Moving Electrons Purposefully through Single Molecules and Nanostructures: A Tribute to the Science of Nongjian Tao

Abstract

Electrochemistry intersected nanoscience 25 years ago when it became possible to control the flow of electrons through single molecules and nanostructures. Many surprises and a wealth of understanding were generated by these experiments. Nongjian Tao was among the pioneering scientists who created the methods and technologies for advancing this new frontier. Achieving deeper understanding of charge transport in molecules and low-dimensional materials was the first priority of his experiments, but he also succeeded in discovering applications in chemical sensing and biosensing for these novel nanoscopic systems. In parallel with this work, the investigation of a range of phenomena using novel optical microscopic methods was a passion of Nongjian and his students. This article is a review and an appreciation of some of his many contributions with a view to the future.

Graphical Abstract

Professor Nongjian Tao (“NJ”, Figure 1) of Arizona State University (ASU) passed away on March 15, 2020 at the age of 56.¹ He was an extraordinary scholar and scientist whose insightful and influential work encompassed a wide range of nanoscale phenomena in electrochemistry, surface science, chemical and biological analysis, electrical engineering, and quantum physics. In this scientific remembrance we recount some of his beautiful and impactful science.

Figure 1.

Professor Nongjian Tao in his laboratory in 2008. Image credit: The Biodesign Institute at Arizona State University.

This article is organized chronologically, representing the major areas in which his research group worked. His contributions in these areas overlapped in time, as he developed new research thrusts in his group, especially as his group expanded and diversified at ASU, so this chronology is approximate. In fact, NJ progressively added new research areas without abandoning many over his career. At Florida International University and ASU in the early 1990s, he pioneered scanning tunneling microscopy (STM) and atomic force microscopy (AFM) investigations of molecular monolayers in electrochemical environments.^2–9 This work led to a brief period, in the late 1990s and early 2000s, during which he investigated transport through single metal^10–12 and polymer^13–19 nanowires prepared between a surface and an STM or AFM tip in electrochemical environments.

Those experiments spawned two research directions that would define his career. The first was the development of an array of analytical methods that followed from advances in surface plasmon resonance (SPR) spectroscopy (1999–2010 and beyond)^{20,21,30–32,22–29} and the re-purposing of polymer nanojunctions for chemical analyses (in 2003).^14,33–35 The second research direction was transport studies of single molecules using the STM-break junction (STM-BJ) methodology invented in his group (initiated in the 2003 time frame).^36–42

The STM-BJ methodology for investigating transport through single molecules was applied to DNA almost immediately after its discovery, in 2004.^41,43–49 This period coincided with the publication of a number of studies from other research groups in which charge transport through DNA was probed.^50–57 Subsequently, this subfield would become a productive one for his research group.

Following contributions to SPR techniques earlier in his career, optical microscopy methods, often hyphenated with other methods, emerged as an emphasis of his program in recent years.^31,58–66 One success involved the application of “differential detection”, used to enhance the sensitivity of force measurement in an AFM system, to surface plasmon spectroscopy.^24,67–71 This invention led to his successful development of new and powerful functional imaging tools. NJ also devoted time and effort to the translation of his research to broad societal benefit, co-founding two companies with his former students and postdocs.

SCANNING TUNNELING MICROSCOPY AND ATOMIC FORCE MICROSCOPY OF MOLECULES IN ELECTROCHEMICAL ENVIRONMENTS

As a graduate student at ASU, NJ studied low-frequency modes in DNA and other condensed matter systems, completing his PhD in 1988.^72–77 He went on to initiate studies of disordered systems as a postdoc in the lab of the late Professor Herman Cummins at CUNY. Heeding the pleas of his thesis advisor, he returned to ASU in 1990. The ASU lab had started a new research program, utilizing STM to image surfaces and molecules under aqueous solutions.³ It was here that NJ began his successful merger of electrochemistry and nanoscience, exploiting the development of large, flat single crystal gold surfaces that were readily cleaned with a hydrogen flame to yield near-perfect single-crystal voltammety.⁷⁸ NJ used these surfaces in electrochemical STM to study surface reconstruction,⁷⁹ potential-induced phase transitions⁸⁰, controlled deposition of lead⁸¹ and halides,⁸² and, returning to the theme of his PhD, an exquisite study of the phases of DNA bases adsorbed on Au(111) under potential control.⁸³ He also helped to perfect AFM as an electrochemical tool.^84,85 As a graduate student, NJ had used Brillioun scattering to map the microelastic properties of bone, a material whose properties are dominated by micro- and nanoscale heterogeneity.⁸⁶ He returned to this problem with one of the first AFM studies to map these elastic heterogeneities on the nanoscale.⁷²

