Harry E. Ruda
University of Toronto
January 21, 2025
This report was prepared as a result of being approached by Micromem Technologies to assess the particular needs of Chevron regarding evaluating sensor technologies and modalities relevant to their specific needs. Our preliminary considerations, as embodied in this report, demonstrate that Chevron’s needs for monitoring and sensing of brackish wastewaters in and around the vicinity of oil wells and their associated flood zones, may be accommodated by our unique ultrasensitive nanowire based multi-modal sensor platform.
This paper examines current and emerging sensor technologies for detecting water contamination caused by oil well operations and surrounding flood zones. It evaluates the suitability of these sensors for real-time monitoring, their sensitivity, specificity, cost-effectiveness, and potential deployment in field conditions. Special attention is given to their ability to detect hydrocarbons, heavy metals, and other oil-related pollutants in diverse environmental conditions. Against this background, we present our unique nanotechnology-based sensing platform that can provide a broad range of sensing solutions suitable for addressing the needs for this specific wastewater contamination problem and enable a path to efficient environmental remediation.
Our team is led by Professor Harry Ruda, the Stanley Meek Chair Professor in Nanotechnology at the University of Toronto, and Director of the Centrefor Advanced Nanotechnology - Canada’s first centre for nanotechnology. After obtaining his BSc from Imperial College in 1979, he joined MIT, and from 1979 to 1982 he worked on the optical and transport properties of II-VI materials and obtained his PhD in 1982. Between 1982 and 1984 he developed one of the first theories for electron transport in selectively doped two-dimensional electron gas heterostructures while working as an IBM post doctoral fellow, and in 1984 he joined 3M Corporation as a senior scientist developing some of the first models for electronic transport and optical properties of wide band gap II-VI semiconductors, while being a key member of the blue laser team. In 1989 he joined the University of Toronto, rising through the ranks to his current Full Professor appointment.
Professor Ruda has published over 300 publications in international refereed journals (with over 8,344 citations; h=45), has co-authored 4 books and has 14 patents. Professor Ruda’s research interestsfocus on nanostructures with application in areas such as sensing andquantum technology.
Harry Ruda serves on the National Science and Engineering Council of Canada and on other government panels including those of the DOE, EPA and NSF in the US, and the RAE and EPSRC in the UK. He serves as a board member of Canadian Solar Inc (CSIQ: NASDAQ). He has also served on the editorial boards of: Journal of Applied Physics, Applied Physics Letters, Journal of Nanoscience and Nanotechnology, Journal of Materials Science: Materials in Electronics, Nanotechnology Research Letters, and Nano-Micro Letters and is currently Chief Editor of IET-CDS. He is a Fellow of the Royal Society of Canada, Fellow of Institute of Physics, Fellow of the Institute of Nanotechnology, Fellow of the Institutionof Engineering and Technology, and Fellow of the Canadian Academy of Engineering.
Prof. Ruda is well known for his work on nano-materials engineering and nanowire-based sensors development (e.g., [8], [11], [21]). Prof. Ruda also was involved in previous sensors projects such as an NSERC sponsored project on position and motion sensing with Micromem, and on anoptical microsystem for plasma emission-based gas sensors that resulted in aspin-off company, Optomem Sensors. Prof. Ruda has 285 publications in international refereed journals (>7,890 cites and h=42), has co-authored 4 books and has 14 patents. He has successfully managed over one hundred grants with industrial partners.
