Cyclic voltammetric detection of breath-based biomarkers of colorectal cancer by nickel-functionalized Titania Nanotube Arrays

Colorectal cancer (CRC) is the third most common cancer worldwide. By 2030, it is expected that approximately 2.2 million new cases of CRC will be diagnosed globally, half of which will be fatal due to belated screenings. Many of the issues preventing early and accurate CRC diagnoses are related to limitations of the current diagnostic technologies. Colonoscopy, being the current gold-standard detection method, has its fair share of challenges: prohibitive costs, extreme invasiveness, extensive preparation, and postprocedure complications. Recent discoveries of some volatile organic compounds (VOCs), in the breaths of CRC patients, have opened up new vistas for developing an inexpensive, noninvasive and rapid diagnostic tool for early detection of CRC. The objective of the present studies was to fabricate and improve a nickel-functionalized titanium dioxide nanotube array (Ni-TNA) sensor that will be able to detect four critical breath-based biomarkers (cyclohexane, 1,3 dimethylbenzene, methylcyclohexane, and decanal) found in CRC patients. A standard anodization process was used to synthesize the TNA on to which nickel was electrodeposited in the form of Ni(OH)2 nanoparticles. Detection of the VOCs, both in solution and vapor samples, was carried out by cyclic voltammetry (CV). For gas-phase CVs, a planar electrode setup with the graphene-embedded polylactic acid (G-PLA) conducting polymeric electrolyte was used. Normal and VOC-spiked breath samples from a healthy subject were also tested with the fabricated sensor. CV measurements provided distinct electrochemical signatures for each of the four iv biomarkers in acidic solution. Gaseous VOCs could individually be detected in the presence of an inert, acidified polar solvent that may have increased conductivity. Pronounced CV features in the spiked-VOC breath samples were also observed. These results demonstrate the sensor's sensitivity to its environment and its applicability in both solution and gaseous/vapor phase detections. G-PLA performed quite well as an electrolyte and facilitated recording of the CVs in gas-phase, something that has never been done before. The research results obtained in these studies have demonstrated the potential future opportunities for the development of a portable, noninvasive, inexpensive, and versatile electrochemical-based sensor for the detection of critical organic compounds found in CRC and other cancer patients.

CYCLIC VOLTAMMETRIC DETECTION OF BREATH-BASED BIOMARKERS OF COLORECTAL CANCER BY NICKEL-FUNCTIONALIZED TITANIA NANOTUBE ARRAYS by Anurag Tripathy A thesis submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Master of Science Department of Biomedical Engineering The University of Utah May 2019 Copyright © Anurag Tripathy 2019 All Rights Reserved The University of Utah Graduate School STATEMENT OF THESIS APPROVAL The thesis of Anurag Tripathy has been approved by the following supervisory committee members: , Chair 02/28/2019 Swomitra Kumar Mohanty , Member 02/28/2019 Hamidreza S. Ghandehari , Member 02./28/2019 Jessica Ruth Kramer and by David W. Grainger the Department/College/School of Date Approved Date Approved Date Approved , Chair/Dean of Biomedical Engineering and by David B. Kieda, Dean of The Graduate School. ABSTRACT Colorectal cancer (CRC) is the third most common cancer worldwide. By 2030, it is expected that approximately 2.2 million new cases of CRC will be diagnosed globally, half of which will be fatal due to belated screenings. Many of the issues preventing early and accurate CRC diagnoses are related to limitations of the current diagnostic technologies. Colonoscopy, being the current gold-standard detection method, has its fair share of challenges: prohibitive costs, extreme invasiveness, extensive preparation, and postprocedure complications. Recent discoveries of some volatile organic compounds (VOCs), in the breaths of CRC patients, have opened up new vistas for developing an inexpensive, noninvasive and rapid diagnostic tool for early detection of CRC. The objective of the present studies was to fabricate and improve a nickel-functionalized titanium dioxide nanotube array (Ni-TNA) sensor that will be able to detect four critical breath-based biomarkers (cyclohexane, 1,3 dimethylbenzene, methylcyclohexane, and decanal) found in CRC patients. A standard anodization process was used to synthesize the TNA on to which nickel was electrodeposited in the form of Ni(OH)2 nanoparticles. Detection of the VOCs, both in solution and vapor samples, was carried out by cyclic voltammetry (CV). For gas-phase CVs, a planar electrode setup with the grapheneembedded polylactic acid (G-PLA) conducting polymeric electrolyte was used. Normal and VOC-spiked breath samples from a healthy subject were also tested with the fabricated sensor. CV measurements provided distinct electrochemical signatures for each of the four biomarkers in acidic solution. Gaseous VOCs could individually be detected in the presence of an inert, acidified polar solvent that may have increased conductivity. Pronounced CV features in the spiked-VOC breath samples were also observed. These results demonstrate the sensor’s sensitivity to its environment and its applicability in both solution and gaseous/vapor phase detections. G-PLA performed quite well as an electrolyte and facilitated recording of the CVs in gas-phase, something that has never been done before. The research results obtained in these studies have demonstrated the potential future opportunities for the development of a portable, noninvasive, inexpensive, and versatile electrochemical-based sensor for the detection of critical organic compounds found in CRC and other cancer patients. iv This work is dedicated to my parents Prabhat Kumar Tripathy and Debajani Tripathy who have always given me an extraordinary amount of support and love and taught me how to be a good human being, above all else. A special feeling of respect and love for my grandfather Braja Kishore Panda who showed me what pursuing excellence and being a good human being mean. “The will to win, the desire to succeed, the urge to reach your full potential… these are the keys that will unlock the door to personal excellence.” - Confucius TABLE OF CONTENTS ABSTRACT ....................................................................................................................... iii ACKNOWLEDGMENTS ................................................................................................. ix Chapters 1. INTRODUCTION AND BACKGROUND ................................................................... 1 1.1 Current CRC diagnostic space and challenges .................................................... 3 1.1.1 Technological standpoint ......................................................................... 3 1.1.2 Patient-related factors .............................................................................. 4 1.1.3 Practitioner-related factors ....................................................................... 4 1.2 A new diagnostic approach .................................................................................. 5 1.2.1 Versatility and value of VOCs ................................................................. 5 1.2.2 Current VOC detection techniques and limitations ................................. 7 1.2.3 Metal-oxide sensors and their capabilities ............................................... 9 1.3 Research objectives ............................................................................................ 11 2. SYNTHESIS AND OPTIMIZATION OF NICKEL FUNCTIONALIZED TiO2 NANOTUBE ARRAY...................................................................................................... 13 2.1 An overview of TNA- based sensors and the need for nickel functionalization ...................................................................................................... 13 2.2 Experimental ...................................................................................................... 15 2.2.1 TiO2 nanotube array (TNA) synthesis ................................................... 15 2.2.2 Nickel electrodeposition on TNA .......................................................... 16 2.2.3 Characterization of Ni-TNA sensor ....................................................... 16 2.3 Results ................................................................................................................ 17 2.3.1 TNA surface characteristics ................................................................... 17 2.3.2 Ni-TNA surface morphology and characteristics .................................. 18 2.4 Discussion .......................................................................................................... 26 3. SOLUTION-PHASE CYCLIC VOLTAMMETRIC DETECTION OF THE SELECTED VOCs............................................................................................................ 29 3.1 Construction of cyclic voltammetry setups ........................................................ 29 3.2 Cyclic voltammetry of exhaled breath condensate ............................................ 30 3.3 Experimental ...................................................................................................... 31 3.4 Results ................................................................................................................ 33 3.4.1 Neutral solutions .................................................................................... 33 3.4.2 Basic pH experiments ............................................................................ 35 3.4.3 Acidic pH experiments .......................................................................... 38 3.5 Discussion .......................................................................................................... 43 3.5.1 Electrocatalytic activity of Ni(OH)2 in the detection of the VOCs ....... 43 3.5.2 Neutral solutions .................................................................................... 44 3.5.3 Basic pH behavior .................................................................................. 45 3.5.4 Acidic pH behavior ................................................................................ 49 3.5.5 Effect of acidic pH on sensor morphology and performance ................ 51 4. GAS-PHASE AND BREATH-BASED CYCLIC VOLTAMMETRIC DETECTION OF THE VOCS ................................................................................................................. 53 4.1 Current trends in the use of conductive polymers in electrochemical gas detection ................................................................................................................... 53 4.2 Experimental ...................................................................................................... 55 4.2.1 Gas-phase VOC detection setup ............................................................ 55 4.2.2 VOC detection under gaseous conditions .............................................. 55 4.2.3 VOC detection in breath samples .......................................................... 57 4.3 Results ................................................................................................................ 57 4.3.1 Gas-phase VOC detection ...................................................................... 57 4.3.2 Breath-based VOC detection ................................................................. 62 4.4 Discussion .......................................................................................................... 66 4.4.1 Pure VOC conditions ............................................................................. 66 4.4.2 VOCs dissolved in acetone .................................................................... 67 4.4.3 VOC detection in breath samples .......................................................... 69 5. LIMITATIONS AND FUTURE WORK ..................................................................... 70 5.1 Limitations ......................................................................................................... 70 5.2 Future work ........................................................................................................ 74 6. CONCLUSION…………………………………………………………….………….76 REFERENCES ................................................................................................................. 81 viii ACKNOWLEDGMENTS I want to extend my sincerest and heartfelt gratitude to Dr. Swomitra Mohanty for providing me with such a wonderful opportunity to work in his laboratory and on this project. I greatly appreciated the autonomy he gave to me and all the resources and tools he provided me to produce the results of this research. I also wish to recognize the contributions of Dr. Dhiman Bhattacharyya who helped me immensely when I first joined the laboratory. Special thanks are also due to Yalda Saffary and Christina Willis who were willing to listen to my ideas and helped me realize some of the research goals I had planned for this project while pursuing their doctoral dissertations. To the other members of my committee, Dr. Jessica Kramer and Dr. Hamid Ghandehari, I wish to express my respect and gratitude for their willingness to be on my committee and exhibiting a tremendous amount of patience and providing valuable inputs. I also want to recognize the entire faculty, staff, colleagues, and friends in the Biomedical Engineering Department for helping me one way or another during my undergraduate and master’s research. Lastly, I want to specifically express my love and appreciation towards my parents, especially to my father whose expertise in electrochemistry and years of experience kept both my research and sanity on track. I want to thank them for their endless support and constant faith in my abilities as I carried out my research work. CHAPTER 1 INTRODUCTION AND BACKGROUND Colorectal cancer (CRC) is a disease that progresses due to an unchecked division and subsequent proliferation of abnormal cells within the large intestine of the gastrointestinal tract (Fig. 1.1). CRC typically arises from a noncancerous growth (polyp) within the inner linings of the colon/rectum and grows over a 10 to 20 year period [1]. The most common polyps are adenomatous polyps—also known as adenomas—and it is speculated that approximately one-third to one-half of all individuals, globally, will develop these [1]. It is the second leading cause of cancer death in women and third for men [1]. In fact, CRC was reported to be the third most common cancer worldwide and the fourth deadliest cancer malignancy, accounting for approximately 700,000 deaths in 2012 [2], [3]. A 60% increase in CRC’s global burden is expected by 2030 with approximately 2.2 million new cases and 1.1 million deaths [2]. Though the survival rate for early diagnosed CRC, before metastasis, is above 90% [3], [4], only 40% of CRC cases are detected early; advanced-stage survival rates, due to late detection, are significantly lower [3]. The most common treatment for CRC is surgery, but other methods such as chemotherapy, radiation therapy, and ablation are also commonly used [1]. However, the chances of complete recovery or remission significantly depend on how early the cancer was diagnosed and/or treated [1], [3]. 2 Figure 1.1 Graphic is showing the formation of adenomatous polyps in colorectal cancer over various stages of the disease [5]. The economic burden due to CRC is significant. In the United States, the Center for Disease Control and Prevention (CDC) reports that the clinical diagnostic costs for CRC range between $304-1,150, nonclinical costs are around $1,000, and extraneous costs can be approximately $450, per person [6]. The financial strain and associated burden due to CRC necessitate the need to pursue research and development for early detection and follow-up treatment at affordable prices. Development of a reasonably simple, effective and versatile diagnostic tool to detect the cancer, fairly accurately, at early stages will go a long way to mitigate the suffering of patients and decrease the number of colonoscopies required. 3 1.1 Current CRC diagnostic space and challenges Early CRC diagnosis of symptomatic patients is a multifactorial problem that has three main components: 1) limitations in available diagnostic techniques; 2) patient delay factors; and 3) physician delay factors [4]. Consequently, it is speculated that alleviation or elimination of some of the issues may contribute to an increase in accurate and timely detection and treatment [1]-[4]. 