In 1992, NJ started his independent career at Florida International University (FIU). At that point, STM and AFM had been in use for several years as imaging tools to study surface morphology at the molecular and atomic scales, but efforts to understand the imaging of redox-active molecules in solution were still at an early stage, impeded by practical experimental problems. Operating an STM in an electrochemical environment with separate control of the surface potential versus a reference electrode presented practical problems because applied voltages in the STM could drive the reactions of redox species in the solution, producing spurious current that interfered with measurement of the tunneling current. Molecular Imaging Inc., cofounded by Stuart Lindsay, developed an “electrochemical STM” with an integral potentiostat that enabled the STM-sample potentials to be defined relative to the electrochemical potential of the sample, solving this problem. NJ, with others, learned how to insulate STM tips except at the apex of the tip to suppress reactions with electron donors and acceptors in solution.

Armed with these tools, NJ set about pioneering applications of STM and AFM to problems in electrochemistry. His 1996 STM investigation of chemisorbed porphyrins (Figure 2a) on graphite was seminal in terms of opening this field (Figure 2).⁸⁷ In an earlier 1995 paper, he had demonstrated that these porphyrins adsorbed flat, forming close packed and crystalline two-dimensional (2D) layers on a graphite surface that was readily imaged at molecular resolution.⁶ The 1996 paper documented the influence of the iron redox center and the electrochemical potential of the surface, which fixed the ratio of Fe(II)/Fe(III) within these monolayers on their apparent height, determined by the conductivity of individual molecules within these monolayers.⁸⁷ At a constant tip-sample bias of −100 mV, iron protoporphyrin (FePP) molecules “light up” within mixed monolayers containing PP, producing larger apparent heights (Figure 2d). When the electrochemical potential of the surface is varied across the range about the reversible potential for FePP, E°, (Figure 2e), the apparent height shows a maximum at E° = −0.44 V versus the standard calomel electrode (SCE), the potential at which Fe states are resonant with the tip-sample potential (Figure 2f).⁸⁷ These data provided early evidence for electrochemically controlled resonant tunneling in molecular systems,⁸⁷ with more detail added by subsequent papers.^{2,6,7,88–95} These papers also established that such precise control of the electrochemical potential of the surface was compatible with atomic and molecular resolution imaging.

Figure 2.

a) Iron protoporphyrin (FePP, left) and protoporphyrin (PP, right) molecules. b) Schematic diagram of a scanning tunneling microscope (STM) with control of the electrochemical potential of the tip and surface for investigations of single-molecular electron transfer. c) Energy level diagrams of Fe(III)-protoporphyrin IX and protoporphyrin IX and their relation to the Fermi levels of the substrate and tip. d) STM images of FePP/PP mixed monolayers on graphite containing FePP and PP at the ratio of 0:1 (A), 1:4 (B), 4:1 (C), and 1:0 (D). (E)–(H) are the corresponding cyclic voltammograms, e) STM image of FePP molecule embedded in an ordered matrix of PP molecules when the substrate was held at −0.15 (A), −0.30 (B), −0.42 (C), −0.55 (D), and −0.65 V (E), versus SCE ref. electrode. (f) Apparent height of FePP relative to PP as a function of the substrate potential via Pt-Ir tip and W tip. Figures a–d reprinted with permission from ref⁸⁷. Copyright 1996 American Physical Society.

