the different sensors. 24 DISCUSSION This study aimed to design a nickel functionalized TiO2 nanotube array (TNA) sensor that can detect four critical breath-based volatile organic compounds (VOCs) that are diagnostic of colorectal cancer [5]. The TNA was synthesized via a standard anodization procedure, and the functionalization of nickel was performed through electrochemical deposition. The sensor was exposed to varying concentrations of VOC vapor at different pre-determined bias voltages corresponding to each VOC. The collected current profiles demonstrate that the synthesized sensor not only detects the 4 VOCs of interest but the response also varies with VOC concentration, i.e., higher concentrations result in greater current peaks. These results demonstrate the potential of this sensor to detect VOCs present in the breath of colorectal cancer patients. To understand the observed detection patterns, the formation of the active nickel complexes on the TNA must be explained. SEM and EDS analysis demonstrated the presence of Ni(OH)2 complexes on the Ni-deposited TNA surface. Thermodynamics does not favor the hydrolysis of NiCl2 to Ni(OH)2 as the calculated free energy of formation for the reaction (Eqn. 2) at 25oC is 94.9kJmol-1. However, under the applied current of ~22.5mA, NiCl2 might have decomposed to give free Ni2+ on the TNA. The nickel ions subsequently formed Ni(OH)2 species while being surrounded by water. This is plausible as no metallic nickel could be detected in the XRD profile of the nickel-deposited TNA (Fig. 8), only a mixture of TiO2 (both anatase and rutile) and Ni(OH)2 phases only. Ni2+ + 2Cl- + 2H2O → Ni(OH)2 + 2HCl (2) It is reported that the gas sensing capabilities of sensors are based on changes in the conductivity of the oxide layer present on or near the surface of the sensor [10], [11], [12]. 25 This conductivity change is profoundly impacted by gas adsorption and/or the formation of complexes on the sensor surface; the formation of complexes is the primary mechanism of detection for these VOCs. Bhattacharyya et al. [11] proposed the formation of a complex involving Co2+ and biomarker methyl nicotinate to be the dominant mechanism of their sensor, like the one created in this study. From both the present study and the work published by Bhattacharyya et al. [11], it is evident that the presence of metal ions not only imparts stability to the sensor but also act as effective oxidizing agents. The formation of the complexes, however, depends heavily on the properties of TiO2 nanotubes. The unique characteristics of TiO2 nanotubes facilitate electron transfers in the oxidation of the VOCs. The interaction between nickel complexes and VOCs was increased due to the great stability and large surface area provided by the nanotubes [11], [12]. The tubular structure maximized electron transfer while minimizing the loss of electrons participating in the reactions [11]. TiO2 nanotubes also enable the adsorption of surface oxygen by donating electrons, creating the superoxide anion of oxygen (O2 (gas) to O2(ads)); the reaction pathway for superoxide anion can be found below [11], [12]. O2 (gas) → O2 (ads) (3) O2 (ads) + e- → O2- (ads) (4) O2- (ads) + e- → 2O- (ads) (5) O- (ads) + e- → O2- (ads) (6) This anion, in turn, generated a facile path for the initiation of an effective bonding between surface nickel and the VOCs by oxidizing the latter, resulting in unstable intermediates [10]. These intermediates were subsequently stabilized by the presence of the nickel complexes. For instance, Ide et al. [15], described the superoxide anion as having 26 played a decisive role in the formation of cyclohexanone and cyclohexanol, two important intermediates in the oxidative conversion of cyclohexane into adipic acid (further details provided below). They further reported that the presence of OH groups on the surface facilitated the direct interaction between TiO2 and the cyclohexane. Bhattacharyya et al. [11] have also supported this by reporting the OH groups as great Brönsted-Lowry bases and key providers of stability to their cobalt-TiO2 organometallic complexes [11]. It is this sequence of steps that is speculated to have contributed to the formation of the various complexes between Ni(OH)2 and VOCs that were first oxidized by generated O2- followed by reduction via the OH groups of Ni(OH)2 leading to a complex formation. The oxidation of cyclohexane to adipic acid can follow one of two reaction pathways: direct conversion to adipic acid or via the formation of the intermediates cyclohexanol and cyclohexanone [24], [25]. For both pathways, the creation of the intermediate