1.1.1 Technological standpoint Available diagnostic methods for CRC can be divided into two categories: visual examinations and stool tests [1]. The first category consists of techniques such as colonoscopy, computed tomographic colonography (CTC), double-contrast barium enema, and flexible sigmoidoscopy [1]-[4]. Stool tests usually involve fecal immunochemical test (FIT), FIT-DNA test, and high-sensitivity guaiac-based fecal occult blood test (gFOBT) [1]-[4]. While each of these tests presents its own merits and drawbacks in terms of efficacy, complication, and cost, the gold-standard detection method is colonoscopy [1], [2], [4]. In fact, any aberrant results from the other tests eventually require a confirmatory colonoscopy diagnosis [1]. Colonoscopy is a direct visualization tool that allows for a complete examination of the colon and rectum including biopsies, and even removal of polyps in some cases [1], [3], [4]. However, its benefits come with severe drawbacks including complete bowel cleaning with laxatives, required sedation, forced absences from work for patients, exorbitant costs, and the highest risk of bowel tears and/or infections than any other method [1], [4]. In other words, while colonoscopy has the highest performance capability, it poses the highest complexity in terms of patient preparation, 4 inconvenience, facilities/equipment required, and general patient discomfort. 1.1.2 Patient-related factors The biggest issues confronting CRC patients are a lack of knowledge and concern regarding the potential risks associated with the seriousness of CRC symptoms [4]. This often leads to misattribution of CRC as a benign condition, and a “wait and see” policy is often used by patients [4]. Better noninvasive diagnostic methods with fast turnaround times and accuracy are expected to combat some of these behavioral trends [4]. Other contributing factors include fear/denial of cancer, embarrassing and/or unpleasant experiences, which have also motivated a lack of patient initiative in screenings until the symptoms become quite severe [4]. It is also speculated that low socioeconomic status and education levels may be contributing but it has been difficult to examine their impacts thus far [4], [6]. 1.1.3 Practitioner-related factors Further complications pertaining to effective early CRC diagnosis are ones created and/or faced by physicians. Several studies have reported that delays in diagnosis caused by physicians manifest due to initial misdiagnoses or attribution of symptoms to different benign conditions [4]. Correspondingly, there is a significant 200-day delay (on an average) between the first physician-patient encounter and an endoscopic referral [4]. A commonly used diagnostic biomarker of iron-deficiency anemia by physicians extends the delay to more than 300 days [4]. Compounding these issues is an extremely common absence of physical examinations by physicians on patients facing lower abdominal 5 symptoms [4]. While some of these factors could be attributed to improper physician training, technological limitations of nonspecific tests and false positives further add to these delays [1], [4]. 1.2 A new diagnostic approach Many aforementioned limitations and complications can be attributed to the weaknesses of current diagnostic methodologies, which have decreased patient compliance and made jobs of physicians more difficult [3], [4]. Consequently, the development of a novel detection method is imperative not only for early and accurate CRC diagnosis but also to alleviate physician concerns and increase patient compliance. Recent research has indicated that the detection of volatile organic compounds (VOCs) is promising for CRC diagnosis [3], [7]. 1.2.1 Versatility and value of VOCs VOCs are gaseous molecules (under standard temperature and pressure conditions) that have been identified in a plethora of biological samples including blood, urine, stool, and breath [3], [8]–[21]. It is believed that abnormal genomic and metabolic processes caused by diseased states facilitate the production of these VOCs, which then proliferate throughout these biological samples [3], [12], [19]. These VOCs are typically not seen in healthy people. Indeed, a vast array of diseases ranging from diabetes and cancer to tuberculosis have been recorded to produce a variety of VOCs corresponding to each of these diseases [22]–[24]. In cancer, specifically, the formation of VOCs has been attributed to the peroxygenation of different cellular membrane components [12], [19]. The VOCs 6 are subsequently released into the bloodstream, which then get excreted from the body in several ways. In the case of CRC, distinctive VOC patterns have been observed in all four of the biological samples listed above. Interestingly, these patterns vary based on the nature of the sample. For instance, Westenbrink et al. [21], [25] and Meiji et al. [16] have used electronic nose (e-nose) and gas chromatography-mass spectrometry (GC-MS) techniques to isolate VOCs in patients’ urine. Wang et al. [15] identified four unique VOCs in the blood of CRC patients while Kornborg et al. [26] showed unique VOCs in stool samples. Though fluid samples have been extensively studied and show great promise as diagnostic targets, arguably, the most useful test may be exhaled breaths (EB) of CRC patients [3], [7], [12], [16], [22]. The advantages of EB are multifaceted. Compared to most fluid/stool samples, VOC detection in EB can be easier due to its relatively simple composition [22]. A typical EB is a mixture of nitrogen (N2), oxygen (O2), carbon dioxide (CO2), and inert gases along with water vapor [22]. Although confounding factors, such as ambient environment and diet, can increase the complexity of breath sampling, it offers the highest potential as an inexpensive, rapid and noninvasive diagnostic tool [3], [8], [12], [16], [22]. A few studies, thus far, have discussed potential breath biomarkers of CRC. Amal et al. [3] listed four VOC biomarkers in different concentrations in healthy controls and CRC patients. Markar et al. [27] listed a single, specific VOC, propanal, as a CRC breath biomarker with a distinguishing capability of 96% between healthy people and CRC patients (cancerous vs. noncancerous). Peng et al. [20] also listed six different CRC breath VOCs. Amongst all of these VOCs, the ones chosen for this study come from the work of Altomare et al. [12] who presented fifteen unique breath VOCs, out of which four were selected. These four VOCs include cyclohexane (CH), 1,3 dimethylbenzene (DMBZ), 7 methylcyclohexane (MCH), and decanal (DEC). The choice was based on two criteria: 1) all four VOCs have a distinguishing capability greater than 92%; and 2) the pattern of multiple VOCs is more important than the presence of any single VOC [12]. It is important to note that the only 1,3 dimethylbenzene was a common VOC found in both the Peng et al. [20] and Altomare et al. [12] studies. Altomare et al. [12] also reported that their reported VOCs did not vary with sex, age or stage of cancer, which made them ideal targets for detection in the present study. Though each of these studies presents a vast array of possible VOC targets, there were only a handful of methodologies used to identify them. 1.2.2 Current VOC detection techniques and limitations A study of the published literature indicates that GC-MS technology has been, by far, the most commonly used technique for the analysis and quantification of EB VOCs of various diseases [22]. This technique is perhaps the most comprehensive and sensitive method to detect VOCs down to parts-per-trillion/parts-per-billion (ppt/ppb) concentrations [3], [7], [8], [22], [28]. Several derivatives of GC technology have been developed and implemented over the years that include GC coupled with ion mobility spectrometry (GC-IMS), GC coupled with flame ionization detection (GC-FID), etc. in addition to GC-MS [22], [27]. Each of these techniques has their fair share of advantages [22], [27]. However, a significant number of inherent drawbacks make GC not a clinically favored technique. GC analysis requires indirect sampling and sample pretreatment, making it a time-consuming process that can lead to analyte loss/degradation and/or potential contamination [22]. It also requires expensive equipment, which can be unaffordable for many point-of-care settings (POC) [22], [28]. For these reasons, intense 8 research activities are being pursued to develop suitable real-time VOC analytical techniques. Thus far, the techniques capable of performing real-time analysis include proton transfer reaction mass spectrometry (PTR), proton transfer reaction-time-of-flight-mass spectrometry (PTR-TOF-MS), IMS coupled with multicapillary columns (MCC/IMS), selected ion flow tube mass spectrometry (SIFT-MS), etc. [22]. Laser spectroscopic techniques, also known as electronic noses (e-noses), have also been applied to EB VOC detection [18], [22]. Some commercial products with innovative technical features have also been developed. Owlstone, a company based on the University of Cambridge, has successfully marketed a “Field Asymmetric Ion Mobility Spectrometry (FAIMS)” chipbased chemical sensor that detects VOC ions as they pass through channels across an oscillating electric field [29], [30]. FAIMS technology has also been utilized in the products of a Canadian company, Breathtec Biomedical, who have refined the method and developed sensitive, cheap, and portable devices for a variety of diseases [29]. Fossil Ion Tech developed their iteration of a mass spectrometer called SUPER SESI (short for Secondary Electro-Spray Ionization) that ionizes VOCs molecules with a cloud of ions, at high temperatures [29]. New England Breath Technologies (NEBT) developed a breathalyzer measuring blood sugar levels by registering the concentration of acetone in patients’ breath, a compound with linear correlation to blood sugar concentration [29]. Fujitsu Laboratories Ltd. synthesized an ammonia sensor using a copper (I) bromide film that could undergo reversible adsorption with the ammonia molecules [31]. Ammonia adsorption/desorption effectively reduces the electrical resistance between electrodes, thereby leading to its detection [31]. A tuberculosis (TB) breathalyzer developed by Rapid 9 Biosensors analyzes a breath VOCs for TB by carrying out a displacement assay using evanescent wave technology [32]. In this sensing technique, a biochemical coating loaded with the sample and fluorescent analogs is exposed to TB antigens, which displace the latter and bond to the antibodies strongly, resulting in a decrease in the magnitude of the fluorescent signal when excited by a laser [32]. These real-time techniques address many of the limitations of the GC based methods by way of reducing cumbersome sample preparation/storage steps and providing comparably high-sensitivity and selectivity in detection [18], [22]. While these properties make them quite favorable for POC settings, they still suffer from major limitations including exorbitant costs, requiring highly specialized and skilled operators, and provide a narrow analysis of the chemical variation in EB VOCs [22], [28]. Additionally, polymeric e-nose technology suffers from issues 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 [18]. Consequently, the quest for a novel, noninvasive screening system, specific to breath-based CRC VOCs that can provide GC-MS resolution in real-time and can be applied in POC settings, is in progress and assumes increasing importance. It appears that the electrochemical detection of VOCs using a metal-oxide sensor (MOS) may perhaps be the most promising way forward to address some of the existing limitations. 1.2.3 Metal-oxide sensors and their capabilities The term “metal-oxide” encompasses a wide assortment of materials ranging from metals to semiconductors and insulators [33], [34]. They possess a dynamic range of electronic, chemical, and physical properties that are highly sensitive to their chemical 10 environments [34]. Within the paradigm of gas sensing, the electrical conductivity of semiconductors varies depending on a specific gas/chemical composition [34]. Historically, MOS have been used for gas sensing applications via surface adsorption and desorption of gases [33], [34]. These processes are carried out by electron transfers on film surfaces [34]. One dimensional (1-D) nanostructure metal oxides (e.g., metal oxide nanotubes, nanowires, etc.) exhibit a high surface area-to-volume ratio and are widely favored for gas sensing applications due to their relatively high chemical and thermal stabilities [33], [35], [36]. In the case of CRC, a few studies have reported the use of MOS to detect VOCs found in the stool samples of CRC patients, such as benzene, decanal and 1-iodononane [35], [36]. Zonta et al. [35] designed oxide films of titanium, tantalum, and vanadium to detect benzene. A combined titanium and tin oxide film provided enhanced detection of decanal and 1-iodononane. Bhattacharyya et al. [8] designed a nickel functionalized TiO2 nanotube array that exhibited detection of the four breath-based VOCs chosen in the present study. While these results demonstrated successful detection of the CRC VOCs, they did so under somewhat unfavorable conditions. For many MOS (excluding the one synthesized by Bhattacharyya et al. [8]), gas detection is facilitated at elevated temperatures [33]–[35]. For instance, the sensor developed by Zonta et al. [35] required temperatures up to 650oC to detect the VOCs. Elevated temperature is necessary for converting atmospheric (O2), adsorbed onto the MOS surface, into the superoxide anion (O2-) that then interacts with the gases [8], [28], [33]. 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 11 power and elaborate reactor/sensor designs making the application of these MOS in POC settings highly inefficient [22]. Bhattacharyya et al. [8] attempted to circumnavigate these issues by using nickel deposits on their sensor to increase specificity and used amperometry, an electrochemical technique used to monitor a gain (reduction) or loss (oxidation) of electrons in the presence of a fixed potential by measuring changes in current. However, their sensor does not operate like a typical MOS because there is no heating involved, and the electrochemical method does not clarify how the reactions are proceeding. Furthermore, their method does not enhance the specificity of the sensor since it also showed reactivity with acetone (AC) and ethanol (EtOH) vapors. Nevertheless, the highly tunable and stable properties of MOS combined with a suitable electrochemical technique can provide enhanced detection of breath-based VOCs found in CRC patients. 1.3 Research objectives The overarching goal of the present research is to design a novel and inexpensive, metal oxide-based sensor that can meet the following two specific objectives: 1) Demonstration of high sensitivity and selective detection of the target VOCs in a relatively short timeframe, both under solution and gaseous conditions; 2) The capability of integration into a portable, noninvasive, hand-held device deployable in POC settings. Using the work reported by Bhattacharyya et al. [8], [28] as the starting point, the specific focus of the present research is to enhance the morphological and structural features/properties of the nickel functionalized TiO2 sensor for the electrochemical detection of the VOCs. The titanium dioxide (TiO2) nanotube array (TNA) system was 12 chosen for a combination of attractive properties (robust mechanical strength, fairly high corrosion resistance, favorable charge transfer, high surface area, and reasonable cost) [8], [33], [34]. Nickel deposition on its surface is expected to enhance the detection of the VOCs, which will be carried out via cyclic voltammetry (CV) [8]. Cyclic voltammetry, being a powerful transient electrochemical technique, is extensively used to investigate the oxidation-reduction behavior of electrochemical reactions/couples [37]. In this technique, cyclic voltammograms (CVMs) are recorded to study the current-voltage characteristics. These patterns identify possible redox reactions, distinctive to the electroactive species being investigated [37]. Reactions of the VOCs with the nickel-TNA (Ni-TNA) sensor are expected to elicit unique CVM trends and patterns that will confirm the detection of the VOCs. To verify the validity of the sensing methodology and platform, possible reaction mechanisms between the sensor and the VOCs will be examined, and the sensor will be exposed to breath samples from a healthy subject. The present thesis has been structured into six chapters: (1) Introduction; (2) Synthesis of Ni-TNA sensors; (3) Solution-phase cyclic voltammetric detection of VOCs; (4) Gas-phase cyclic voltammetric detection of VOCs; (5) Limitations and Future Work; and (6) Conclusion. CHAPTER 2 SYNTHESIS AND OPTIMIZATION OF NICKEL FUNCTIONALIZED TiO2 NANOTUBE ARRAY 2.1 An overview of TNA- based sensors and the need for nickel functionalization TiO2 nanotube arrays (TNAs) are highly ordered, one-dimensional semiconductors that possess a high surface area-to-volume ratio enabling efficient charge transfer reactions [28]. Their unique electrical, optical, and thermal properties have rendered them a favorable metal oxide substrate with wide-ranging applications: hydrogen gas generation, supercapacitors, solar cells so on and so forth [33], [34]. Anodized TNAs have been widely used in various sensing applications