In some of the earliest in-situ investigations of the electrochemical reactivity of molecules, AFM and STM were used in combination and provided complimentary electronic and topographical information.^2,7,95,96 For the electrochemical oxidation of guanine monolayers on graphite,⁹⁶ AFM produced lower noise images, documenting the formation of nanometer-scale voids in these monolayers during the oxidation process. The higher resolution of STM images provided an opportunity for changes in the ordering of individual molecules to be observed in some cases,^96,97 but STM images acquired for guanine and adenine monolayers on graphite showed potential-dependent superpositions of the graphite lattice with that of the monolayer.⁷ Because an organic monolayer on an electrode surface is located within the electrical double layer, the influence of an electric field operating on the monolayer must also be considered.⁹⁸ The thickness of the monolayer and the radius of the adsorbed ions can be as small as a few Ångstroms, thus, the electric field between these two charged layers—the electrode surface and oppositely ions located at the inner Helmholtz plane—can be large and controllable over a broad range (e.g., −10⁹ to 10⁹ V/m) using the applied electrode potential. These observations provide the basis for the conclusion that the local distribution of electron densities inside of a molecule can be modified and controlled by the electric field.^2,7 Because the local electron distribution of a molecule determines its reactivity, this discovery is significant in understanding or designing electric-field-induced redox reactions, and even chemical reactions.^99–101 For in situ investigations focusing on the reactions of molecular monolayers, AFM operating in the recently developed tapping mode was used effectively, in preference to STM, a pervasive trend at this time.^102–106

As the electrode potential changes, the surface charge density can be continuously changed, reaching values as high as 0.1 electron per surface atom. Another important contribution is the discovery of charge-induced order-disorder transition of an organic monolayer at a solid–liquid interface.⁹¹ Using 2,2’-bipyridine (22’BPY) as an example, he found that the molecules stack into polymeric chains that are randomly oriented on the surface at low charge densities (Figure 3a).⁹⁴ Increasing the charge density to a critical value, the polymeric chains are aligned along one of the Au(111) lattice directions in a fashion similar to the isotopic-nematic transitions of liquid crystals (Figure 3b,c).^91,94 The surface-charge-induced order–disorder phase transition is reversible and this phenomenon is general, existing in other organic molecules such as 4,4’-bipyridine,⁹⁴ 1,10’-phenanthroline,⁹³ and even anions¹⁰⁵ as well as large biomolecules such as lipids.¹⁰⁴

Figure 3.

Potential-induced disorder–order transition in a 2,2’-bipyridine monolayer on reconstructed Au(111). Images a, b, and c were obtained at sample potentials of 0.11 V, 0.24 V, and 0.38 V versus SCE, respectively, in aqueous 0.10 M NaClO₄. Lines in a and b indicate the direction of the reconstruction strips. Arrows in b point to the formation of the ordered phase as the potential is raised to 0.24 V. Reprinted from ref⁹⁴. Copyright 1996 American Chemical Society.

The surface-charge-induced reversible order–disorder phase transitions of lipid films, providing an explanation of the unusually high diffusion of myoglobin (Mb) in a solid-like lipid film.¹⁰⁴ Similarly, the surface charge induced structural changes of monolayers of phosphate ions, helping understand the unusual electrochemical potential dependence of the electron transfer behavior of cytochrome c.¹⁰⁵ Immobilizing metalloproteins with intact and/or optimized conformations on electrode surfaces is not only important for the study of biological processes, but also integral to the development biosensors and the investigation of biosynthetic processes. However, these molecules often denature upon adsorption onto a solid electrode. These studies were foundational for his development of practical sensors/biosensors, as described below.

QUANTUM TRANSPORT IN ONE-DIMENSIONAL METAL AND CONDUCTIVE POLYMER NANOSTRUCTURES

In the late 1990s, metal quantum point contacts (QPCs) became tools for the investigation of the conductivity of single molecules. These QPCs exhibited quantized conductance in units of G₀ (Eq. 1) were prepared using mechanically controlled break junctions^107–112 and by forming QPCs at a metal surface by using the tip of a scanning tunneling microscope (STM) to contact a metal surface.^12,113,114

$$G=n2e^2/h=n77.4\mathrm{\mu }\text{S},n=1,2,3\dots$$ [1]

Li and Tao demonstrated the first electrochemical formation of a copper metal QPC in an STM in 1998 (Figure 4).¹¹⁵ In these experiments, a partially insulated STM tip was positioned a few nanometers above a Au(111) surface in an aqueous copper plating solution and copper metal was electrodeposited with the aid of reference and counter electrodes also immersed in this solution (Figure 4a). In this configuration, the deposition of copper and its dissolution from the nanowire could be controlled using the potential applied to the tip (Figure 4b). This configuration provided a means by which the diameter of the copper nanowire could be precisely controlled in situ. Measurements of the tip–sample current versus time recorded during Cu electrodeposition and subsequent dissolution produced conductivity steps either up (during deposition) or down (during dissolution) in multiples of G₀ (Figure 4c).¹¹⁵ Conductance histograms for this system, compiled from hundreds of deposition, stripping, and re-deposition traces (Figure 4b) showed prominent peaks at G values predicted by Eq. (1). This 1998 experiment was the direct progenitor of the STM-break junction method¹¹⁶ that would be applied in 2003 to measurements of the single-molecule conductance.