cyclohexyl radical is necessary. In the photocatalytic experiments of Ide et al. [15], the intermediate radical was formed on TiO2 due to the direct interaction of cyclohexane with either the valence band hole of TiO2 or a combination of the valence band hole and its reduction by an OH group. TiO2 is an n-type semiconductor so the interaction of its valence band hole with cyclohexane will be minimal under electrochemical settings such as ours; it was a poor oxidizing agent. This was observed in our cyclic voltammetry studies (not shown) which proved the need for functionalization of TiO2 nanotubes with a more active metal for the voltammetric detection of the VOCs. The synthesized Ni(OH)2 complexes, more specifically the OH groups, proved to be crucial for the detection of the VOCs via the oxidation of the compounds. The presence of OH groups on the surface increased the general conductivity of n-type TiO2 sensor and served as great Brönsted-Lowry acids. 27 Figure 15 below depicts a possible reaction mechanism of the interaction of Ni(OH)2 and cyclohexane; cyclohexane could be oxidized to a nickel cyclohexyl complex with two cyclohexane molecules binding to one nickel atom. The comparatively high current response from cyclohexane detection (Fig. 11) indicates a strong interaction between nickel and cyclohexane, something that is well corroborated by the mechanism shown in Figure 15 which demonstrates a 1:2 ratio of nickel to cyclohexane. This observation also indicates that the formation of this complex between nickel and two cyclohexyl molecules results in the largest increase in conductivity out of all the other VOCs, due to the presence of the Ni atom. It should be noted that the precise reason for this complex having the highest conductivity increase relative to the other complexes is unknown; this is beyond the scope of the project. Figure 15: Proposed complex formation between Ni(OH)2 complex and two cyclohexane molecules. 28 Figure 16: Schematic diagram showing the formation of the nickel cyclohexyl complex on the TNA. Similar complexes were formed with other VOCs (shown below) during their respective detection. Much like cyclohexane, 1,3 Dimethylbenzene, also known as m-xylene, and its interaction with nickel followed a similar oxidation process. Oxidation processes convert m-xylene into isophthalic acid, and this process constitutes the formation of a series of intermediates [14], [28], [29]. These intermediates were formed by the adsorbed oxygen formed by TiO2 and were stabilized by their interaction with OH groups of Ni(OH)2. Figure 17, below, shows a potential reaction scheme for this interaction. The primary difference between the cyclohexane and m-xylene reaction schemes is the inverse ratio of binding between nickel and the biomarker- there was 1 nickel atom per 2 cyclohexane molecules while 2 nickel atoms were complexed with 1 m-xylene molecule. 29 Figure 17: Proposed complex formation during the oxidation of m-xylene Correspondingly, the peak detection current of the biomarkers represents that difference in binding ratios- the peak currents were significantly lower for m-xylene than they were for cyclohexane indicating that while there was an increase in conductivity, it was not to the same degree as the nickel-cyclohexyl complex. While the cyclohexane and m-xylene reaction schemes made use of adsorbed oxygen and nickel for the oxidation processes, the interaction between the nickel complexes and methylcyclohexane involved a series of dehydrogenation steps (the simplest form of oxidation) [16]. This dehydrogenation process is not dependent on the adsorbed oxygen on the TiO2 surface, and instead, when methylcyclohexane is adsorbed onto the nickel complexes, the axial hydrogen atom of the former is removed [16]. This dehydrogenation is heavily favored on the metallic nickel complexes and eventually forms toluene [16]. Figure 18 below depicts a potential intermediate complex that is formed during this process. The detection of methylcyclohexane followed a pattern akin to that of m-xylene. In this case, 3 nickel atoms interact with 1 methylcyclohexane molecule. Correspondingly, the peak currents observed for this process were lower than the other two indicating that this complex formation experiences the lowest conductivity increase out of all the VOCs. 30 Figure 18: Proposed complex formation during the oxidation of methylcyclohexane Figure 19: Proposed complex formation during the oxidation of decanal The detection of the fourth and final VOC of interest, decanal, also consisted of an oxidation reaction that result in conversion to a peroxy acid [13]. Much like cyclohexane and