including gas/vapor phase systems [28], [33], [38], [39]. For instance, TNAs were used for both room and elevated temperature hydrogen gas sensing [40], [41]. Previous studies have also demonstrated reactivity with gases such as ethanol, formaldehyde, carbon monoxide, acetone, and VOCs, and in many of these cases, TNAs outperformed several other MOS systems [33], [38], [39], [41]. It is evident that TNAs synthesized via electrochemical anodization of titanium foils offer a mechanically robust and electrically active substrate that can not only undergo a variety of modifications but also be instrumental in facilitating the detection of VOCs [8], [28], [39]. It is important to highlight here that although plain/unmodified TNAs do have the ability to detect certain 14 VOCs, there have been many instances of poor sensitivity, meager response magnitude and sluggish response [39]. In our cyclic voltammetry studies (not shown) as well as those of Bhattacharyya et al. [8], [28], unfunctionalized TNAs have not demonstrated any observable interactions with the selected VOCs used in the present study. In contrast, upon addition of a suitable electroactive functional group such as cobalt, gold, and nickel enhanced reactivity (by amperometric detection method) could be observed [8], [28]. Consequently, nickel functionalization of TNA was chosen in the present study. As mentioned previously, the decision to use nickel to detect the four breath-based CRC VOCs was based on its successful application, as has been reported by Bhattacharyya et al. [8] as well as in various other industrial catalytical systems. For instance, the oxidation of cyclohexane (CH) and other hydrocarbons for the production of adipic acid (important for Nylon-6 and Nylon-66 synthesis) has been achieved with nickel doped TiO2 under photocatalytic conditions [42]. In the work of Gaur et al. [43], nickel impregnated carbon fibers were shown to exhibit catalytic oxidation of 1,3 dimethylbenzene (DMBZ). The Ni/Al2O3 catalysts synthesized by Yolcular et al. [44] showed dehydrogenation of methylcyclohexane (MCH). Cooper et al. [45] showed nickel-based oxidation reactions with decanal. Further evidence in support of the efficacy of nickel-based catalytic systems came from the work of Hosseini et al. [46] who electrodeposited nickel nanoparticles on a TiO2 substrate to carry out methanol oxidation. Li et al. [41] used nickel doped TiO2 substrates to detect VOCs in a hydrogen environment. Though nickel has successfully demonstrated both catalytic and sensing capabilities, detailed information about its morphological features that could enhance the latter functionality are only sketchy and hence require further investigations. Indeed, the 15 surface properties of these sensors are pivotal for the detection of VOCs [28], [33], [34]. For these reasons, optimization of nickel nanoparticle morphology on the TNA surface was carried out by modifying two specific electrodeposition parameters: 1) nickel concentration in the electrolyte; and 2) applied cathode current density. Although other factors such as pH, temperature and stir rate may influence the morphology of these deposits [37], analyses of their effects on sensor response were beyond the scope of the present study. 2.2 Experimental 2.2.1 TiO2 nanotube array (TNA) synthesis The TiO2 nanotubes were grown on titanium foils using the method described by Bhattacharyya et al. [8], [28]. The foils (GI grade 0.009”x12”x25”, ESPI Metals, 99.99%) were cut into 1.5x1.5cm coupons. One side of the coupons was mechanically polished with a fine-grade emery paper, while the other side was covered with a Kapton tape to ensure that the TNA grows on one side only. For anodization, the coupons were first placed in a 50-50 vol.% solution of ethanol and acetone and degreased via ultrasonication for 25 mins in a Branson 5510 ultrasonicator. Anodization was performed in a mixture of ethylene glycol (Fisher Scientific Grade), ammonium fluoride (Acros Chemicals, 98+%, extra pure), and deionized (DI) water solution according to a specific recipe (wt.%): ethylene glycol96.5; ammonium fluoride- 0.5; DI H2O- 3. A two-electrode setup with Ti coupon as the anode and platinum foil as the cathode was used for anodization. A constant 30V (DC Power Supply HY3005F-3LH) was applied for 60 mins to the coupons in a Teflon beaker with the solution being stirred at 60 rpm. Following anodization, the coupons were ultrasonically cleaned in DI H2O for ~5s and subsequently thoroughly rinsed with 16 isopropanol. The coupons were then placed in an oven (VWR Symphony) at 110oC overnight for drying. Finally, annealing of the dried coupons was carried out under oxygen flow (4.5L/min) at 500oC for 2h with ramp rate 3oC/min in a furnace (Thermo Scientific Lindberg Blue M Model). 2.2.2 Nickel electrodeposition on TNA Deposition of nickel nanoparticles on the annealed TNA surface was carried out in an aqueous 1.5M NiCl2 solution (NiCl2-6H2O Fisher Chemicals) [46], [47]. A twoelectrode electrochemical cell with Pt foil as the anode and TNA coupon as the cathode was used to effect nickel electrodeposition. To determine the effect of current density on nickel deposition, three constant currents magnitudes (11mA, 22mA, and 44mA, corresponding to 2.5 to 10mA/cm2 current densities, respectively) were applied to the coupons for 1 min (BK Precision 9110 power supply). During the electrodeposition runs, an anode-to-cathode (ACD) distance of ~2.5cm was maintained. After electrodeposition, the Ni-TNA coupons were thoroughly rinsed thoroughly with DI H2O, dried overnight at 110oC, and stored in a desiccator to mitigate any potential moisture damage. 2.2.3 Characterization of Ni-TNA sensor 2.2.3.1 Scanning electron microscope (SEM) Morphological features of the unfunctionalized annealed TNA and the Ni-TNA samples were examined under a Hitachi S-4800 scanning electron microscope (SEM) with an Energy-dispersive X-ray spectroscopy (EDS) attachment (Oxford make)+. EDS analysis was used alongside SEM images to approximate element distributions on coupon 17 surfaces. 2.2.3.2 X-ray photoelectron spectroscopy (XPS) Surface composition of the Ni-TNA sensors was analyzed by the XPS (Kratos Axis, Ulta DLD model, Utah Core Facility). 2.2.3.3 X-ray powder diffraction (XRD) A Rigaku Miniflex XRD (CuKα = 1.54059o) was used to record the X-ray diffraction (XRD) patterns of the sample from in the 2θ range from 10 to 80o with a step size of 0.015o and dwell time of 1o/min. Diffraction patterns were subsequently analyzed using Rigaku PDXL2 software and indexed using the standard JCPDS cards. 2.3 Results 2.3.1 TNA surface characteristics Scanning electron microphotographs of the TNA coupons (Fig. 2.1) showed morphological properties that corresponded well with ones reported by Bhattacharyya et al. [8]. The structure of the TNA revealed a highly ordered and regular-sized arrangement with tubes having diameters in the range of 50-75nm and wall thicknesses of 10-20nm. The length of the nanotubes was in the order of 1.3-1.7μm, and as expected, the highly resistive TNA films consisted of mostly anatase (minor levels of rutile) TiO2, created after annealing in oxygen. 18 Figure 2.1 SEM images of the unfunctionalized TNA: (a) Top side view of the TNA; (b) Tubular view of the TNA; (c) Lateral cross-section of the TNA at a scratched surface. 2.3.2 Ni-TNA surface morphology and characteristics The electrodeposition of nickel was performed at three different current densities to examine their effects on the morphology of electrodeposited nickel. In all the three cases, SEM images (Fig. 2.2) showed aggregated globules of nickel nanoparticles on the TNA surface, similar to those seen by Bhattacharyya et al. [8] in their nickel sensors. Though individual particulates were present on the tubes, the greater proportion of nickel was globular. The diameter of the nickel globules was in the range of 120-200nm, prepared at all the three current densities. However, current density influenced the amount of nickel deposited on the surface. EDS studies (not shown) showed the following wt. % of nickel on the surfaces: 8-13 at 2.5mA/cm2; 14-23 at 5mA/cm2; 24-33 at 10mA/cm2, respectively. 19 Figure 2.2 SEM images of the unfunctionalized TNA and Ni deposited TNA at various current densities: (a) Top view of the unfunctionalized TNA; (b) 2.5mA/cm2; (c) 5.0mA/cm2; (d) 10mA/cm2. Nickel deposition at 10mA/cm2 was further profiled in different regions. EDS studies showed a gradient in nickel deposition with the top one-third (Fig. 2.3), the middle third (Fig. 2.4) and bottom third (Fig 2.5) having 25-35 wt. %, 40-60 wt. %, and 65-85 wt. % nickel, respectively. A somewhat different observation sometimes made only at the top one-third of the sensors was the formation of a thin nickel film encompassing the nickel globules. Despite the presence of relatively higher concentrations of nickel in the middle and bottom third sections, the formation of no such thin films was seen. 20 Figure 2.3 SEM-EDS images of the top one-third of 10mA/cm2 Ni-TNA sensor: (a) SEM image of the Ni deposition; (b) EDS image of Ti, Ni, and O on the surface; (c) EDS showing Ni deposition on the surface exclusively; (d) Calculated wt. % distribution of elements on the Ni-TNA surface. Note that the 1.9% Cl contamination comes from electrodeposition process seen in some sensors. 21 Figure 2.4 SEM-EDS images of the middle one-third of 10mA/cm2 Ni-TNA sensor: (a) SEM image of the Ni deposition; (b) EDS image of Ti, Ni, and O on the surface; (c) EDS showing Ni deposition on the surface exclusively; (d) Calculated wt. % distribution of elements on the Ni-TNA surface. Note the lack of Cl in this spectrum and the increase in Ni wt. %. 22 Figure 2.5 SEM-EDS images of the bottom one-third of 10mA/cm2 Ni-TNA sensor: (a) SEM image of the Ni deposition; (b) EDS image of Ti, Ni, and O on the surface; (c) EDS showing Ni deposition on the surface exclusively; (d) Calculated wt. % distribution of elements on the Ni-TNA surface. The XPS spectrum of the Ni-TNA surface (Figs. 2.6 and 2.7) revealed a multitude of peaks across the entire spectrum, notably due to both nickel and oxygen. The 1s peak of oxygen is seen at 532eV (Fig. 2.6), while nickel was observed to exhibit prominent Ni 2p peaks at 855.6eV (Fig. 2.7). These values have been reported by both Bhattacharyya et al. [8] and Wu et al. [48] to be due to Ni(OH)2, matching the curve-fitting results. Minor Ni metal peak at 852.6eV was also seen and can be attributed to the electrodeposition process, which forms reduced nickel metal along with Ni(OH)2. Ni3s and Ni3p peaks were observed at the lower end of the XPS spectrum, and their low energy values and miniscule peak heights perhaps indicated their insignificant role in globule formation. 23 Figure 2.6 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 2s, Ni2p, and O1s peaks are visible indicating the formation of Ni(OH)2 on the Ni-TNA surface. 24 Figure 2.7 Hi-res XPS spectrum is showing only prominent curve-fitted nickel peaks (including satellite peaks) from the nickel globules on the surface of the Ni-TNA. Relative percentages of Ni metal and Ni(OH)2 have also been calculated at the different eV. 25 XPS data corroborated well with the XRD profile (Fig. 2.8) of the Ni-TNA surface, which showed two distinct crystalline states of Ni(OH)2: α-Ni(OH)2 and β-Ni(OH)2. Between these two, the latter is significantly more prominent on the surface. XRD pattern also substantiated the presence of the anatase phase of the TNA (over the rutile phase), which was expected due to the annealing process. Interestingly, little-to-no metallic nickel formed on the surface, which indicated that the nickel deposition was dominated by Ni(OH)2 formation. The overlapped Ti and Ni peaks at 53o and 77o indicate their distinct presence (i.e., Ni is not integrated into the TNA lattice). For all the subsequent biomarker detection tests, Ni-TNA sensors fabricated at a cathodic current density of 10mA/cm2 were used due to the formation of relatively higher wt.% of nickel on the TNA surface. Figure 2.8 XRD profile of the Ni-TNA sensor’s surface showing the crystalline phases of both titanium and nickel. 26 2.4 Discussion In order to analyze the conditional and varied response of the sensor (observed in the CVMs), it is essential first to understand how the electrodeposited nickel nanoparticles were formed. Scanning electron microphotographs, together with the EDS measurements and XRD profiles indicated that nickel deposition on the TNA surface took place via the formation of Ni(OH)2 nanoparticles. This observation is also supported by the work of Bhattacharyya et al. [8] who proposed that the formation of Ni(OH)2 could be attributed to the hydrolysis of NiCl2 whose thermodynamic barrier was overcome by the application of an appropriate current density [46]. Equation 1 shows the reaction mechanism proposed by them. The present study has indicated that a current density of 10mA/cm2 vis-à-vis 5mA/cm2 (as proposed by Bhattacharyya et al. [8]) resulted in the deposition of higher amounts of nickel on the TNA surface. NiCl2 + 2H2O → Ni(OH)2 + 2HCl (1) The hydrolysis of NiCl2 was further enhanced upon carrying out the electrodeposition at pH 7 in a high-molarity NiCl2 solution. According to Cheshideh et al. [47], nickel electrodeposition involves two competing reactions: hydrogen evolution and nickel deposition. At low pH, hydrogen evolution is significantly enhanced, and electrochemically generated H2 bubbles adsorb onto the electrode surface instead of bubbling out of the electrolyte [47]. Consequently, they occupy suitable sites intended for nickel deposition and therefore force nickel nucleation at unoccupied spaces between the bubbles, creating large nanoparticles placed sporadically [47]. However, at high pH, 27 hydrogen evolution is significantly reduced making more sites available for nickel nucleation resulting in smaller-sized particles arranged more uniformly [47]. Nickel nucleation was further enhanced by the high nickel concentration of the electrolyte. Yao et al. [49] reported that at higher concentrations, mass transfer of the electroactive component at the cathode becomes considerably faster, which leads to increased nucleation of the nanoparticles. Significant retention of nuclei on the substrate restrains the overall growth of the nanoparticles, therefore, reducing their size [49]. Correspondingly, we observed that the combination of pH 7 and high Ni2+ concentration (1.5M NiCl2 solution) generated smaller nanoparticles that were distributed more uniformly than those reported by Bhattacharyya et al. [8]. That said, the percentage of nickel deposits, in the present work, across the TNA surface varied from ~25% at the top one-third region of the sensor to ~80% at the bottom third region of the sensor. Such a large gradient of nickel deposition might have occurred due to two possible reasons: 1) irregularities in TNA morphology; and 2) anode-to-cathode distance (angular placement of the TNA inside the electrodeposition chamber). In the first case, despite the highly ordered nature of the array, minute irregularities such as cracks, dishevelment, etc. could vary current flow on the surface thereby decreasing uniformity in the deposition of Ni(OH)2 [47]. In the second instance, the TNA and Pt electrodes should have been placed in the cell, parallel to one another. However, due to the shape and size of the electrodes, the bottom third of the TNA was observed to be closer to the Pt electrode (during electrodeposition) than the top third, which was somewhat away from the Pt electrode. Because of such a placement of the TNA, more nickel got deposited on the lower surface. It is believed that the combination of these factors resulted in the nonuniform deposition of nickel on the surface of the Ni-TNA sensors. 