Figure 4.

Quantum transport in metallic nanowires. a). Schematic drawing of the experimental setup. b,c) Plots of deposition/dissolution current versus the potential for an Au(111) substrate (b, left) and a segment of Pt_0.8Ir_0.2 wire (the tip material, b, right) are obtained by sweeping the potential linearly at a rate of 50 mV/s. Note that the actual deposition/dissolution current for the coated tip used in the experiment is too small to be measured, c) As Cu is deposited onto the tip, the growing front touches the substrate electrode, resulting in a stepwise increase in the conductance. The top curve is shifted upward by 5 G₀ for clarity. The tip and the substrate potentials were held at 30 and 25 mV, respectively. d) Conductance histogram for Cu nanowires fabricated by electrochemical deposition and dissolution at room temperature. Reprinted with permission from ref¹¹⁵. Copyright 1998 American Institute of Physics.

Analogous experiments with poly(aniline) (PANI) nanowires were less readily interpreted.^15,17,18 Initially, a lithographically fabricated gold nanogap was used, but an STM tip–sample gap was also studied because it facilitated investigations of the influence on conductance of stretching of the PANI nanowire.^15,17,18 Whereas metal QPCs in the STM showed discrete conductance steps spaced by G₀ as the tip–sample distance was increased by a few Ångströms,^12,113,114 decrementing the conductance of PANI nanojunctions required increasing the tip–sample spacing by many nanometers.¹⁹ Stepped conductance traces acquired while stretching PANI junctions showed steps corresponding to 10⁻⁷–10⁻⁸ S, orders of magnitude smaller than G₀. These conductance steps were not attributed to conductance quantization.¹⁹ Instead, it was hypothesized that they were caused by transitions between polymer chain configurations for the multiple polymer chains present within the tip–sample gap.¹⁹ Conduction through PANI nanowires also varied as a function of its oxidation state. The conductivity of the PANI bridge “telegraphically” alternated between zero and one level for cases where a small amount of PANI was deposited, and between several current levels when larger amounts were deposited.¹¹⁷ The number of accessible conductivity states could be adjusted by equilibrating the PANI nanowire at various potentials. This discrete conduction behavior was attributed to fluctuations of the polymer between doped, conducting, and undoped insulating redox states.¹¹⁷ This initial series of experiments on PANI occurred in the 1998–2003 time period, immediately before Tao’s group re-focused on understanding the conductivity of smaller organic molecules.

Years later, Tao and colleagues revisited investigations of PANI nanowires in the STM.^14,115,118 The application of tensile stress to these junctions was observed to produce two behaviors: Initial stretching caused an increase in conductivity that was attributed to alignment of polymer chains; with continued stretching, a maximum conductivity was observed, followed by a staircase of declining conductance steps resembling those seen in the copper nanowire system.¹¹⁵ In 2007, Lindsay and Tao conducted a comprehensive investigation of electrodeposited PANI nanowires in planar, nanofabricated (static) metal gaps immersed in aqueous solution.¹⁴ In these experiments, negative differential resistance, a common feature of single-molecule conduction including studies of aniline oligomers,¹¹⁸ was observed for PANI systems where a relatively small amount of the polymer had been electrodeposited.¹⁴

CHEMICAL AND BIOLOGICAL ANALYSIS

Among NJ’s first forays into chemical sensing was his development in 1999 of high-resolution SPR.^21,119 This advance was achieved by replacing the commonly used single photodiode with a quadrant photodiode chip, which permitted real-time subtraction of “background” and “active” surface regions and common mode noise reduction caused by light source fluctuations, mechanical vibrations, and thermal drift.^21,119 This high-resolution SPR modality increased the refractive index sensitivity by orders of magnitude, affording a resolution of 10⁻⁸ refractive index units (RIU) and a response time of 1 μs.^21,119 This level of refractive index sensitivity enabled new and heretofore impossible applications of the SPR. One of these applications involved the detection of adsorbed layers of toxic elements, such as lead¹¹⁹ and arsenic.¹²⁰ In these examples, reference and sample areas of the sensor were coated with an unfunctionalized n-alkanethiol self-assembled monolayer (SAM), and a carboxylate (-COOH)-terminated SAM, respectively. Metal cations in a contacting solution phase were bound by the carboxylate-terminated thiol SAM, but not at the unsubstituted SAM.^119,120 The limit of detection for this enhanced SPR technique for Pb²⁺ (40 ppt)¹¹⁹ and As (as H_xAsO₄^(3-x)-) was 10 ppb,¹²⁰ as required by the United States Environmental Protection Agency. This technique would form the basis in 2004 of the startup company Biosensing Instruments Inc. (vide infra).