m-xylene, adsorbed oxygen formed by TiO2 functioned as the primary catalyst for this conversion while the OH performed as intermediate complex stabilizers. Figure 19 above shows a potential reaction mechanism for the formation of the complex between Ni(OH) 2 and decanal. Decanal’s interaction with the nickel complexes follows a simple 1:1 ratio. Peak currents observed for decanal detection are markedly higher than those of methylcyclohexane and m-xylene but lower than those seen in the detection of cyclohexane. This demonstrates a conductivity increase almost as high as that of the nickelcyclohexyl complex, but markedly more than those of the other two complexes. A conflicting limitation of this work includes the possibility of one of the selected VOCs not being specific to colorectal cancer. Recently, Markar et al. [30] published their findings on breath-based VOCs of stomach and esophageal cancers where decanal was labeled as one of the critical diagnostic biomarkers for those cancers. Due to this finding, it is speculated that decanal may not be a breath biomarker specific to colorectal cancer, 31 but may be indicative of gastrointestinal cancers in general. However, it should be noted that this methodology of diagnosing colorectal cancer requires the specific presence of all 4 of the critical VOCs and so, even though decanal may not be specific to CRC, its detection with the other three VOCs is a strong indication of colorectal cancer. The collective pattern of all 4 VOCs is more important than the presence of any single VOC. An important limitation which recontextualizes the results collected in this study (and pervasive throughout the field of breath diagnostics) is the lack of consensus on the concentrations of these VOCs present in the breath of CRC patients. While this study used 10mM, 1mM, and 0.1mM as standard values, the exact concentration of each of VOCs is unknown. Furthermore, no standardized methods have been proposed to determine the precise concentrations of these VOCs in a patient breath. This is an important limitation to keep in mind because information about the concentration of VOCs can drive design decisions to improve the sensitivity of the sensors. For the successful application of a breath-based sensor, a keen understanding of VOC concentrations is a fundamental necessity. Some previous studies have attempted to resolve this dilemma in other diseases like tuberculosis and lung cancer through rigorous GC-MS analyses with carefully selected calibration samples. However, none of the quantitative values have been strongly supported and so determining the strict concentrations of VOCs for all conditions is a difficult and highly active area of research in the field of breath diagnostics, as a whole. A secondary limitation of this study is the methodology of biomarker detection. Here we used amperometric detection with bias voltages “specific” to each biomarker for their detection. However, an examination of the voltages (section D of methods) shows that the difference between, for instance, decanal’s and 1,3 dimethylbenzene’s bias 32 voltages, is a mere 50mV. This a highly insignificant difference considering the proposed reaction mechanisms between the sensor and the biomarkers is oxidation of the latter. In other words, it is highly likely that if 1.4V is applied for the detection of decanal and 1.45V for that of 1,3 dimethylbenzene, comparable currents will be observed, attributed to either or both and not selective. This is supported by the moderately-sized error bars in Figure 11 which show comparable currents for the two (and the rest of the biomarkers in general), especially at low concentrations. Consequently, it is speculated that these “bias” voltages may indeed not be specific to their respective biomarkers and could facilitate the detection of any of the VOCs. This is problematic because if sharp current increases are observed with the application of 1.0V or 1.4V on CRC patient breath or even mimicked environments with all four VOCs mixed, it would be unclear as to what specific VOC (among the four or even other confounding ones) is causing that increase. Though amperometric detection provided helpful preliminary information about the sensor and its interactions with the biomarkers, future work will be aimed at not only further improving the morphological properties of the sensor but also exploring a better electrochemical technique that can provide “true” biomarker sensing specificity, individually as well as in mixed environments. It is believed that cyclic voltammetry might meet the specificity requirements due to its capability to distinguish between