28 Section 5.1 presents some of the ways by which to increase the homogeneous nickel deposition including controlled anode-to-cathode distance between the electrodes and cyclic voltammetric electrodeposition. CHAPTER 3 SOLUTION-PHASE CYCLIC VOLTAMMETRIC DETECTION OF THE SELECTED VOCs 3.1 Construction of cyclic voltammetry setups Cyclic voltammetry (CV) is an electrochemical technique typically run in conductive, aqueous, and nonaqueous solutions consisting of an analyte (also known as a functional electrolyte) and a salt (known as an auxiliary electrolyte, such as NaCl, KCl, etc.) [37]. The dissolved salts, in the solution/electrolyte, increase the conductivity of the electrolyte allowing for significant quantities of an oxidation-reduction (also known as redox) reaction to take place [37]. Running a CV requires three separate electrodes: 1) a working electrode (WE) where the reduction reactions take place; 2) an inert counter electrode (CE) that completes the electrical circuit and allows for charge flow; and 3) a reference electrode (RE), with no current flowing through it, to monitor/measure the potential of the WE [37]. All three electrodes are placed inside the solutions while running a CV [37]. However, in cases where the analyte is a nonpolar/organic substance, aqueous phase electrochemistry becomes considerably more difficult and often requires the use of complicated electrode systems or expensive organic salts to ionize the solutions [50], [51]. The four VOCs (CH, DMBZ, MCH, and DEC) being investigated in the present study pose a similar difficult: they are not soluble in aqueous/polar solvents and instead require 30 organic/nonpolar solvents for dissolution. Consequently, in the present study, the recording of CVs was performed by dissolving the VOCs in ethanol. Though ethanol was used as a solvent, normal, healthy human breath does contain minor levels of ethanol so detection of the VOCs in ethanol-based solutions can be extrapolated to breath samples [52]. This means that from a clinical perspective, ethanol can enhance the detection of the VOCs. To compensate for the nonpolar nature of solutes and solvent, CVs were recorded in both acidic and basic conditions and the voltammograms were compared with those recorded in neutral pH. It is important to note that while the intended application of the Ni-TNA sensor is in gas-phase detection (of the VOCs), solution-phase cyclic voltammograms (CVMs) may also provide important qualitative information about the possible redox reactions that might occur in gas-phase CV recordings. Additionally, due to CV’s versatility as an electrochemical technique and the vast number of setups that it can facilitate, previous studies have implicated its usage for the detection of various compounds in exhaled breath condensates (EBC) [53]. 3.2 Cyclic voltammetry of exhaled breath condensate Thus far, a significant number of studies have reported the value and possibilities of EBC in the field of noninvasive diagnosis of diseases as well as general monitoring of body conditions [53]–[56]. EBC sampling has become a fairly optimized and standardized process with various commercially manufactured devices available [56], [57]. Detection of EBC based compounds has been carried out with common laboratory techniques such as fluorescence, capillary zone electrophoresis, liquid chromatography combined with 31 ultraviolet-visible (UV-Vis) spectroscopy, GC-MS, etc. [53]–[56]. However, only one study by Gholizadeh et al. [53] has reported the application of cyclic voltammetry as a detection method to measure nitrite content in reduced graphene oxide. Their results show the significant potential of electrochemical analysis of EBC via a viable diagnostic method that can also supplement the gas-phase detection of similar compounds. The experimental design, adopted in the present studies for solution-phase VOC detection, holds significant promise in its applicability with EBC systems, especially because compounds such as ethanol (used as a solvent in the present study) and acetone are found in EBC. Successful application of the Ni-TNA sensor in solutions has significantly advanced its versatility for both EBC and exhaled breath VOC detection. 3.3 Experimental The electrochemical detection of the four VOCs was performed via cyclic voltammetry (CV) in a standard three-electrode electrochemical setup (Fig. 3.1). In this system, while the working electrode (WE) was the Ni-TNA sensor, silver (Ag) foils (Alfa Aesar 0.01” thick, hard, Premion, 99.998%, 100x100mm) acted as both counter (CE) and reference electrodes (RE), respectively. The electrodes were placed at different depths in the electrolyte (50mL solutions of ethanol and VOCs each, Table 3.1). The Ni-TNA coupon was dipped ~3mm into the solutions, while the counter Ag electrode was dipped ~1.5cm, and the reference Ag electrode was ~1cm deep. The electrodes were arranged in a triangular fashion with a working-to-counter electrode distance of ~1.5cm, working-toreference electrode distance of ~1cm, and a counter-to-reference electrode distance of ~1.8cm. The RE was placed as close as possible to the WE to minimize resistive loss (also 32 Figure 3.1 Experimental setup used for the solution-phase cyclic voltammetry of the VOCs. NT: Ni-TNA sensor Table 3.1: List of solutions made in ethanol for the solution-phase CVs of the VOCs Individual VOCs Mixed Solutions of VOCs 50mM Cyclohexane (CH) 50mM CH + DEC 50mM 1,3 Dimethylbenzene (DMBZ) 50mM CH + DEC + DMBZ 50mM Methylcyclohexane (MCH) 50mM CH + DEC + DMBZ + MCH 50mM Decanal (DEC) 50mM CH + 5mM DEC (only acidic condition) known as voltage drop) due to an IR drop. The electrodes were connected to a potentiostat (model: EmStat 3+ PalmSens) which, in turn, was USB-linked to a computer loaded with the accompanying PSTrace (v. 5.4) software for data recording and analysis. To investigate the influence of pH on the detection of the VOCs, solutions of neutral (pH 7), acidic (pH 5-6), and basic (pH 8-9) pH were made. Concentrated HCl (Fisher Scientific Grade) was used for acidification while NH4OH (MiliporeSigma GR ACS Grade) was used for increasing the basicity. All solutions (including neutral) were stirred at 100 rpm and heated to 52oC for 40 mins to ensure adequate mixing before the addition of the acid/base. After 5 mins, stirring was stopped before the CV tests and the electrolyte 33 pH was checked with a Whatman indicator paper (pH 0-14, type CF) both before and after the CV runs. CV runs of pure ethanol were run as a negative control test. For each of these tests, a new sensor was used to mitigate any potential interfering effects that may arise from a used sensor. CV data were extracted from the PSTrace software and plotted in Microsoft Excel. 3.4 Results Detection of the four biomarkers was carried out via cyclic voltammetry in an ethanol-based medium at neutral, basic and acidic pH. Ethanol was chosen as the solvent for these experiments because it satisfied the requirements of being a conductive medium that could solubilize the biomarkers. Pure biomarker samples are organic liquids and cannot carry charge. Detection was evaluated based on whether or not the systems demonstrated linear and/or resistive behavior with low currents since that is indicative of a lack of electrochemical interaction between the Ni-TNA sensor and the biomarkers. The following sections detail the interactions of the sensors and biomarkers in different environments. 3.4.1 Neutral solutions CVMs of both the individual biomarkers and their mixtures showed linear resistive behavior (Figs. 3.2 and 3.3). Both of these figures showed the absence of current responses between the sensor and the VOCs. As CV detects the current response of an electrochemical reaction within a specified voltage window, their absence throughout the entire potential window rendered CV’s application in neutral pH solutions inconsequential. 34 Figure 3.2 Cyclic voltammograms of the individual VOCs and EtOH under neutral conditions exhibiting linearity across the whole sweeping window. Figure 3.3 Cyclic voltammograms of mixed solutions of VOCs showing linear behavior across the entire sweep range. 35 3.4.2 Basic pH experiments Experiments in the basic condition were run identical to ones performed under neutral condition. Figure 3.4 shows the voltammograms (CVMs) of all four individual biomarkers. Figure 3.5 shows the CVMs of CH, DMBZ, MCH, and EtOH, excluding DEC. Unlike in the case of neutral condition where all the biomarkers demonstrated linear, resistive behavior, CVMs of DMBZ, MCH and DEC showed different trends; only CH retained the same linearity. In the case of DMBZ, the forward anodic sweep transitioned from linear to significant “noise-like” hysteresis with a current rise at ~1.28V. The hysteresis persisted in the reverse anodic direction until the same potential, after which, it resumed a linear behavior again. This hysteresis, not seen in the EtOH and CH curves, was Figure 3.4 Cyclic voltammograms of VOCs and EtOH at basic pH showing reaction of DEC with the Ni-TNA sensor as a current “hump” at ~1.78V. Cathodic (negative) sweeps not beyond -0.5V not shown due to a lack of cathodic/reduction current. All CVMs are shown between -0.5 to 2.0V. 36 Figure 3.5 Cyclic voltammogram is showing all VOCs except decanal (DEC) at basic pH showing clear indicators of reaction with the Ni-TNA sensor in the form of hysteresis and “humps” in the current. perhaps an indication of some form of interaction of the sensor with DMBZ. Similarly, MCH demonstrated clear markers of interaction with the sensor but not in the form of hysteresis. In the forward anodic sweep of MCH, the linear rise in current was interrupted by two significant “humps” at ~0.95V and ~1.56V, respectively. Its reverse anodic sweep demonstrated a linear decrease until ~0.67V when there was a significant “drop” in the current. This behavioral trend of “humps” in current was also seen in the CVM of DEC. Unlike the two humps of MCH, the forward anodic sweep of DEC only had one significant current hump at ~1.78V after maintaining a linear increase; the reverse anodic sweep was not significant in this case. This hump probably serves as evidence of an interaction between the Ni-TNA sensor and DEC. Like the CVMs of the individual biomarkers, the CVMs of the mixed solutions of 37 the biomarkers also showed trends unique to the mixtures (Fig. 3.6). In the mixed solution of CH and DEC, the CVM showed significant hysteresis in both the forward and reverse anodic sweeps starting at ~0.24V. Additionally, the observed current at 2.0V of this mixture was significantly greater than the currents at that potential for either of the biomarkers, individually. These two observations are indicative of a relatively stronger interaction between the sensor and the two biomarkers. Addition of DMBZ to the CH and DEC mixture altered the interaction of the sensor with the biomarkers. The hysteresis observed in the former case was completely squashed, and the current at 2.0V of this mixture increased dramatically. Despite the lack of hysteresis, the sensor still demonstrated noteworthy anodic interactions with the CH, DEC, and DMBZ mixture because the Figure 3.6 Cyclic voltammograms of mixed solutions of VOCs at basic pH showing similar characteristics of hysteresis and significantly high magnitudes of current indicative of reaction with the Ni-TNA sensor. 38 magnitude of the current was several orders of magnitude higher than the currents observed for any of the individual biomarkers. Though the addition of DMBZ to the mixture of biomarkers eliminated the hysteresis effect, it was observed again when MCH was mixed with the other three biomarkers. In the forward anodic sweep, hysteresis started at ~1.21V and persisted through the reverse anodic sweep until the same potential, after which the linear behavior resumed. Much like the other two cases, the current at 2.0V was also significantly greater than any of the individual biomarkers. These observations indicated a strong interaction between the Ni-TNA sensor and the mixture of all four biomarkers. 3.4.3 Acidic pH experiments CVs in the acidic solutions were run in a similar manner as those recorded in both neutral and basic solutions. Figure 3.7 shows the CVMs of each of the biomarkers separately, which highlights the anodic peak potentials, while Figure 3.8 shows CVMs of the mixed solutions of biomarkers and their anodic peak potentials. Unlike in the neutral or basic conditions where the Ni-TNA sensor exhibited minor reactivity with a few of the biomarkers, acidification of the biomarker solutions significantly enhanced these interactions with all of the biomarkers. Additionally, the recorded CVMs look like “true” voltammograms with the formation of clear anodic peaks with significantly high currents. The CVMs of the individual biomarkers, CH, MCH, and EtOH, exhibited anodic peaks at ~1.50V, ~1.53V and ~1.47V, respectively. All three potentials are within 60mV of each other but the peak current recorded for EtOH is significantly lower than that of CH and MCH, which is indicative of an enhanced interaction of the Ni-TNA sensor with the biomarkers. In contrast, DMBZ and DEC had anodic peaks at ~1.28V and ~0.88V, 39 Figure 3.7 Cyclic voltammograms of VOCs and EtOH at acidic pH showing significant and unique anodic peaks for each. Figure 3.8 Cyclic voltammograms of mixed solutions of VOCs in acidic pH showing significant and unique anodic peaks for each mixture. Note the migration of peaks to lower anodic potentials as the number of VOCs in solution increases. A shift in the anodic peak of CH+DEC mixture to a higher potential is also observed due to a change in DEC concentration. 40 respectively. While DMBZ’s current is comparable with that of CH and MCH, DEC’s anodic current is relatively smaller; all of the biomarkers had significantly larger currents than the value recorded for pure EtOH. Like the individual biomarkers, CVMs of their mixtures also exhibited discernable anodic peaks specific to the biomarkers present. The anodic peak potential of the CH and DEC mixture was ~0.92V, and when the same mixture was recreated with a lower concentration of DEC, the anodic peak shifted to a more positive value of ~1.07V. This shift is indicative of the influence of biomarker concentration on the interaction of the NiTNA sensor and the biomarkers. This trend of peak-shifting to smaller potentials was also seen upon the addition MCH to the mixture of the other three biomarkers. The anodic peak potential for the mixture containing all the four biomarkers was ~0.55V, which was significantly lower than any of the individual biomarkers. The aforementioned anodic peaks were seen in the first CV scan of all the solutions. Subsequent scans carried out in the same solutions with the respective sensors, immediately after the first scans, produced significantly different CVMs. Figures 3.9 and 3.10 show the second and third CV scans of each of the biomarkers and ethanol, respectively. Compared to their corresponding first scans, the second and third scans featured significantly lower currents and no anodic peaks at the potentials from their first scans. Instead, in the second scan CVMs, all biomarkers showed small anodic peaks in the 0.15-0.70V range; ethanol was the only exception to this pattern showing no peaks at all. The third scan CVMs also repeated this behavior with smaller currents than the first and second scan CVMs along with smaller anodic peaks in the comparatively narrower range of 0.06-0.5V, with closer alignment in all the peaks. This trend was also seen in the second and third scans of the 41 Figure 3.9 Cyclic voltammograms of VOCs and EtOH at acidic pH showing the second sweep. Note the lack of any anodic peaks at the potentials observed in the first scan but the presence of new peaks within the range of 0.15V-0.70V. Figure 3.10 Cyclic voltammograms of VOCs and EtOH at acidic pH showing the third scan. Note the repetition of the same peaks seen in scan 2 but their slight shifts to lower anodic potentials, significantly greater overlap and lower magnitudes. 42 mixed solutions of the biomarkers (not shown). Figure 3.11 is an image of a sensor after recording the CV in an acidic solution. Unlike in the neutral and basic solutions where the sensor’s physical appearance remained unchanged after the test, the portion of the surface of the sensors that was submerged in the acidic solution had been stripped of the black, nickel layer. Subsequent SEM and EDS analyses (Fig. 3.12) showed that the CV tests in acidic conditions resulted in the stripping of all the nickel nanoparticles from the surface of the Ni-TNA sensor, leaving only the anatase TNA substrate. Further analysis indicated that the majority, if not all, of the nickel stripping took place after the first CV, and minimal surface Ni(OH)2 was present during the second and third scans. This was an interesting observation because the sensor retains functionality after sitting in ambient air and in water, and so it begs the question of whether Ni stripping would be observed in breath testing since breath is slightly acidic. Figure 3.11 Photographs of the sample sensors showing the physical appearance of the sensors after synthesis and used under different pH conditions. Note the lack of black Ni(OH)2 nanoparticles on the portion of the sensor submerged in acidic solutions. 