At the same time, the Tao group began to investigate applications of conductive polymer nanojunctions to problems in chemical analysis. This new research direction was inspired both by his previous work on STM/conducting polymer conductivity measurements,^{14,15,17,18,114,117} and by observations that the quantum conductance of these systems was influenced by molecular properties as well as the electronic structure of the solid state.^16–18 This research thrust was launched by a 2003 paper titled, “A Conducting Polymer Nanojunction Sensor for Glucose Detection,”¹²¹ in which a PANI/poly(acrylic acid) channel bridging two lithographically fabricated metal nanoelectrodes was doped with the enzyme glucose oxidase (GO_x). The detection of glucose by this sensor occurs when GO_x produces H₂O₂ in the presence of glucose, oxidizing the PANI and enhancing its electrical conductivity.¹²¹ The nanoscale dimensionality of the sensor made possible a rapid (200 ms) response time for glucose detection.¹²¹ This paper showed that nanojunctions, previously the domain of physicists, provided analytical chemists with a new and versatile tool for designing chemical sensors.

An unexpected advantage of such nanojunction nanosensors is their ability to operate in unprocessed, real samples. Inspired by Justin Gooding’s work on peptides,¹²² a conducting polymer nanojunction sensor for the detection of heavy metals in drinking water was developed,¹²³ followed by a conducting polymer nanojunction breath sensor, that detected ammonia at parts-per-billion levels.¹²⁴ Elsewhere, several groups reported on the amazing sensitivity and other overall properties of nanosensors.^125–127 Taking inspiration from this work, in 2004, NJ and Larry Nagahara of Motorola Labs decided to explore the properties of low-dimensional carbon materials in sensors. The capability of Motorola to create silicon-based chips with 1 or 2 single-wall carbon nanotubes (CNTs) in a micron-scale gap enabled modification of the CNTs, leading to the demonstration of ultra-low parts-per-trillion level sensing of heavy metal ions¹²⁸ and pM level detection of viral RNA strands for Hepatitis C.¹²⁹ The sensitivity of the latter would eliminate the need for amplification by polymerase chain reaction, PCR, of RNA strands, a requirement of many current methods used for virus detection.

The versatility of PANI nanojunction sensors was demonstrated in 2007 with the operation of these sensors both in conductometric and amperometric modes (Figure 5a,b).³⁴ Conductometric detection is obtained by transducing the channel conductance, just as in chemical field-effect transistors (FETs). Amperometric detection exploits the direct oxidation/reduction of a target molecule at the PANI channel (Figure 5c). Nanojunction sensors can be operated in either of these modes, or in both simultaneously (Figure 5d). The attributes of these two modes were compared for the detection of the neurotransmitter dopamine.³⁴ The work came to fruition with the selective detection of dopamine in the presence of 1000-fold excesses of an interferent, ascorbic acid.^34,130,131 Versions of this technology were commercialized for the detection of exhaled nitric oxide,^132–134 an airway inflammation biomarker, and for the ultrasensitive detection of explosives.^135,136

Figure 5.

(a) Schematic illustration of a hybrid amperometric and conductometric sensor. WE1 and WE2 represent two working electrodes (source and drain electrodes) connected with a conducting polymer bridge. RE and CE are reference and counter electrodes, respectively. The electrochemical gate potential (V_g) E₁ is applied between the drain electrode (WE₁) and RE, and a bias voltage (V_bias) is applied between WE₁ and WE₂. (b) Optical and scanning electron microscope images of the device made of an array of polymer bridges on a silicon chip. (c) Cyclic voltammogram of 3.9 mM hydroquinone (HQ) on polyaniline-modified microelectrode at 60 mV/s. (d) ΔI_d and ΔI_2,ec simultaneously recorded during the injections of supporting electrolyte (1–2) and 1.4 mM HQ (3) to a close (bridge) and open polymer bridge (open) amperometric and conductometric device. V_bias) 0.25 V and E₂) 0.45 V. Inset of (d) illustrates the reaction mechanism in a close polymer bridge device. Reprinted from ref³⁴. Copyright 2007 American Chemical Society.