Prof. Ruda’s team working on semiconductor-based nanostructure sensors as pertinent to the current gas and aqueous sensors includes:
An ideal chemical sensor exhibits excellent responsivity, response/recovery times, enhanced sensitivity, and is highly selective with strong stability/repeatability, while still being low-cost and easily scalable to particular applications. A variety of candidate chemical sensors exist which, to name a few, includes: resistive sensors [1], [2] where the electrical conductance/resistance of the device is modulated upon introduction of a chemical species; optical based sensors [3], [4] where processes such a slight absorption or luminescence indicates the presence of volatile compounds; acoustic wave sensors [5], [6] which rely on resonant elements where measured frequency shifts correspond to analyte adsorption. In our case, we focus on furthering development of our multi-nano wire field-effect transistor (NWFET)[7], [8] sensors – where resolving response parameters is determined by perturbations in device conductance (i.e., a resistive sensor). Nanoscale device architectures commonly face issues where stability is notably important to operation. Low signal levels require hefty investment into bulky, specialized measurement equipment, restricting the possibility for portability and miniaturization. Our nanowire-based sensors, with multiple wires running in parallel, amplify the device current which offers significant noise reduction, eliminating size-based restrictions. Furthermore, InAs NWFET’s are exceptionally sensitive to local charge fluctuations at the surface [9]–[11]. For chemical sensing the semiconducting channel acts to detect the presence of various chemical compounds through electrostatic interactions between the adsorbate molecule and electrons at the NW surface. Typically, this interaction may be described by two processes, physisorption and chemisorption [12], [13]. Physisorption occurs when molecules adsorb and interact via Van der Waalsforces, instead of being chemically bound. This process is normally reversible (due to weak binding ~10 – 100 meV) and is especially useful when describing sensing of ultralow concentrations (<1 ppm) of chemical species. Chemisorption, on the other hand, occurs when molecules chemically react with the surface and form a bond (strong binding ~1 – 10 eV) – it is irreversible,and responses occur over much larger time scales. For NWFET chemical sensors, one can distinguish these two processes by examining the time scale in which the NW conductance varies and by monitoring change in conductance relative to a pre-defined baseline. Our fabricated devices operate in both regimesand have been formerly studied for ethanol and NO2 while operating at room temperature (compared with typical ceramic/defect sensors requiring programmed heating), another attractive feature of these devices [8], [11]. Beyond sensing of gaseous species, additional fabrications steps have been instituted for liquid-based sensing all while keeping the same form factor. This is advantageous as our NWFET sensing platform can effectively be tailored for multiple molecule sensing applications in differing mediums. Hence, it is highly plausible to outfit our devices as a portable unit with multiplexing capabilities. Using Raspberry Pi single-board computers and high-resolution on-board Analog/Digital converters, we can keep the cost low and efficiency loss minimal (due to the highspeed of the boards).
Some important previous results with regards to wastewater characterization include developing nanowire/nanotube based chemical oxygen demand (COD) sensors by a photoelectrochemical approach.
This was accomplished by measuring net photocurrents between a blank solution (i.e. no organics present) and a solution with organics. Various concentrations of oxalic acid in 0.1M Na2SO4 were used as organic solutions for determining the detection limits and sensitivities of the different samples. To calibrate the photoelectrochemical system, the COD of the organic solutions was determined by the conventional dichromate method with a Hach DRB200 digestion reactor coupled with a Hach DR870 colorimeter. The measured net photocurrents were then plotted against these COD values (Fig. 1 left).
Alinear fit of the data was applied to determine the COD at which the detection limit was reached. The saturation of the photocurrent at higher organic concentrations could be attributed to the occupation of the NW surface sites by organic molecules. Once saturation of the surface was achieved, the proportionality was no longer observed, and the net current deviated from the linear behaviour. As the average pore size increased with anodization voltage, this saturation process occurred at larger concentrations of oxalic acid. The sensitivity (defined as the produced photocurrent per mg/L of COD) is representative of the accuracy of the sensor and was significantly affected by the sample morphology as well (Fig. 1 right).
.png)
Role of pH was also studied for NW sensors and Fig XC below shows one result for an InAs NW FET in a PBS buffer. As is evident, these FETS areable to sensitively monitor pH in solution.
Similar structures were also shown to provideexcellent measure of the solutionc onductivity (Fig. 3, Upper) and indeed, we showed how a simple microfluidic system could be developed for online monitoring (Fig. 3,Lower).