43 Figure 3.12 SEM images and EDS analyzed elemental wt. % of a Ni-TNA sensor used under acidic conditions: (a) Sensor surface pre-CV run; (b) Calculated elemental wt. % of elements on the surface of (a); (c) Sensor surface post-CV run; (d) Calculated elemental wt. % of elements on the surface of (c). Note that the presence of F and Cl in the first SEM image can be attributed to the anodization and electrodeposition processes, respectively. Presence of C in the second SEM image can be due to some adsorbed organic compounds or other unknown contaminating particulates after use. 3.5 Discussion 3.5.1 Electrocatalytic activity of Ni(OH)2 in the detection of the VOCs The presence of Ni(OH)2 complexes on the TNA surface is crucial for the detection of the VOCs. Previous work by Bhattacharyya et al. [8], [28] and our cyclic voltammetry studies (not shown) have demonstrated that the nonfunctionalized TNA surface could not interact with the target VOCs in our experimental conditions. The necessity of the Ni(OH)2 44 complexes was explored in the works of Hossieni et al. [46] and Cheshideh et al. [47] who used nickel nanoparticles for the electrooxidation of methanol. Their cyclic voltammetry studies demonstrated a series of three, highly likely, sequence of events that were involved in the oxidation/detection of the target compound [46], [47]. As seen in the XRD profile, nickel deposited on the TNA surface existed in two different crystalline forms: the highly hydrated α-Ni(OH)2 and the less hydrated β-Ni(OH)2 (Fig. 2.7). During the stages of the positive anodic sweep, (~0.25-0.52V vs. standard calomel electrode, SCE) α-Ni(OH)2 was converted into β-Ni(OH)2 [47]. At more positive potentials (~0.61V vs. SCE), the βNi(OH)2 complexes were oxidized into the electro-active oxidizing agent of Ni(OOH) that could then react with the VOCs [47]. Equations (2)-(4) show the probable sequence of events that might have occurred in the solution (liquid) electrochemical conditions [46], [47]. α-Ni(OH)2 → β-Ni(OH)2 β-Ni(OH)2 + OH- → NiOOH + H2O + e- (fast) NiOOH + VOC → β-Ni(OH)2 + oxidation product(s) of VOC (2) (3) (4) 3.5.2 Neutral solutions Equation (3) shows two essential species for the formation of Ni(OOH): β-Ni(OH)2 and OH-. Bhattacharyya et al. [8], [28] have reported that the OH- species is particularly important due to it being a strong Brønsted-Löwry base, and in this case, it is important for oxidizing β-Ni(OH)2 into Ni(OOH). The importance of this conversion was amply highlighted in our neutral pH cyclic voltammetry experiments where the Ni-TNA sensor 45 failed to detect any of the VOCs. The cyclic voltammetry solutions consisted of a combination of 190 proof EtOH (95% EtOH and 5% H2O) and an organic biomarker. This is a nonconductive solution due to the extremely low disassociation constants of EtOH and H2O (Ka values of 1.3x10-16 and 1.8x10-16 at 25oC, respectively), which means that little to no OH- species were available to activate the Ni-TNA sensor. Lack of Ni(OOH) combined with the inherent nonconductive nature of the solution might have prevented any interaction of the sensor with the VOCs. 3.5.3 Basic pH behavior In contrast to the neutral solutions, cyclic voltammetry runs in solutions with a pH in the range of pH 8-9 facilitated the detection of DMBZ, MCH, and DEC but not CH. Under these conditions, the addition of NH4OH, a relatively weak base, provided moderate ionicity to the solution via free NH4+ and OH- species. In addition to improving the conductivity of the medium, the latter molecules could perhaps increase the concentration of Ni(OOH) and thereby allow the Ni-TNA sensor to undergo redox reactions with the three VOCs (Eq. 4). The recorded CVMs in Figures 3.4 and 3.5 give some insight as to what these oxidation products might have been. It is important to note that the lack of corresponding cathodic peaks indicated that the reactions between the sensor and the VOCs were irreversible. In the case of DMBZ, the significant hysteresis beginning at ~1.28V indicated gas evolution at the Ni-TNA sensor and that gas/vapor might potentially have been some of the intermediates or the final product(s) due to the oxidation of DMBZ. It has been reported by Long et al. [58] that isophthalic acid is the oxidation product of DMBZ, and according 46 to Bhattacharyya et al. [8], [28], the intermediates formed during the oxidation, got adsorbed onto the sensor surface. However, the hysteresis in the CVM along with the lack of any carbon deposits on the used sensors indicated that the complexation theory proposed by Bhattacharyya et al. [8], [28] is unlikely. Instead, these intermediates or even the isophthalic acid might have been generated, which either could have remained in the electrolyte or bubbled out of it. Similar to DMBZ, MCH also underwent an oxidation reaction with the Ni-TNA sensor. According to Yolcular et al. [44], toluene is the final oxidation product of MCH, and while in their experimental design they did not form any aromatic intermediates, it is possible that our experimental setup might have produced some. In the forward anodic sweep of the CVM, the two discernable “humps” as possible current peaks perhaps indicated the sensor’s reaction with two separate compounds. Unlike in the case of DMBZ, no gas evolution was observed (no hysteresis), and so it is possible that all the oxidation products remained in the electrolyte. That said, the identity of the oxidative products could not be ascertained in the absence of further corroborative studies by techniques such as chromatographic analyses via GC-MS. These results also indicated the unlikelihood of the occurrence of the complexation theory proposed by Bhattacharyya et al. [8], [28]. In a similar pattern to MCH, DEC underwent oxidation with the Ni-TNA sensor as seen by a current hump in the CVM at ~1.78V. The presence of a single jump indicated that perhaps only one oxidative product was formed. The potential at which the hump occurred was considerably larger than the onset of hysteresis for DMBZ and the current humps of MCH. This behavior indicated that under basic conditions, DEC might have become relatively more stable compared to the other two. Cooper et al. [45] have suggested 47 that the oxidation product of decanal is peroxy acid, and since no hysteresis was observed, it can be expected that the oxidation product remained in the solution. Further analyses would be required to establish the exact nature of the products in the solution. Again, these behavioral patterns did not support the formation of any complex between the VOC and the Ni-TNA sensor. While DMBZ, MCH, and DEC seemed to have undergone some form of oxidation reactions with the Ni-TNA sensor, CH’s CVMs mirrored the linear resistive behavior seen in the CVM of EtOH. This means that under basic conditions, CH along with EtOH did not undergo any noticeable oxidation. Though the exact cause for this lack of oxidation is unknown, it is possible that CH may have been highly stable under basic conditions or within the potential window of -2.0V to 2.0V. Perhaps, a voltage higher than 2.0V will be required to form oxidation products, if any. The CH/DEC mixture exhibited clear signs of oxidation and gas evolution with a hysteresis onset at ~0.24V. This is significantly lower than the ~1.78V of isolated decanal, indicating that this mixture was considerably more unstable and readily underwent oxidation. The magnitude of the current change (rather the absolute value of the current) was considerably larger than current change for either of the biomarkers, which served as further evidence of oxidation taking place. The increased instability and high current response might have been due to two possible reasons: 1) interactions between DEC and CH; and 2) interactions between the oxidation product(s) of DEC and CH (if any) since the oxidation of the latter was unlikely. Further chromatographic analyses are necessary to understand which of these possibilities could have become the likely scenario. Again, these CVs also indicated the unlikelihood of the complex formation theory. 48 The addition of DMBZ to the above mixture also demonstrated clear trends of oxidation. In the mixture of CH, DEC, and DMBZ, the lack of hysteresis indicated that no gas evolution took place; instead, the oxidation products may have remained in the solution. The magnitude of the current change was significantly large much like the in the case of the CH/DEC mixture, and this is yet another evidence supporting an oxidation mechanism. The possible reasons attributed to the high current response in the above case may apply here as well. In other words, interactions between the VOCs themselves, interactions between the oxidation products and the VOCs or interactions between the oxidation products amongst themselves, are the plausible reasons for the trends and patterns observed in the CVMs. The unique patterns observed for each of the mixtures were also seen with the of MCH. In the mixture of all four VOCs, clear signs of oxidation were visible in the CVM with a hysteresis onset at ~1.21V and an appreciable magnitude of current change. The hysteresis is comparatively smaller than the one seen in the CH and DEC mixture, indicating that minor levels of gas evolution might have occurred. The possible reasons for these behavioral patterns in oxidation are akin to the general ones mentioned in the above case. It is also worth reiterating that the oxidation patterns varied between mixtures. Though cyclic voltammetry could qualitatively differentiate the compositions of the mixtures in basic solutions, the distinct changes between individual biomarkers and mixtures in acidic conditions were significantly more pronounced, and the interactions of the VOCs with the Ni-TNA sensor were greatly enhanced. 49 3.5.4 Acidic pH behavior Dropwise addition of concentrated HCl to the biomarker solutions increased the electrolyte conductivity, which facilitated the formation of the Ni(OOH) active species. The initial reaction between HCl and H2O, shown in (5), subsequently followed by the reactions shown in (3) and (4) may have taken place inside the electrolyte and contributed to the observed effects. HCl + H2O → H3O+ + Cl- (5) The most significant difference between the CVs recorded in both basic and acidic solutions was the enhanced reactivity between the Ni-TNA sensor and the VOCs, in the latter condition. CVMs in the acidic solutions showed somewhat “ideal” anodic peaks instead of small current “humps” or hysteresis, indicating the occurrence of the complete oxidation of the VOCs. Besides, the magnitude of the current changes was in the range of 2.5-5.5mA. Such changes were significantly higher than any current changes observed in basic solutions. In addition to high current responses, the positions of the anodic peaks were also indicative of a particular or a group of VOCs. For instance, CH, MCH, and EtOH had anodic peaks at ~1.50V, ~1.53V and ~1.47V, respectively, and while these values were fairly close to one another, the change in current, in the case of EtOH, was significantly lower than the other two. This may indicate that the sensor was not as efficiently oxidizing EtOH as the biomarkers. Oxidation of CH leads to the formation of adipic acid and may follow one of two reaction pathways: 1) direct conversion to adipic acid; or 2) via the formation of the intermediates cyclohexanol and cyclohexanone [42], 50 [59]–[61]. It is unclear whether both or one of these pathways occurred on the Ni-TNA electrode. It is also worth noting that these oxidation reactions did not take place under neutral and basic solutions. The negligible difference in the anodic peak positions of CH and MCH may be attributed to their significant structural similarity; MCH has a single CH3 group attached to the cyclohexane ring. In contrast to CH and MCH, DMBZ’s anodic peak was at observed ~1.28V, which was lower than both of their potentials. This lower potential could have been due to the comparatively higher instability of the DMBZ structure, making it more prone to oxidation than CH and MCH. In the same vein, DEC’s anodic peak was at even lower potential (~0.88V), indicating that it may possess the highest tendency to oxidize under acidic conditions. It is speculated that the oxidation products of these VOCs, under acidic conditions, are identical to those that may have been created in the basic conditions. However, further chromatographic analyses are needed in order to determine the oxidation products qualitatively. Significant anodic peaks were also observed in mixtures of the biomarkers under acidic conditions. Recorded CVMs showed a trend that qualitatively distinguished between mixtures based on the number of VOCs that were present. In other words, as the number of VOCs in the mixture increased, the anodic peak potential shifted to less positive values (Fig. 3.8). Also, the concentrations of the biomarkers seemed to change the anodic peak positions. For example, the mixture of 50mM CH and 5mM DEC registered an anodic peak at a value of ~150mV greater than the current peaks observed with an equimolar concentration of the two. The anodic peak shifts also support the possibility of oxidation products of the VOCs either interacting amongst themselves or with the VOCs. A decrease in the DEC concentration did not shift the current peak closer to that obtained for pure CH. 51 If the oxidation products were relatively inert, then not only would there have been two distinct anodic peaks (one for CH and one for DEC) but the peak(s) would also have not existed between the two peaks corresponding to each. Chromatographic studies are necessary to understand the exact nature of the interactions of the various compounds amongst each other and with the Ni-TNA sensor under acidic conditions. Additionally, for effective qualitative application of the sensor and detection method pertaining to VOC mixtures, a library containing all possible VOC mixtures at various concentrations will require to be rigorously tested. 3.5.5 Effect of acidic pH on sensor morphology and performance The anodic peaks observed in the acidic conditions were only seen in the first scan. Subsequent scans with the same sensor in the solutions showed anodic peaks that were significantly smaller in magnitude and within the small voltage range of 0.15-0.70V; the third scan had a similar range of 0.06-0.5V. It is speculated that these peaks were not oxidation peaks corresponding to the biomarkers because SEM/EDS analysis showed that the portion of the sensor that remained completely immersed in the acidic solution during the recording of the CVs was completely stripped of nickel. Instead, these anodic peaks may have formed due to one of three possible reasons: 1) adsorption of VOC/VOC oxidation products onto the sensor; 2) conversion of α-Ni(OH)2 to β-Ni(OH)2; and 3) experimental artifact. In the first case, the lack of Ni(OH)2 layer on the TNA surface may have allowed for the adsorption of the VOCs or their oxidation products, the latter being more likely since the VOCs did not show reactivity with the unfunctionalized TNA surface. In the 52 second case, it is possible that during the cathodic sweep of the second and third scans, some nickel may have deposited on the surface since the scan was carried out all the way to -2.0V (data are only shown to -0.5V), which is sufficient for nickel deposition. In the same vein, the conversion of α-Ni(OH)2 to β-Ni(OH)2 may have also occurred at the potentials of these anodic peaks since they are within the same range listed by Cheshideh et al. [47]. It is important to point out at that their reference electrode was SCE and their electrolyte was NaOH, while our reference electrode was Ag, and the electrolyte was EtOH so the expected potentials for those conversions would be different. Nevertheless, this is another possible reason for the formation of those peaks. The last possible reason could be that these peaks were merely experimental artifacts. Irrespective of the reason(s), it is clear that the Ni-TNA sensor gets damaged in acidic solutions and would not function beyond a single scan. We have observed (not shown) that the use of a new sensor for each of the three scans mitigates the issues observed with a single sensor used for all three scans. As a general principle, it is assumed that the sensors are single-use, and each test, regardless of the conditions, necessitates a news sensor. This is an important observation to keep in mind for future research work. CHAPTER 4 GAS-PHASE AND BREATH-BASED CYCLIC VOLTAMMETRIC DETECTION OF THE VOCS 4.1 Current trends in the use of conductive polymers in electrochemical gas detection At the frontier of research activity in electrochemistry is the preparation, characterization, and application of electroactive polymers [62]–[64]. Conductive polymers present a wide range of promising applications including electroanalysis, energy storage, bioelectrochemistry, electrocatalysis, photoelectrochemistry, organic electrochemistry, etc. [63], [65]. Charge propagation in these polymers is determined by their chemical structures and usually takes place in one of either two ways: 1) electronconducting; and 2) proton (ion)-conducting [62], [63]. Usage of conductive polymers in electrochemistry typically involves a polymer film electrode with three separate phases contacted successively [63], [66]. The polymeric phase is sandwiched between the liquid electrolyte (containing the analyte) and a metal electrode whose surface it typically adheres or adsorbs onto [63]. For gas sensing purposes, the only difference in the electrochemical setup is the lack of an electrolyte solution containing the analyte [63]–[67]. Conducting polymer-based sensing systems utilize a variety of sensing principles such as measuring changes in resistance, current, material mass, etc. [63]–[65]. Changes in these variables are 54 observed when the oxidation state of the polymer is altered due to a transfer of electrons from the analyte to the polymer or vice versa [63]–[65]. For example, many studies have shown how a reducing gas, such as ammonia, increases the resistance of a polyaniline film [63]. Correspondingly, numerous conductive polymer films and hybrids have been constructed for the detection of volatile organic compounds (VOCs) [63], [64], [66], [68]. Mallya et al. [66] designed a polymer-carbon black composite that showed selective detection of toluene vapor over compounds such as acetone and cyclohexane vapors. Zampetti et al. [67] synthesized a highly sensitive nitrogen dioxide (NO2)sensor based on poly(3,4- ethylenedioxythiophene)/polystyrene sulfonate (PEDOT-PSS) coated on TiO2 nanofibers. While the above examples demonstrate the versatility of conductive polymers in electrochemical cells, to our knowledge, no one has reported the use of these polymers, specifically electron-conducting polymers, as an electrolyte in electrochemical cells. Recent advances in renewable/rechargeable battery designs have reported conductive polymer electrolytes, but these systems use the polymers as ion-conducting media that transfer an electroactive species between the cathode and the anode [69], [70]. In contrast, the present study used graphene embedded polylactic acid (G-PLA) conducting polymer to perform electron-exchange reactions. Literature studies have revealed that no previous study has reported on the use of G-PLA to perform electrochemical measurements. Some of the recently published literature describes the application of an advanced manufacturing process to print G-PLA for applications in energy storage devices, fabrication of antibacterial composites, and tissue engineering, etc. [71]–[75]. One of the objectives of the present study is to examine the possibility of using G-PLA for the electrochemical 55 detection of gaseous breath-based CRC VOCs by cyclic voltammetry. 