competing for redox reactions like thos proposed above. Cyclic voltammetry may also facilitate a more in-depth examination of the reaction mechanisms and subsequent adjustments. Improvements in sensor synthesis process could also enhance VOC detection via improved nanotube geometry and homogeneity of nickel deposition on the surface, both being explored currently. Additionally, a whole new sensor design consisting of 50/50 (% 33 elemental composition) nickel-titanium alloy called nitinol, is being considered as an option. It is hypothesized that only the standard anodization and annealing steps (as described above) will be sufficient to generate the surface Ni(OH)2 layer desired for the detection of the biomarkers [31]. It may outperform the current sensor design since the nanotubes will have Ni(OH)2 complexes integrated within the walls instead of just lying on the surface thereby increasing the overall active surface area of detection. Following the optimization of the sensor design and the selection of a successful electrochemical technique for VOC detection, a future test will involve taking healthy human breath samples, directly injecting them with the target VOCs, and examining whether the sensor can selectively detect those VOCs in the breath sample. This will distinguish the designed sensor as not a mere gas sensor but rather a breath sensor. It is the expectation that the success of this sensing technology will contribute to an earlier and easier screening of potential colorectal cancer patients thereby increasing the likelihood of earlier treatment, and a net decrease in mortality rates around the world. 34 ACKNOWLEDGMENTS The author would like to extend his sincere and utmost gratitude to his PI and mentor, Dr. Swomitra Mohanty, for accepting him into his laboratory group and placing a significant amount of trust and responsibility on him to conduct this research and report in a timely and appropriate manner. He would also like to thank Dr. Mohanty for allowing him to present his work and learn from the work of others at the Biomedical Engineering Society (BMES) Conference in October 2016. The author would also like to thank Dr. Dhiman Bhattacharyya for his immense patience and guidance in setting up the experiments, teaching him how to use the various equipment, editing papers and proposals, and so much more. Additional thanks are given to Dr. Mano Misra for introducing the author to Dr. Mohanty, the rest of the Mohanty Laboratory, the Undergraduate Research Opportunities Program (UROP), University of Utah Honors College, and University of Utah Department of Biomedical Engineering. 35 REFERENCES 1. "Colorectal Cancer Statistics." Centers for Disease Control and Prevention. Centers for Disease Control and Prevention, 20 June 2016. Web. 18 Sept. 2016. 2. "Colorectal Cancer Statistics | World Cancer Research Fund International." Colorectal Cancer Statistics | World Cancer Research Fund International. World Cancer Research Fund International, 16 Jan. 2015. Web. 14 Sept. 2016. 3. P. Vega, F. Valentín and J. Cubiella, "Colorectal cancer diagnosis: Pitfalls and opportunities", World Journal of Gastrointestinal Oncology, vol. 7, no. 12, p. 422, 2015. 4. S. Markar, S. Chin, A. Romano, T. Wiggins, S. Antonowicz, P. Paraskeva, P. Ziprin, A. Darzi and G. Hanna, "Breath Volatile Organic Compound Profiling of Colorectal Cancer Using Selected Ion Flow-Tube Mass Spectrometry", Annals of Surgery, p. 1, 2017. 5. Altomare, " Exhaled volatile organic compounds identify patients with colorectal cancer", Br J of Surg, vol. 100, no. 7, pp. 980-980, 2013. 6. Peng, M. Hakim, Y. Broza, S. Billan, R. Abdah-Bortnyak, A. Kuten, U. Tisch and H. Haick, "Detection of lung, breast, colorectal, and prostate cancers from exhaled breath using a single array of nanosensors", Br J of Cancer, vol. 103, no. 4, pp. 542-551, 2010. 7. R. Arasaradnam, M. McFarlane, C. Ryan-Fisher, E. Westenbrink, P. Hodges, M. Thomas, S. Chambers, N. O'Connell, C. Bailey, C. Harmston, C. Nwokolo, K. Bardhan and J. Covington, "Detection of Colorectal Cancer 36 (CRC) by Urinary Volatile Organic Compound Analysis", PLoS ONE, vol. 9, no. 9, p. e108750, 2014. 8. Ahlquist, J. Skoletsky, K. Boynton, J. Harrington, D. Mahoney, W. Pierceall, S. Thibodeau and A. Shuber, "Colorectal cancer screening by detection of altered human DNA in stool: Feasibility of a multitarget assay panel", Gastroenterology, vol. 119, no. 5, pp. 1219-1227, 2000. 9. O. Kronborg, C. Fenger, J. Olsen, O. Jørgensen and O. Søndergaard, "Randomised study of screening for colorectal cancer with faecal-occultblood test", The Lancet, vol. 348, no. 9040, pp. 1467-1471, 1996. 