.jpg)
Fig 3: InAs based NW FETs measuring solutionconductivity (Upper) and microfluidic circuitbased on PDMS for inline monitoring (Lower).
Our progress with NW FET sensors in gas phase sensing has also been remarkable, and show allow for stand off sensing of VOCs. For example, in Fig. 4 below, we show how ethanol and DMMP can be detected down to parts per billionwith these sensors. The BET isotherm modeling for each molecule provides an exceptional fit showing that there is an ability to use the interaction strength parameters to distinguish mixtures of both.
Our preliminary work on machine learning models using both principal component analysis and linear discriminant analysis suggest that we should be able to use such an approach to differentiate similar VOCs from each other with these sensors.
Oil well operations can cause water contamination due to the operations themselves, as well as the associated flood zones in the vicinity of the wells. This section reviews current and emerging sensor technologies for detecting this water contamination, evaluating their suitability for real-time monitoring, and including performance assessments (especially their sensitivity and specificity) as well as cost-effectiveness, and potential deployment in field conditions. Special attention is given to their ability to detect hydrocarbons, heavy metals, and other oil-related pollutants in diverse environmental conditions. Contaminants such as hydrocarbons, heavy metals, and salts can leach into groundwater and surface water, endangering ecosystems and human health. Effective monitoring solutions are critical for timely detection and mitigation of contamination. Below we review sensor technologies tailored to these challenges, including their operational principles, advantages and limitations.
The primary contaminants of concern include:
Research opportunities include the development of multi-modal sensors and enhanced algorithms for predictive modeling (see section 4 below). Selecting appropriate sensors for monitoring water contamination in oil wells and flood zones dependson specific application requirements, including sensitivity, specificity, and deployment conditions. Advances in nanotechnology and remote sensinghold promise for next generation monitoring systems capable of addressing current limitations and ensuring environmental safety.
The proposed work would focus on multimodal nanotechnology-based sensors with enhanced algorithms for sensitive differentiation of chemical contaminants:
InAs NWs are grown by molecular beam epitaxy (MBE) via the vapour-liquid-solid (VLS) mechanism [14], a common method for growing single-crystalline NWs with superior quality and lower background dopant concentration as compared with other growth methods [15]. MBE is carried out in-house with our commercial system (ATC-EP3) that provides a clean (p~3.5x10-10Torr) environment and high purity sources for controlled depositions by dual filament cells (with As in a valved cracker). GaAs (111) substrates arethermally cleaned within the system and a thin Au film is deposited via electronbeam (eB) evaporation followed by thermal annealing to form Au nanoparticles inaccordance with the VLS method for NW growth. Itself, MBE allows for strict control of NW dimensions, in which the physical properties influence various electrical properties of the wires – e.g., differences in electron effective mass depends on the crystal phase, varying ON-OFF ratio for NWFETs, and the presence of quantum confinement effects in ultrathin NWs [15], [16]. Ergo, optimising InAs NW growth via MBE will lead to a set of conditions favorable for device operation and sensor responsivity.
Device fabrication will be carried out in our modular cleanroom unit (class 100/1000) where the process is confined to this well-controlled environment (aside from MBE & contact metallization occurring in an exterior vacuum chamber). Prepared donor InAs substrates from MBE are used for NW transfer to acceptor p+-Si/SiO2 (100 nm) substrates using one of two methods:contact-printing (CP) or drop casting. Optimisation of the latter is an underlying task where we have determined the former method, which is sensitive to vertically applied pressure and thus requires ample precision, is susceptible to scratching the surface oxide as well as deposits unwanted material. For device fabrication to move towards a manufacturable process, reducing the number of steps and complexity is key.