4.2 Experimental 4.2.1 Gas-phase VOC detection setup For gas phase detection of the VOCs, a custom-made planar three-electrode system, mimicking the one from the liquid tests, was designed. Graphene embedded PLA (G-PLA) filament (Black Magic 3D, 1.75mm, 100g) was 3D printed using Flash Forge Creator Pro Printer into a 1.0(H) x 25(W) x 50(L) mm rectangle. Silver paint (Silver print, MG Chemicals) was used to print three separate connection pads, on the polymer, for the electrodes. A custom-made sensing chamber, made of acrylic, was also 3D-printed to house the sensor-fitted polymeric pad and allow for the inlet/outlet connections for VOC vapor. Figure 4.1 shows the dimensions of the pads and box along with an image of the setup. While printed silver pads served as the RE and CE in the electrochemical cell, the Ni-TNA coupon, placed at the center silver square pad, served as the WE. All three electrodes were attached to alligator clip extensions with one side connected to the electrodes, and the other side was insulated with Kapton tape and attached to the bottom side of the polymer pad. The alligator clip extensions were soldered to a palmsens strip (Metrohm DropSens strip) which, in turn, was connected to the EmStat potentiostat for recording the CV data. 4.2.2 VOC detection under gaseous conditions To generate VOC vapor for the gas-phase CV runs, nitrogen gas was bubbled through aliquots of 25mL samples of each of the VOCs at a flow rate 200sccm (standard cubic centimeters per minute) in order to vaporize them. VOC vapor was stored inside one- 56 Figure 4.1 Schematic of the G-PLA base with silver connection pads printed on the surface (left) and a photograph of the experimental setup used for gas-phase cyclic voltammetry (right). R: Reference electrode, W: Working electrode, C: Counter electrode. The 3D printed acrylic box had outer dimensions of 35(H) x 85(W) x 85(L) mm and inner dimensions of 35(H) x 70(W) x 70(L) mm. liter Tedlar bags (SKC Tedlar sample bag), filled into the maximum possible extent. Before VOC exposure, a baseline CV for each sensor was run in ambient air, and then VOC vapors from the bag were gently introduced into the sensing chamber via a pump (model: SKC Pocket Pump 210-1002A) at a flow rate of 25mL/min. This procedure was repeated with 50mL acidic solutions of the individual VOCs as well as mixtures of the VOCs, dissolved in acetone and bubbled into the Tedlar bags. The solutions were prepared in the same way as the ethanol-VOC solutions used in the liquid system test. The vapor-phase compositions were identical to ones made for the liquid-phase conditions. It is worthwhile to mention here that acetone was preferred for these (vapor-phase) tests and acidic (liquid) solutions (neutral and basic conditions not tested here) because EtOH vapor reactivity with the sensor was indistinguishable from that of VOC (not shown). CVMs were recorded and plotted in Excel. 57 4.2.3 VOC detection in breath samples The breath samples from a single healthy subject (not known to have colorectal cancer) were collected in two 5L bags. A 25mm syringe filter with a 0.2μm PTFE membrane (VWR) was inserted between the bag and the sampling tubes to condense water vapor out of the breath samples and onto the tubes. One of the 5L bags was filled in with breath and N2 gas while the other one was filled in with breath and all four VOCs’ vapors bubbled in from a mixed solution (not dissolved in acetone). Baseline CVs of the sensors were recorded before the exposure to breath samples. Breath samples were flown over the sensor at 50mL/min because of the availability of relatively larger sample quantities. CVMs were recorded and plotted in Excel. 4.3 Results 4.3.1 Gas-phase VOC detection Like in the case of the liquid phase detection of the biomarkers, the objective in the gas-phase detection was to study if significant current responses could be recorded as a result of the sensor’s interaction with the biomarkers. In order to test the Ni-TNA sensor’s sensitivity to the biomarkers, pure samples of the VOCs were collected and exposed to the sensor. Figure 4.2 shows the CVMs of the pure biomarker samples and ambient air. While DMBZ, MCH, and DEC showed CVMs within a comparable range, similar to ambient air’s, the CVM of CH exhibited a significantly larger current signal. Though no clear anodic peaks were observed to form, such a behavior nevertheless was indicative of the sensor’s ability to interact with CH under neutral and ambient conditions. The sensor did not seem to interact with the other VOCs in any meaningful way. To identify conditions 58 Figure 4.2 Cyclic voltammograms of pure VOC vapors, individually. Note the dramatically higher response of CH compared to the rest of the VOCs and ambient air. Not shown are the respective ambient air CVs of the sensors used for DMBZ, MCH, and DEC since those CVMs looked identical to the CVMs of those 3 VOCs. under which all VOCs could be detected by the sensor, the biomarkers were subsequently dissolved in ethanol and acetone and prepared in the same way as those prepared for the measurements in acidic-liquid CVs. The biomarkers were then bubbled into the bags (by flowing inert N2 gas into them) and exposed to the sensor. Acidic conditions were preferred since they provided comparatively better detection sensitivities (with the biomarkers) than either basic or neutral solutions. Preliminary CVMs collected from pure ethanol and ethanol-VOC vapors (not shown) showed current profiles very similar to the vapor phase CVM of CH. Due to the possibility of the sensor’s selective interaction with ethanol (rather than with the VOCs), acetone was used as the solvent. Figure 4.3 shows the CVMs of acetone vapor and ambient air, and Figures 4.4 - 4.7 show all three CV scans of the four 59 Figure 4.3 Cyclic voltammogram of acetone vapor and ambient air using the same sensor. Only the first scan has been shown since scans two and three were effectively identical to scan one for each. Figure 4.4 Cyclic voltammogram of CH vapor and ambient air using the same sensor. Note the significant hysteresis beginning at ~1.50V in the CH scans with the first scan exhibiting more than the second and third scans. 60 Figure 4.5 Cyclic voltammogram of DMBZ vapor and ambient air using the same sensor. Note the significant magnitude of change in current, which increases from scan one to scan three, which is the opposite behavior seen in ambient air. Figure 4.6 Cyclic voltammogram of MCH vapor and ambient air using the same sensor. Note the significant magnitude of current change only from scan one to scan two. 61 Figure 4.7 Cyclic voltammogram of DEC vapor and ambient air using the same sensor. Behavioral trends in this CVM match those seen in the DMBZ case. acetone-biomarker vapors compared to their respective sensor’s ambient air baseline. In all four cases, ambient air CVMs demonstrated the same consistent pattern of decreasing current decreasing with multiple scans, and the current at 2.0V was in the range of 0.510μA. However, the biomarkers showed varied responses. CVMs of acetone-CH vapor showed a marginal current increase as compared to the baseline but the trend of decreasing current signals over the three scans were identical to the baseline behavior. A stark difference between the CVMs was the “noise-like” appearance in the CH signal at ~1.5V, which persisted in both the forward and reverse anodic sweeps. This behavior is indicative of some possible minor interactions between the Ni-TNA sensor and the biomarker vapor. In contrast to the CV of CH, DMBZ’s CV scans showed a significantly higher current than those recorded in all the three baseline scans. Additionally, the CVMs of DMBZ showed the opposite trend as those of baseline’s: in 62 each scan, the current response increased, rising to a maximum of ~190μA. No “noiselike” behavior was seen, and the interaction between the sensor and DMBZ appeared to be more significant than that with CH. Many distinct trends observed in the CVMs of DMBZ were also observed in the CVMs of MCH. In contrast to baseline, MCH CV scans showed currents that were significantly higher, and a similar trend of increasing current signals from one scan to the next was also observed. However, a key difference between the CVMs of MCH and DMBZ was that the current associated with the scan three (in the case of MCH) did not exceed the reported current of scan two. Unlike in the case of MCH, the CVMs of DEC replicated the patterns of DMBZ. All three DEC CV scans had significantly higher currents than baseline values and with each scan, the current signal increased. These trends suggested that the Ni-TNA sensor had significant interactions with the DEC vapor. Much like the liquid-phase CV runs, various mixed solutions of the biomarkers were made in acetone. The VOC mixture was bubbled into a bag and then pumped onto the sensor followed by an ambient air CV run. Figure 4.8 shows the CVMs of each of the mixtures and ambient air. Though the four biomarkers demonstrated varying degrees of interaction with the sensor, in a mixed environment, no significant change in the magnitude of either current or “noise” could be observed in the CVMs. 4.3.2 Breath-based VOC detection Thus far, the Ni-TNA sensor was tested for liquid/gaseous phase VOC detection. However, the ultimate goal is to be able to design as a sensor that can directly detect distinguishing biomarkers from patients’ breath. To test the ultimate detection ability of fabricated sensors, breath samples from a healthy (without any history or diagnosis of 63 Figure 4.8 Cyclic voltammogram of mixed VOC vapors and ambient air using the same sensor for all mixtures. Note the lack of any meaningful indicators of reaction between the Ni-TNA sensor the VOCs (i.e., no hysteresis or increase in magnitudes of change in current). CRC) subject and breath samples spiked with the biomarkers of interest were run over the sensor. While Figures 4.9 and 4.10 show the CVMs of the normal breath samples, Figures 4.11 and 4.12 exhibit the CVMs of VOC-spiked breath collected from the same subject. In all four CVMs, the ambient air scans of the sensors showed negligible differences in current signals, but the CV scans of normal and VOC breath showed considerably varying trends. CVM of the first normal breath sample showed currents signals that were significantly higher than ambient air conditions and increased with each successive scan. Over the course of five scans, the current signals appeared to reach an equilibrium/plateau region as seen in the virtual overlap of scans four and five. Similar trends were observed in the CVM of the second breath sample, which also showed a higher current trend than that of the baseline ambient air. Unlike the first sample, the current signals were not 64 Figure 4.9 Cyclic voltammograms of normal patient breath showing increases in the magnitude of change in current from one scan to the next, reaching an upper limit by scan five. Figure 4.10 Cyclic voltammograms of normal patient breath showing a similar trend with the magnitude of change in current increasing to a cap by scan four. 65 Figure 4.11 Cyclic voltammograms of VOC-spiked breath showing a decrease in current from one scan to the next and minor levels of noise hysteresis. Figure 4.12 Cyclic voltammogram of VOC-spiked breath showing the same trend of decreasing magnitude of change in current and minor hysteresis. 66 markedly higher than the baseline values. However, the trend of current signals increasing with each scan to some plateau region could be replicated. Much like the CVMs of normal breath, VOC-spiked breath CVMs showed similar but opposite trends. In the first VOC-spiked breath sample, the current signals were significantly higher than the respective baseline, and these signals decreased from one scan to the next; the latter behavior is the opposite of the one seen in normal breath CVMs. Though CVMs of the second VOC-spiked breath sample replicated the behavioral trends of the first, small differences were observed. In this case, the magnitude of the current increase was not as significant, with respect to baseline, and the decrease in current signals from scan to scan was much lower. 4.4 Discussion 4.4.1 Pure VOC conditions While the solution-phase experiments successfully demonstrated the fabricated sensor’s ability to interact with the VOCs under varying conditions, in a real application scenario, the sensor is intended for the gas/breath-phase detection. Correspondingly, NiTNA sensors were first exposed to pure VOC vapor. Under these ambient conditions, only CH vapor was successfully picked up by the sensor showing a significant magnitude of change in current. It is believed that the reaction between the sensor and CH was most likely the same oxidation reactions that were observed in the acidic solution. However, further analyses are required to be performed to understand the increased current phenomenon. Other three biomarkers, namely, DMBZ, MCH, and DEC did not show any interactions with the sensor, which perhaps indicates the comparatively greater stability of 67 these compounds under ambient conditions. It is also possible that for the successful oxidation of these biomarkers, a better conductive medium may be required. 4.4.2 VOCs dissolved in acetone Dissolving the biomarkers in acetone facilitated the detection of all four VOCs. In the case of CH, hysteresis in all three scans with an onset of ~1.50V (similar to the value obtained in the acidic solution) indicated oxidation reactions; the first scan exhibited the most pronounced hysteresis. However, the magnitude of the change in current was not significant relative to that obtained in ambient air and considerably lower than the current change seen under pure conditions. Additionally, the currents recorded for CH decreased from one scan to the next matching the trend observed in air. These behaviors indicate that while it is possible that CH oxidation took place, significant side reactions between the oxidation products, CH or acetone may have also occurred, contributing to the low current response. Further analysis is needed to determine the reason for the differences observed in CH detection, both under pure and dissolved conditions. Compared to CH, detection of dissolved DMBZ and DEC demonstrated more significant oxidation reactions. Not only was the magnitude of the current changes in both the cases significantly greater than their respective ambient air currents but they also showed the opposite trend: increasing currents from one scan to the next. This latter behavior indicated that as the biomarker concentration increased inside the sensing chamber, more VOC molecules could be oxidized resulting in the larger currents. Furthermore, it is possible that the potential side reactions that could have occurred between the VOCs, their oxidation products or acetone (as seen in the CH case), may not 68 have happened or were comparatively insignificant. Like in the cases of the DMBZ and DEC, the Ni-TNA sensor also exhibited significant reactivity with the MCH vapor showing a higher magnitude of current change. This large change in current was a clear indication of oxidation, and it is hypothesized that the reaction products are akin to the ones observed in the liquid conditions. A key difference observed in the MCH case was a slight decrease in current in scan three relative to scan two. Though MCH initially followed the same pattern as DMBZ/DEC, scan three did not follow the same pattern. This could have happened due to three possible reasons: 1) the sensor used for the detection had significant morphological flaws; 2) the alligator clips may have slipped during the scan, scratching the sensor in the process, thereby, reducing detection; and 3) the oxidation products, MCH vapor, or acetone might have interacted with each other, thereby, decreasing the amount of available MCH molecules. More in-depth analysis is necessary to determine which one of these reasons might have caused the current decrease. When the biomarkers were mixed in acetone and exposed to the sensor, a completely different behavior was observed. This was a significant departure from the trends seen in both acidic liquid conditions and in the individual biomarkers. The reason for such a behavior is unclear, but it is highly probable that the biomarkers were interacting with each other, altering the response of the sensor to the mixtures. Since CV is being used qualitatively, these behaviors give us an interesting qualitative trend and are indicative of something occurring. However, more studies are necessary to determine the exact reason for why these responses were so drastically different. 