10. Cosandey F, Skandan, Gand Singhal A 2000 Materials and processing issues in nanostructured semiconductor gas sensors JOM-e 52 1–6 11. D. Bhattacharyya, Y. Smith, M. Misra and S. Mohanty, "Electrochemical detection of methyl nicotinate biomarker using functionalized anodized titania nanotube arrays", Mat. Res. Exp., vol. 2, no. 2, p. 025002, 2015 12. Y. Smith, D. Bhattacharyya, S. Mohanty and M. Misra, "Anodic Functionalization of Titania Nanotube Arrays for the Electrochemical Detection of Tuberculosis Biomarker Vapors", Journal of The Electrochemical Society, vol. 163, no. 3, pp. B83-B89, 2015. 13. Cooper and H. Melville, "444. The kinetics of the autoxidation of n-decanal. Part I. The mechanism of reaction", J of the Chem Soc. (Resumed), p. 1984, 1951. 37 14. V. Gaur, A. Sharma and N. Verma, "Catalytic oxidation of toluene and mxylene by activated carbon fiber impregnated with transition metals", Carbon, vol. 43, no. 15, pp. 3041-3053, 2005. 15. Y. Ide, N. Kawamoto, Y. Bando, H. Hattori, M. Sadakane and T. Sano, "Ternary modified TiO2 as a simple and efficient photocatalyst for green organic synthesis", Chem. Comm, vol. 49, no. 35, p. 3652, 2013. 16. S. Yolcular and Ö. Olgun, "Ni/Al2O3 catalysts and their activity in dehydrogenation of methylcyclohexane for hydrogen production", Cat. Today, vol. 138, no. 3-4, pp. 198-202, 2008. 17. M. de Haan, R. van Gelder, A. Graser, S. Bipat and J. Stoker, "Diagnostic value of CT-colonography as compared to colonoscopy in an asymptomatic screening population: a meta-analysis", European Radiology, vol. 21, no. 8, pp. 1747-1763, 2011. 18. G. Zonta, B. Fabbri, A. Giberti, V. Guidi, N. Landini and C. Malagù, "Detection of Colorectal Cancer Biomarkers in the Presence of Interfering Gases", Procedia Engineering, vol. 87, pp. 596-599, 2014. 19. C. Malagù, B. Fabbri, S. Gherardi, A. Giberti, V. Guidi, N. Landini and G. Zonta, "Chemoresistive Gas Sensors for the Detection of Colorectal Cancer Biomarkers", Sensors, vol. 14, no. 10, pp. 18982-18992, 2014. 20. T. de Meij, I. Larbi, M. van der Schee, Y. Lentferink, T. Paff, J. Terhaar sive Droste, C. Mulder, A. van Bodegraven and N. de Boer, "Electronic nose can discriminate colorectal carcinoma and advanced adenomas by fecal volatile 38 biomarker analysis: proof of principle study", International Journal of Cancer, vol. 134, no. 5, pp. 1132-1138, 2013. 21. R. Rouseff and K. Cadwallader, Headspace analysis of foods and flavors. New York: Kluwer Academic/Plenum Publishers, 2001. 22. M. Hosseini, M. Momeni and M. Faraji, "Highly Active Nickel Nanoparticles Supported on TiO2 Nanotube Electrodes for Methanol Electrooxidation", Electroanalysis, vol. 22, no. 22, pp. 2620-2625, 2010. 23. Z. Wu, X. Huang, Z. Wang, J. Xu, H. Wang and X. Zhang, "Electrostatic Induced Stretch Growth of Homogeneous β-Ni(OH)2 on Graphene with Enhanced High-Rate Cycling for Supercapacitors", Scien. Rep., vol. 4, 2014. 24. P. Makgwane and S. Ray, "Efficient room temperature oxidation of cyclohexane over highly active hetero-mixed WO3/V2O5 oxide catalyst", Cata. Comm., vol. 54, pp. 118-123, 2014. 25. Alshammari, A. Koeckritz, V. Kalevaru, A. Bagabas and A. Martin, "Significant Formation of Adipic Acid by Direct Oxidation of Cyclohexane Using Supported Nano-Gold Catalysts", ChemCatChem, vol. 4, no. 9, pp. 1330-1336, 2012. 26. B. Sarkar, P. Prajapati, R. Tiwari, R. Tiwari, S. Ghosh, S. Shubhra Acharyya, C. Pendem, R. Kumar Singha, L. Sivakumar Konathala, J. Kumar, T. Sasaki and R. Bal, "Room temperature selective oxidation of cyclohexane over Cunanoclusters supported on nanocrystalline Cr2O3", Green Chemistry, vol. 14, no. 9, p. 2600, 2012. 39 27. Z. Li, D. Ding, Q. Liu, C. Ning and X. Wang, "Ni-doped TiO2 nanotubes for wide-range hydrogen sensing", Nanoscale Research Letters, vol. 9, no. 1, p. 118, 2014. 28. X. Long, Z. Wang, S. Wu, S. Wu, H. Lv and W. Yuan, "Production of isophthalic acid from m-xylene oxidation under the catalysis of the H3PW12O40/carbon and cobalt catalytic system", J of Ind and Eng. Chem., vol. 20, no. 1, pp. 100-107, 2014. 29. Shell Development Company, "Oxidation of xylene and toluic acid mixtures to phthalic acids", 2,723,994, 1955. 30. S. Markar, T. Wiggins, S. Antonowicz, J. Lagergren, M. Mughal and G. Hanna, "Breath volatile organic compound analysis for the diagnosis of oesophago-gastric cancer; multi-centre blinded validation clinical trial", EU J of Cancer, vol. 72, pp. S3-S4, 2017. 31. Hang, R., Liu, Y., Zhao, L., Gao, A., Bai, L., Huang, X., Zhang, X., Tang, B. and Chu, P. (2014). Fabrication of Ni-Ti-O nanotube arrays by anodization of NiTi alloy and their potential applications. Scientific Reports, 4(1), pp.1-9.