Drop casting of NWs is a simple, effective way of reducing processing complexity. Refinement of this process will require an apparatus specific to drop casting and NW alignment by way of rapidly spinning the substrate, measurement of forces relating directional alignment, and measuring NW surface density via scanning electron microscopy (SEM) image analysis. We may then arrive at an optimal set such that consistent quality and operation standards are met. Post NW transfer, defining contacts will be done by photolithography (PL) with a Karl Suss double-sidedmask aligner. Our photomask accommodates seven multi-NWFET devices (refer to Fig. 5 for illustrative purposes) on a single chip such that multiple devices may be measured in tandem. Our in-house thermal/eB evaporation system is used for contact metallization [Ti (10 nm)/Au (100 nm)] where prior surface passivation is completed via dipping the NWFET chip in an (NH4)2Sx solution [17] to ensureohmic contact formation between the metal and InAs.
The chips are then mounted on a wire bonder (Kulickeand Soffa) where Au wire is bonded from the substrate contacts to a leadless 28-pin ceramic carrier. Further processing steps are then carried out to realise aqueous devices wherea series of tubes are bound to the device substrate. These act as liquid reservoirs where an innermost glass tube holds 10 μL and isolates the seven NWFET devicesand an adhesive covers the outer bonding pads. A second & third reservoir will be added to increase capacity (0.1 mL), allow for electrodes to be introduced into the analyte solution, and protect against overflow. Significant efforts will also be under way in understanding and applying surface functionalization schemes for bio-sensing studies. Refinement of all processes will be done iteratively over the proposed years where optimal parameters will be stored/shared between teammates and applied to enhance fabrication.
To determine specific sensor responses, various analytes will be chosen for both sensing platforms. Building a library of responses is imperative in resolving specificity parameters (based upon responsivity, time constants, sensitivity & limits of detection) and applying those to a multiplexing scheme. Analyte selection and response extraction will be based on choosing families of molecules with similar structures but varying underlying properties (e.g., physical size, charge, dipole moment, chemical reactivity) and measuring electrical responses. For gaseous phases, a custom-built environmental chamber will be used to introduce varying concentrations of gasdiluted in N2 through a mixer (Environics 6100) andelectrical I-V measurements (apply Vds andmeasure Ids) are done using a high-precisionmulti-instrument system (Moku Lab). Liquid phase measurements will be initiated with a Keithley 4200 source-measurement unit system (w/4x SMUs) which is deployed as a quasi-potentiostat where electrodes are used to maintain a set liquid gate potential and device current is measured as done in the gassensing setup. A caveat of sensing in liquid is the inherent irreversibility of chemical binding events due to the functionalized surface, which requires a newset of devices for each experiment. We plan to tackle this issue by testing etching schemes that remove a small portion of the InAs surfaceoxide layer (where binding occurs) to desorb chemisorbed molecules. One such way of doing this is with our passivation solution ((NH4)2Sx) where different concentrations have varying etching effects on the InAs wires. In tandem with sensing experiments, data will be analyzed toresolve a definitive set of parameters that determines selectivity, as discussedabove. The gathered datasets will be post processed and PCA, a statisticalprocedure that represents multivariate data as smaller, grouped sets, will beimplemented.
To bridge the gap between lab-based experiments to industrial application, a design for portablea low- cost, miniaturized system is required. The design will include multipleanalog/digital (AD/DA) converters which provide high-resolution voltage outputs/inputs with a Raspberry Pi 3B+ single-board computer (RPI) as the base. This setup is propitious as size (5 x 10 cm, < 2 lbs) and cost (~$250 CAD for hardware) are minimized. Furthermore, all software/hardware used will be fully open source, allowing us to expand and modify any components readily. Python will be the programming language of choice to develop front-end software which will tie into a back-end cloud database (Amazon Web Services, Microsoft Azure, GoogleCloud) for storage and subsequent PCA analysis as discussed in Task B. Post processing can be done both on the RPI board or on the back end depending on location and a stable connection to the cloud database. Assembly and testing of the unit will begin by incorporating single chips for general FET response analysis (gate pulse transients, output/transfer curves) to optimise and ensure proper device operation. For aqueous based sensors, extra components will be incorporated to accommodate the sensing reservoirs and external electrodes used in said setup. Expansion to using these devices assensors will follow a similar, where measurements using the developed PCAtechnique will be incorporated.