69 4.4.3 VOC detection in breath samples Following the mimic vapor tests, Ni-TNA sensors were exposed to both normal breath samples collected from a single healthy subject, and breath samples (from the same subject) spiked with all four biomarkers. The collected CVMs showed that in both of these cases, the magnitudes of current change was significantly greater than seen in ambient air indicating that the sensor oxidized one or more things in the samples. CVMs of normal breath showed the magnitude of the current change increasing from one scan to the next and eventually reaching some equilibrium/plateau region beyond which the current did not increase any further. This behavior may indicate that the sensor actively reacted with a compound(s) in healthy breath and either became saturated (i.e., all active sites were used) or the oxidized products slowly decreased the level of interaction between the sensor and the incoming breath by interacting with the latter. In contrast, CVMs of VOC-spiked breath showed the opposite trends of decreasing magnitudes of change in current from one scan to the next and no such plateauing. These trends indicate that the presence of the VOCs altered the reactivity of the sensor with the rest of the breath. There are two possible reasons for what the effect could have been: 1) oxidation products of the breath and VOCs interacted amongst each other or with the VOCs, thereby, preventing any subsequent interaction with the sensor; or 2) the oxidation products remained adsorbed onto the surface of the Ni-TNA, effectively reducing the amount of available active Ni sites for further reaction. Though the exact reason is unknown, it is clear that the presence of the VOCs dramatically altered the way the Ni-TNA sensor responded to the breath samples. CHAPTER 5 LIMITATIONS AND FUTURE WORK 5.1 Limitations The present study, carried out thus far, has some potential limitations primarily based around the sensor design and experimental setup. The sensor fabrication process began with a standard anodization process, which was followed by elevated-temperature annealing. Though the adopted process created a fairly robust TNA substrate, there were minor structural issues such as gaps, dishevelment, and cracks, etc., which not only impacted uniform Ni deposition but also unfavorably impacted charge transfer reactions across the surface. Recent work from Willis et al. [76] has addressed some of these issues by carrying out anodization of the TNA at carefully chosen temperatures. Their TNA substrate showed comparatively better geometry and fewer defects. The TNA, fabricated by them, may improve the quality of the results. The fabricated Ni-TNA sensor also some showed heterogeneity in Ni deposition on the surface. While irregularities on the TNA surface contributed to some of the nonuniformity, electrodeposition, in general, creates heterogenous deposits. Additionally, placement of the electrodes in the solution (anode-to-cathode distance) further influences the deposition process. The electrode placement created a gradient nickel concentration on the TNA surface. Subsequently, these gradients and heterogeneity influenced the overall 71 efficacy of the sensor in detecting the VOCs. Steps such as reducing irregularities in the TNA surface, improving electrode placement inside the solution, and potentially adopting a 3-electrode setup for CV-based electrodeposition (as described by Cheshideh et al. [47]) may increase the homogeneity of the Ni deposits on the Ni-TNA surface, which in turn, will improve the performance characteristics of the sensor. Our preliminary studies have indicated that effective stirring of the electrolyte, during the deposition process, improves the deposit homogeneity because of the improved mass transfer process. Detection of the biomarkers by the Ni-TNA sensors was carried out via cyclic voltammetry, a powerful electrochemical technique that could isolate specific reactions between the VOCs and the sensor. While detection was achieved under different liquid and gaseous conditions, the electrochemical cell used in this study was far from an ideal configuration. The two main issues that have contributed to the nonideal nature of the electrodes are 1) the working electrode surface of the Ni-TNA sensor was heterogeneous; and 2) Ag counter and reference electrodes are not truly inert and can get oxidized/tarnished, both in liquid and gaseous solutions because of the reactions within the electrolyte and VOCs. It is well understood that for recording appropriate CVs, a uniform (both in terms of shape and composition) working electrode is necessary. However, we could reduce the working electrode heterogeneity to some extent to lessen its impact on the measurements. Similarly, the counter electrode should have consisted of an inert material that is relatively inert and whose surface areas should not have changed during the measurement. Ag counter electrodes used in our liquid experiments showed some tarnishing/oxidation over time, which could have influenced the magnitude of the measured current. It is possible that even though the reference Ag electrode did not have 72 any current flowing through it, the adsorption of any organic compounds (organic compounds can easily adsorb onto the electrode surfaces, particularly in the absence of barriers to isolate the electrodes in the electrochemical cells) might have also occurred. The adsorption phenomenon could potentially create some uncertainty in the measuring voltages as a true (and nonreacting) reference electrode is usually preferred to determine the working electrode voltage. Indeed, in our studies, we observed that shifts in peaks/onset voltages were within a 50-75mV range, which is not significantly large but sufficient to add some difficulty in analyzing the CVMs. It is also worth noting that while CV can provide quantitative data, we primarily used the technique for qualitative measurements because of the limitations described above. Though the synthesized Ni-TNA could detect all the four VOCs of interest, it did so under somewhat specific conditions. In solution form, the most favorable and significant detection took place in acidic solutions. It is speculated that the comparatively weaker responses in basic solutions were due to use of a weak base, such as NH4OH. Use of a stronger base, such as NaOH, may produce better anodic peaks. In the gas phase, only CH could be detected in its pure state but upon the addition of a slightly polar and predominately nonpolar solvent (acetone), to solubilize the VOCs in an acidic solution, allowed the detection of all four VOCs. However, when the same biomarker vapors were mixed, the sensor failed to respond to any of them. VOC-spiked breath showed a qualitatively significant trend that was different from normal breath. These behaviors are not only indicative of the sensitivity of the sensor to its environment but also demonstrated that the mere presence of the biomarkers might not have been sufficient for their detection. It is postulated that for effective detection of the biomarkers, breath sample preparation 73 may become a necessary component before running the CV. Based on our results, preparatory steps might include acidifying breath samples from patients and adding some vapors like acetone or other polar compounds known to be unreactive to the sensor but could facilitate or enhance the interactions of the biomarkers and the sensor. It is difficult to exactly conclude what preparatory steps should be taken into consideration because it is unclear how the oxidation reactions between the Ni-TNA sensor and the VOCs proceeded. Further work to elucidate these mechanisms is necessary and may prove to be vital/essential in determining the sample preparation procedures and improving sensor efficacy. An important limitation also exists in the selection of VOCs. Recently, Markar et al. [77] published their findings on breath-based VOCs of stomach and esophageal cancers and labeled decanal as one of the critical diagnostic biomarkers for those cancers. Similarly, both Oguma et al. [23] and Phillips et al. [24] have implicated cyclohexane as an exhaled VOC in lung cancer and tuberculosis, respectively. Oguma et al. [23] also mentioned xylenes as additional breath-based biomarkers found in lung cancer. These studies indicate that these VOCs, individually, may not be unique to colorectal cancer. For instance, decanal’s presence in stomach, esophageal, and colorectal cancers may indicate that it is a biomarker found in gastrointestinal cancers, in general. However, it should be noted that this methodology of diagnosing CRC requires the specific presence of all four critical VOCs, chosen in the present study. Present studies also underscore the importance of the pattern/combination of the VOCs. 74 5.2 Future work Future work in this area will primarily be focused on not only improving the quality of the Ni-TNA sensor but also elucidating the mechanisms by which it interacts with the VOCs. Improvements in sensor synthesis such as adjusting both anodization temperature and electrodeposition parameters/setup could ensure better surface morphology (of the TNA surface) and homogeneity of surface nickel deposition/distribution. Additionally, it is worthwhile to mention here that a new sensor design consisting of 50/50 (wt. %) nickeltitanium alloy (also known as nitinol) is being considered for examining its potential sensing capabilities. It is hypothesized that the nitinol-based sensor may outperform the current sensor because of two possible reasons: 1) increased surface area due to the formation of Ni(OH)2 complexes; and 2) a better working electrode with its comparatively more uniform geometry and surface. It may also obviate the need for the nickelfunctionalization step. Supplementing improved sensor design with step-wise cyclic voltammetry studies and chromatographic analyses (e.g., GCMS, HPLC, etc.) can help identify the reactions taking place between the Ni-TNA sensor and the target VOCs. It is believed that an increased understanding of these reaction mechanisms could facilitate more informed sample preparation steps to enhance VOC detection process. These analyses can be performed both in the solution and gas-phase settings to examine if there are any differences in the two measuring practices. These analytical steps can be integrated/extended to incorporate more breath samples, collected from a diverse cohort of CRC patients, healthy people, and potentially from other cancer patients. It is hoped that the introduction of effective practical approaches to design the sensor and perform 75 electrochemical measurements, combined with a powerful quantitative analytical tool, such as GC-MS, will enable the development of a robust, inexpensive and versatile sensing technology. CHAPTER 6 CONCLUSION It is a well-established understanding that research/innovation and implementation of sound medical practices go hand-in-hand with providing effective diagnosis and quality health care opportunities. This is particularly so and can prove to be very demanding in managing and/or treating cancers, such as colorectal cancer, where early and accurate diagnoses can quite literally make the difference between life and death. While there are some extant diagnostic methodologies for CRC, colonoscopy, by far, has been the most widely used technique to detect and treat CRC. Although sensitive, colonoscopy is not well-suited for rapid, resource-limited, point-of-care (POC) settings. Some of the drawbacks of this technique include low portability, invasiveness, extensive sample/patient preparation stages, and cost unaffordability. Consequently, there is a need to develop a simple and effective alternative diagnostic technique that can be used complementarily with colonoscopies and other detection methods for early diagnosis of CRC. The objective of the present studies is to demonstrate the feasibility of a TiO2 nanotube array (TNA) based sensing platform that can detect four critical breath-based volatile organic compounds (VOCs) in a colorectal cancer patient. To achieve both sensitivity and specificity in the detection of the VOCs, a combination of nickel deposition and cyclic voltammetry (CV), was used. Optimization of sensor response to the VOCs was 77 approached by investigating both the sensor fabrication and sample preparation parameters. Improvements in the nickel-functionalization of the TNA sensor could be achieved by applying a relatively higher current density during electrodeposition of nickel. Such an approach resulted in the significant distribution of relatively small nickel nanoparticles, in the form of nickel hydroxide, Ni(OH)2, on the TNA surface. These Ni(OH)2 nanoparticles played an integral role in the detection of the VOCs. Though the morphology of the nickel deposited TNA (Ni-TNA) sensor was somewhat heterogeneous, future steps such as controlling the electrodeposition setup, improving the TNA substrate’s regularity, and even using nitinol may further improve the performance of these sensors. Cyclic voltammetry, in solution, of the VOCs with the Ni-TNA sensor showed that the detection of the VOCs was highly dependent upon the solution pH. Careful measurements also showed that significant and specific detection could be achieved under acidic solution conditions. Not only did each VOC (except cyclohexane and methylcyclohexane) exhibit unique anodic peak potentials but also their mixtures could be identified by another set of anodic peak potentials. These measurements confirmed that the Ni-TNA sensor was sensitive to both compositions and relative concentrations of VOCs in the mixtures. Though the Ni-TNA sensor detected three of the VOCs under basic conditions as well, the weak nature of NH4OH prevented the formation of any discernable/pronounced anodic current peaks at specific potentials that were otherwise seen in acidic conditions. As CV is a powerful electrochemical technique that explores the oxidation-reduction (redox) behavior of electroactive species within a given potential window, this study also proposed potential redox reactions that might have occurred 78 between the Ni-TNA sensor and the VOCs. These solution-phase results were quite promising because they demonstrated the initial feasibility of the sensor for the detection of the VOCs in actual exhaled breath condensate from CRC patients. In other words, the synthesized Ni-TNA sensor showed reasonably good versatility for both solution and gaseous/vapor samples. In the case of gaseous/vapor samples, optimal detection could take place when the VOCs were dissolved in an acidic acetone medium. Each of the VOCs, except cyclohexane, could be detected by the Ni-TNA sensor (cyclohexane showed a somewhat opposite but meaningful trend). Under ambient conditions, only cyclohexane was detected, and when all the VOCs were mixed in acetone, the Ni-TNA sensor produced a completely different qualitative pattern. It is speculated that perhaps the interactions between the sensor and the VOCs are identical in solution and gas phases. These results strongly indicate that for the detection of gaseous/vapor phase VOCs, the addition of traces of HCl to an unreactive (with the Ni-TNA sensor) polar solvent may become a necessity for achieving better detection ability. It is also possible that separating/filtering the VOCs may be another alternative approach. This is a critical observation that may become integral in guiding future steps as far as sample preparation is concerned. Currently, no study is available in the literature on the benefit of the addition of a polar solvent, such as strong acid, as a possible way to improve the gas-phase detection ability of the TNA sensor. It is possible that these steps can be applied to real patient breath samples to enhance the VOC detection. The gas-phase experiments also validated that the applicability of grapheneembedded polylactic acid (G-PLA) conducting polymers as potential electrolytes for developing new and possibly improved electrochemical-based sensing devices. These 79 experiments have possibly opened up new avenues in the field of materials electrochemistry with various types of electron-conducting polymers being used for solidstate 3-electrode systems. They offer a tremendous advantage over metallic electrolyte systems, which tend to produce significant overpotentials or simply short circuit due to their significantly greater conductivities. Exposure of the Ni-TNA sensor to normal and VOC-spiked breath samples, obtained from a healthy subject also showed meaningful and opposite qualitative trends. Despite the stark differences between the trends, there is room for improvement in VOC detection in breath samples. It is speculated that deliberate modification of breath samples by adding strong acids, polar solvents, etc. may be critical in producing detection characteristics of the VOCs, similar to those seen in the acidic solutions. The performance characteristics of the present sensing platform have presented encouraging results and the following novel features (original contributions not found in published literature): • Elucidation of the effects of solution molarity and applied current density on the electrodeposition of nickel nanoparticles. • Demonstration of the superior effects of acidic conditions on the CV-based detection of four organic compounds, both under liquid and vapor settings. • Successful deployment of graphene-embedded polylactic acid (G-PLA) as the electrolyte for recording gas-phase CVs. • Successful application of an experimental setup using the G-PLA substrate, and a custom 3D printed acrylic sensing chamber optimized to collect gas/breath samples for CV measurements. 