Quantitative data collected from above will be used to determine and validate present sensing mechanisms based on molecular adsorption and relatedphysi/chemisorption events. Modeling the surface binding events that take place, and relating to device responses, can largely be applied through phenomenologicalrepresentations.
For gas phase sensing, physisorption responses in our system scan be attributed to a uniform dipole potential of physisorbed molecules and the relative scattering contribution of potential fluctuations caused by them. Chemisorption, due to stronger bonding and gradual sensor responses, will be treated as adding an ethanol adsorption an acceptor/donor surface state to the NW, where chemical bonds between the molecule and the surface dictate the amount of charge at the NW surface. Descriptions of both will be provided by a dynamic carrier transport model that relates interactions between adsorbate molecules and electrons at the surface. Connecting the sensor response to a given concentration will be done by applying Langmuir [18] and BET [19] isotherms where a description for low and high coverage regimes will be provided. These models have been applied to ethanol and NO2 adsorption (see Fig. 7) and will be reapplied to each unique response after evaluation of the relevant sensing mechanism.
In the liquid phase, a different approach must be taken to account for the interaction between the semiconducting channel, liquid solution, and molecules/ions. Just as in the gas phase, a net charge will be presentat the semiconductor-liquid interface where we can model the final charge density based on the number of amphoteric sites and local density ofions.
The altered charge density at the interface leads to perturbations in the semiconductors surface potential, which is reflected by changes in the active materials conductance. The Nernst-Planckequation is particularly useful for describing time-dependent behavior of ions in the presence of an electric field and concentration gradient, which is extended to our NWFETs. Described above is the site-binding model [20] which will be applied to evaluate device sensitivity and plausible enhancement strategies. We have applied this model in the past to sensing of FeCN ions (Fig. 8) where Vgsweep rates were found to be a critical parameter for obtaining optimal device sensitivity.
In addition, we will adapt our previous work on COD and pH inaqueous solutions to this platform. For exploring ionic conductivity of different electrolytes, key will be to measure threshold voltage shift with buffered solutions of various molar concentrations (e.g.,NaCl, KCl, Na2HPO4, and KH2PO4) – we will also investigate the dependence of Vthon ionic strength as well as the Debye screening length of the electrolyte. Secondly, we can explore the ionic conductivity of specific ions - for example, we can selectively examineNa+ by measuring Vth in KCl solution and incrementally adding Na+by adding NaCl while maintaining the ionic strength of the overall electrolyte. In addition, we can measure Vthin NaCl solution at various temperatures that changes the mobility of the Na+ ions and thus the ionic conductivity.
This White Paper outlined current state of the art sensor technologies for detecting contaminants in the wastewater in the vicinity of oil well operations and surrounding flood zones. The suitability of these sensors for real-time monitoring, their sensitivity, specificity, cost-effectiveness, and potential deployment in field conditions are discussed. In particular, attention is given to their ability to detect hydrocarbons, heavy metals, and other oil-related pollutants in diverse environmental conditions.
Against this background, we present our unique nanotechnology-based sensing platform that can provide a broad range of sensing solutions suitable for addressing the needs for this specific wastewater contamination problem and enable a path to efficient environmental remediation. Our platform is based on an ability to detect specific contaminant to levels as low as ppb in a compact portable and cost- effective format. Moreover, these features can enable multi-sensor networked systems to be deployed for real time intelligent (with sensor fusionand machine learning algorithms) information. Such an approach can take the place of single point HPLC type expensive bulky slow and sampled systems for inexpensive and intelligent near real-time monitoring and in-principle, control. The latter could be through remediation using advanced electrochemicaloxidation technology monitored by the approach above. The key next step would be to initiate development of specific species sensors to establish sensitivity, specificity and limits of detection, as well as other parameters including sensor temporal dynamics and packaging suitable for field use.