80 It is envisaged that the incorporation of suitable modifications in sensor design, its fabrication, deployment in recording quality experimental data, and elucidation of reaction mechanisms by combining a suitable analytical instrument (chromatography) will be required to design an effective yet inexpensive sensor for future clinical trials. These trials will aim to investigate the capability of this sensor as an inexpensive, hand-held, portable POC sensing device for rapid and accurate diagnosis of colorectal cancer. In a clinical setting, the designed sensor is intended to complement the existing methods of CRC diagnosis like colonoscopies. It is hoped that the availability of a breath test for CRC will not only give physicians more laxity to determine whether further, invasive testing is necessary but also decrease the economic and medical burdens associated with colonoscopies, and other diagnostic methods. Furthermore, it is worth mentioning that though sensing technology designed here has been proposed for breath-based VOC detection, the catalytic nature of Ni(OH)2 on the surface may allow detection of colorectal cancer VOCs from blood, stool and urine samples as well. Should the sensing technology be successful after rigorous sensitivity and specificity testing, it may serve as a complementary diagnostic tool for colorectal cancer, providing early and accurate diagnoses to reduce patient mortality, globally. REFERENCES [1] A. C. Society, “Colorectal Cancer Facts & Figures 2017-2019,” of nickel Atlanta, 2017. [2] M. Arnold, M. S. Sierra, M. Laversanne, I. Soerjomataram, A. Jemal, and F. Bray, “Global patterns and trends in colorectal cancer incidence and mortality,” Gut, vol. 66, no. 4, pp. 683–691, 2017. [3] H. Amal et al., “Breath testing as potential colorectal cancer screening tool,” Int. J. Cancer, vol. 138, no. 1, pp. 229–236, 2016. [4] P. Vega, F. Valentín, and J. Cubiella, “Colorectal cancer diagnosis: Pitfalls and opportunities,” World J. Gastrointest. Oncol., vol. 7, no. 12, p. 422, 2015. [5] The Hans India, “Using liquid biopsy to monitor colorectal cancer,” Online, 2017. [Online]. Available: https://www.thehansindia.com/posts/index/Health/2017-1213/Using-liquid-biopsy-to-monitor-colorectal-cancer-/345065. [Accessed: 22-Feb2019]. [6] Centers for Disease Control and Prevention, “CDC - Economic Evaluation of CDC’s Colorectal Cancer Control Program,” 2017. [Online]. Available: https://www.cdc.gov/cancer/dcpc/research/articles/crccp_evaluation.htm. [Accessed: 17-Jan-2019]. [7] E. Westenbrink et al., “Detection of lung, breast, colorectal, and prostate cancers from exhaled breath using a single array of nanosensors,” Int. J. Cancer, vol. 138, no. 1, pp. 229–236, 2014. [8] D. Bhattacharyya, P. Kumar, S. K. Mohanty, Y. R. Smith, and M. Misra, “Detection of four distinct volatile indicators of colorectal cancer using functionalized titania nanotubular arrays,” Sensors (Switzerland), vol. 17, no. 8, 2017. [9] G. Konvalina and H. Haick, “Sensors for breath testing: from nanomaterials to comprehensive disease detection,” Acc. Chem. Res., vol. 47, no. 1, pp. 66–76, Jan. 2014. [10] G. Peng et al., “Diagnosing lung cancer in exhaled breath using gold nanoparticles,” Nat. Nanotechnol., vol. 4, no. 10, pp. 669–673, 2009. 82 [11] G. Shuster et al., “Classification of breast cancer precursors through exhaled breath,” Breast Cancer Res. Treat., vol. 126, no. 3, pp. 791–796, 2011. [12] D. F. Altomare et al., “Exhaled volatile organic compounds identify patients with colorectal cancer,” Br. J. Surg., vol. 100, no. 1, pp. 144–150, 2013. [13] N. Peled et al., “Volatile fingerprints of cancer specific genetic mutations,” Nanomed. Nanotechnol., Biol. Med., vol. 9, no. 6, pp. 758–766, Aug. 2013. [14] N. Peled et al., “Non-invasive breath analysis of pulmonary nodules,” J. Thorac. Oncol., vol. 7, no. 10, pp. 1528–1533, 2012. [15] C. Wang et al., “Blood volatile compounds as biomarkers for colorectal cancer,” Cancer Biol. Ther., vol. 15, no. 2, pp. 200–206, 2014. [16] N. K. H. de Boer et al., “The scent of colorectal cancer: Detection by volatile organic compound analysis,” Clin. Gastroenterol. Hepatol., vol. 12, no. 7, pp. 1085–1089, 2014. [17] H. Amal et al., “The scent fingerprint of hepatocarcinoma: In-vitro metastasis prediction with volatile organic compounds (VOCs),” Int. J. Nanomedicine, vol. 7, pp. 4135–4146, 2012. [18] T. G. De Meij et al., “Electronic nose can discriminate colorectal carcinoma and advanced adenomas by fecal volatile biomarker analysis: Proof of principle study,” Int. J. Cancer, vol. 134, no. 5, pp. 1132–1138, 2014. [19] H. Haick, Y. Y. Broza, P. Mochalski, V. Ruzsanyi, and A. Amann, “Assessment, origin, and implementation of breath volatile cancer markers,” Chem. Soc. Rev., vol. 43, no. 5, pp. 1423–1449, 2014. [20] G. Peng et al., “Detection of lung, breast, colorectal, and prostate cancers from exhaled breath using a single array of nanosensors,” Br. J. Cancer, vol. 103, no. 4, pp. 54–551, 2010. [21] E. Westenbrink et al., “Development and application of a new electronic nose instrument for the detection of colorectal cancer,” Biosens. Bioelectron., vol. 67, pp. 733–738, 2015. [22] J. Pereira et al., “Breath analysis as a potential and non-invasive frontier in disease diagnosis: An overview,” Metabolites, vol. 5, no. 1, pp. 3–55, 2015. [23] T. Oguma et al., “Clinical contributions of exhaled volatile organic compounds in the diagnosis of lung cancer,” PLoS One, vol. 12, no. 4, pp. 1–10, 2017. [24] M. Phillips et al., “Breath biomarkers of active pulmonary tuberculosis,” 83 Tuberculosis, vol. 90, no. 2, pp. 145–151, 2010. [25] R. P. Arasaradnam et al., “Detection of colorectal cancer (CRC) by urinary volatile organic compound analysis.,” PLoS One, vol. 9, no. 9, p. e108750, 2014. [26] A. V. Biankin et al., “Overexpression of p21WAF1/CIP1 is an early event in the development of pancreatic intraepithelial neoplasia,” Cancer Res., vol. 61, no. 24, pp. 8830–8837, 2001. [27] S. R. Markar et al., “Breath volatile organic vompound profiling of colorectal cancer using selected ion flow-tube mass spectrometry,” Ann. Surg., vol. 269, no. 5, pp. 903–910, 2017. [28] D. Bhattacharyya, Y. R. Smith, M. Misra, and S. K. Mohanty, “Electrochemical detection of methyl nicotinate biomarker using functionalized anodized titania nanotube arrays,” Mater. Res. Express, vol. 2, no. 2, p. 025002, 2015. [29] Nanalyze, “9 Companies Pioneering Breath Diagnostics - Nanalyze,” 2018. [Online]. Available: https://www.nanalyze.com/2018/03/9-breath-diagnosticscompanies/. [Accessed: 27-Feb-2019]. [30] Owlstone Medical, “FAIMS technology.” [Online]. Available: https://www.owlstonemedical.com/science-technology/faims-technology/. [Accessed: 27-Feb-2019]. [31] Fujitsu, “Fujitsu Successfully Develops a Breath Sensor Device That Quickly Measures Gas Components in People’s Breath - Fujitsu Global,” 2016. [Online]. Available: http://www.fujitsu.com/global/about/resources/news/pressreleases/2016/0418-01.html. [Accessed: 27-Feb-2019]. [32] Rapid Biosensor, “TB Breathalyser,” 2015. [Online]. Available: http://www.rapidbiosensor.com/tbbreathalyser.asp. [Accessed: 27-Feb-2019]. [33] F. Cosandey, G. Skandan, and A. Singhal, “Materials and processing issues in nanostructured semiconductor gas sensors,” JOM-e, vol. 52, no. 10, pp. 22–25, 2000. [34] R. J. B. Balaguru and B. G. Jeyaprakash, “Mimic of a gas sensor, metal oxide gas sensing mechanism, factors influencing the sensor performance and role of nanomaterials based gas sensors,” NPTEL–Electrical Electron. Eng. Nanodevices, vol. 3, no. 4, pp. 1–30, 2004. [35] G. Zonta et al., “Detection of colorectal cancer biomarkers in the presence of interfering gases,” Sens. Actu., B Chem., vol. 218, pp. 289–295, 2015. [36] C. Malagù et al., “Chemoresistive gas sensors for the detection of colorectal cancer 84 biomarkers,” Sensors (Switzerland), vol. 14, no. 10, pp. 18982–18992, 2014. [37] A. Bard and L. Faulkner, Electrochemical Methods: Fundamentals and Applications 2nd Edition, 2nd ed. New York: John Wiley & Sons, Inc., 2001. [38] S. Lin, D. Li, J. Wu, X. Li, and S. A. Akbar, “A selective room temperature formaldehyde gas sensor using TiO2 nanotube arrays,” Sens. Actu., B Chem., vol. 156, no. 2, pp. 505–509, 2011. [39] V. Galstyan, E. Comini, G. Faglia, and G. Sberveglieri, “TiO2 nanotubes: Recent advances in synthesis and gas sensing properties,” Sensors (Switzerland), vol. 13, no. 11, pp. 14813–14838, 2013. [40] J. a. Rodríguez et al., “High water-gas shift activity in TiO2 supported Cu and Au nanoparticles: Role of the oxide and metal particle size,” J. Phys. Chem. C, vol. 113, no. 110, pp. 7364–7370, 2009. [41] Z. Li, D. Ding, Q. Liu, and C. Ning, “Hydrogen sensing with Ni-doped TiO2 nanotubes,” Sensors (Switzerland), vol. 13, no. 7, pp. 8393–8402, 2013. [42] 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. Commun., vol. 49, no. 35, pp. 3652–3654, 2013. [43] V. Gaur, A. Sharma, and N. Verma, “Catalytic oxidation of toluene and m-xylene by activated carbon fiber impregnated with transition metals,” Carbon N. Y., vol. 43, no. 15, pp. 3041–3053, 2005. [44] S. Yolcular and Ö. Olgun, “Ni/Al2O3 catalysts and their activity in dehydrogenation of methylcyclohexane for hydrogen production,” Catal. Today, vol. 138, no. 3–4, pp. 198–202, 2008. [45] H. R. Cooper and H. W. Melville, “The kinetics of the autoxidation of n-decanal, part I. The mechanism of reaction,” J. Chem. Soc., no. 0, pp. 1984–1993, 1951. [46] M. G. Hosseini, M. 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. [47] H. Cheshideh and F. Nasirpouri, “Cyclic voltammetry deposition of nickel nanoparticles on TiO2 nanotubes and their enhanced properties for electro-oxidation of methanol,” J. Electroanal. Chem., vol. 797, pp. 121–133, July 2017. [48] Z. Wu, X.-L. Huang, Z.-L. Wang, J.-J. Xu, H.-G. Wang, and X.-B. Zhang, “Electrostatic induced stretch growth of homogeneous β-Ni(OH)2 on graphene with enhanced high-rate cycling for supercapacitors.,” Sci. Rep., vol. 4, p. 3669, 2014. 85 [49] G. Yao et al., “Effects of electrodeposition electrolyte concentration on microstructure, optical properties and wettability of ZnO nanorods,” J. Electrochem. Soc., vol. 162, no. 7, pp. D300–D304, Apr. 2015. [50] I. Weissman and D. Aurbach, “Nonaqueous electrochemistry,” in Nonaqueous Electrochemistry, 1st ed., New York: CRC Press, 1999. [51] J. F. Coetzee, “Current trends in electrochemical studies of nonaqueous solutions,” Pure AppI. Chem, vol. 58, no. 8, pp. 1091–1104, 1986. [52] M. Zain, J. Greenberg, J. Herrera, R. N. Cataneo, M. Phillips, and S. Krishnan, “Variation in volatile organic compounds in the breath of normal humans,” J. Chromatogr. B Biomed. Sci. Appl., vol. 729, no. 1–2, pp. 75–88, 2002. [53] A. Gholizadeh et al., “Toward point-of-care management of chronic respiratory conditions: Electrochemical sensing of nitrite content in exhaled breath condensate using reduced graphene oxide,” Microsystems Nanoeng., vol. 3, pp. 17-22, 2017. [54] A. A. Karyakin, S. V. Nikulina, D. V. Vokhmyanina, E. E. Karyakina, E. K. H. Anaev, and A. G. Chuchalin, “Non-invasive monitoring of diabetes through analysis of the exhaled breath condensate (aerosol),” Electrochem. Commun., vol. 83, pp. 81–84, Sep. 2017. [55] E. Mohamadian, A. Shayanfar, M. Khoubnasabjafari, V. Jouyban-Gharamaleki, S. Ghaffary, and A. Jouyban, “Analysis of deferiprone in exhaled breath condensate using silver nanoparticle-enhanced terbium fluorescence,” Anal. Methods, vol. 9, no. 38, pp. 5640–5645, 2017. [56] K. E.M., L. A.S., T. A.S., and B. P.K., “Exhaled Breath Condensate: Technical and Diagnostic Aspects,” Sci. World J., vol. 2015, pp. 1–25, 2015. [57] S. Dodig and I. Čepelak, “Exhaled breath condensate - From an analytical point of view,” Biochem. Medica, vol. 23, no. 3, pp. 281–295, 2013. [58] 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. Ind. Eng. Chem., vol. 20, no. 1, pp. 100–107, 2014. [59] P. R. Makgwane and S. S. Ray, “Efficient room temperature oxidation of cyclohexane over highly active hetero-mixed WO3/V2O5 oxide catalyst,” Catal. Commun., vol. 54, pp. 118–123, Sep. 2014. [60] A. Alshammari, A. Köckritz, V. N. Kalevaru, A. Bagabas, and A. Martin, “Direct oxidation of cyclohexane to adipic acid using nano-gold catalysts,” Appl. Petrochemical Res., vol. 2, no. 3–4, pp. 61–67, 2012. 86 [61] B. Sarkar et al., “Room temperature selective oxidation of cyclohexane over Cunanoclusters supported on nanocrystalline Cr2O3,” Green Chem., vol. 14, no. 9, pp. 2600, 2012. [62] L. Sasso, P. Vazquez, I. Vedarethinam, J. Castillo-León, J. Emnéus, and W. E. Svendsen, “Conducting polymer 3D microelectrodes,” Sensors, vol. 10, no. 12, pp. 10986–11000, 2010. [63] G. Inzelt, “Conducting polymers: past, present, future György,” J. Electrochem. Sci. Eng., vol. 8, no. 1, pp. 3–37, 2018. [64] S. J. Park, C. S. Park, and H. Yoon, “Chemo-electrical gas sensors based on conducting polymer hybrids,” Polymers (Basel)., vol. 9, no. 5, 2017. [65] H. Bai and G. Shi, “Gas sensors based on conducting polymers,” Sensors, vol. 7, no. 3, pp. 267–307, 2007. [66] A. N. Mallya, R. Kottokkaran, and P. C. Ramamurthy, “Conducting polymer-carbon black nanocomposite sensor for volatile organic compounds and correlating sensor response by molecular dynamics,” Sens. Actu., B Chem., vol. 201, pp. 308–320, 2014. [67] E. Zampetti et al., “A high sensitive NO2 gas sensor based on PEDOT-PSS/TiO2 nanofibres,” Sens. Actua., B Chem., vol. 176, no. 2, pp. 390–398, 2013. [68] B. Lakard, S. Carquigny, O. Segut, T. Patois, and S. Lakard, “Gas sensors based on electrodeposited polymers,” Metals (Basel)., vol. 5, no. 3, pp. 1371–1386, 2015. [69] J. B. Goodenough and P. Singh, “Review—Solid electrolytes in rechargeable electrochemical cells,” J. Electrochem. Soc., vol. 162, no. 14, pp. A2387–A2392, 2015. [70] G. M. Wu, S. J. Lin, and C. C. Yang, “High performance composite solid polymer electrolyte systems for electrochemical cells,” J. Power Sources, vol. 244, pp. 287– 293, 2013. [71] C. W. Foster et al., “3D Printed Graphene Based Energy Storage Devices,” Sci. Rep., vol. 7, no. 1, p. 42233, 2017. [72] D. A. C. Brownson and C. E. Banks, The Handbook of Graphene Electrochemistry, 1st ed. London: Springer-Verlag, 2014. [73] I. S. Bayer, “Thermomechanical properties of polylactic acid-graphene composites: A state-of-the-art review for biomedical applications,” Materials (Basel)., vol. 10, no. 7, pp. 1–33, 2017. 87 [74] X. An, H. Ma, B. Liu, and J. Wang, “Graphene oxide reinforced polylactic acid/polyurethane antibacterial composites,” J. Nanomater., vol. 2013, pp. 1–7, 2013. [75] Y. Fu, L. Liu, and J. Zhang, “Manipulating dispersion and distribution of graphene in PLA through novel interface engineering for improved conductive properties,” ACS Appl. Mater. Interfaces, vol. 6, no. 16, pp. 14069–14075, 2014. [76] C. Willis, Y. Saffary, N. M. Yusof, M. Misra, and S. K. Mohanty, “Effect of Anodization Temperature on Performance of Metal-Functionalized TiO2 Nanotube Sensors for Biomedical Applications,” Poster presented at: Department of Chemical Engineering Symposium, University of Utah, Sep 2017, Salt Lake City. [77] S. R. Markar, J. Lagergren, and G. B. Hanna, “Research protocol for a diagnostic study of non-invasive exhaled breath analysis for the prediction of oesophagogastric cancer,” BMJ Open, vol. 6, no. 1, pp. 8–11, 2016.