In 2000–2004, NJ’s team was also undertaking chemical analyses of gases using inexpensive quartz crystal resonator tuning forks, typically used in electronics and watches for timing. Tuning forks could be repurposed as sensitive mass measurement transducers for chemical sensing.^137–141 Working with Motorola scientists Larry Nagahara and Ray Tsui, NJ’s team created an alcohol sensor embedded in a cell phone. In this device, a polymer nano/microwire composite sensing material bridged between the prongs of the tuning fork. This system later evolved to become a wearable mobile sensor for measuring the mechanical properties of light-sensitive materials¹⁴² and for the spatial-temporal tracking of exposure to pollutants^{134,143–146} and particles.¹⁴⁷ The latter systems were introduced to the public by Dr. Francis Collins, Director of the National Institutes of Health, in the 2015 (Mobile) mHealth Summit in Washington, D. C.

In 2009–2010, aware of the fabrication limitations of silicon-based sensors and the relatively high cost of maintenance of clean room operations, NJ sought a relatively inexpensive way to create sensors. Inspired by the work of Ken Suslick on colorimetric sensor arrays,¹⁴⁸ NJ and Erica Forzani began the search for sensing applications with high market demand and created a sensor array for selective detection of oxygen and carbon dioxide in breath, enabling them to measure energy expenditure (kcal/day) via indirect calorimetry.^149–151 This innovation later led to several practical applications to improve the efficacy of weight and disease management in overweight and diabetic patients¹⁵² as well as exercise practices,¹⁵³ and to spin off the second of NJ’s start-up companies, TF Health Co. (vide infra).

In 2008, NJ was named Director of the Center of Bioelectronics and Biosensors at The Biodesign Institute, Arizona State University, which has ~30 permanent researchers and over 20 institutional collaborators, such as Banner Neurological Institute and The Mayo Clinic. Under his leadership, the Center’s scientific contributions expanded in new directions including the fields of chemical and physical sensors for noninvasive environmental sensing^154–158 and medical applications.^159,160

ELECTRICAL CONDUCTANCE OF SINGLE MOLECULES

A long-standing objective in the subfield of molecular electronics has been to develop molecular devices that will complement—and eventually even supersede—current semiconductor-based technologies.^161–163 NJ Tao’s flagship research in molecular electronics, especially in molecular transport junctions (MTJs), established his as a leading group in this area. In 2003, Tao and one of his students invented the scanning tunneling microscope break junction (STM-BJ) method, possibly his most impactful contribution to molecular electronics.¹¹⁶ The STM-BJ method and its variants have been adopted as a standard measurement protocols for the investigation of electron transport in single molecules.

Although a number of methods had been developed to measure the charge transport through single molecules by 2003, the main problems with measurement were creating the electrode–molecule–electrode junction in a reproducible way and handling the inevitable variability in the properties of single molecules. The STM-BJ method starts with repeatedly driving a conducting STM tip into contact with and then pulling away from an Au (111) surface coated with molecules containing anchoring groups that can bond to Au; the conductance curves of the whole processes are simultaneously recorded. As shown in Figure 6a–A, at the start of the STM tip pulling, Au quantum contacts were formed, as evidenced by the conductance steps at multiple integers of the conductance quantum (Eq. 1). Remarkably, after the breaking of the of the quantum contacts, molecules were contacted with the tip and the substrate to form an Au-molecule(4,4’-bipyridine)–Au junction, as indicated by the conductance steps being 100 times smaller than the conductance quantum (Figure 6a–C). No conductance steps were observed in the molecule conductance window when no molecules were applied. In this method, the reliability issue was successfully resolved by forming and measuring the junctions thousands of times. The conductance histogram (Figure 6a–B,D,F) shows statistically significant results that capture the inherent variability of single-molecule measurements.

Figure 6.

The scanning tunneling microscopy-break junction (STM-BJ) method. (a) STM-BJ using 4’4’-bipyridine molecules. Reprinted with permission from ref 293. Copyright 2003 American Association for the Advancement of Science. (b) Conductive atomic force microscopy measurement of Au-thiol binding in Au–octanedithiol–Au molecule junction. Reprinted from ref 38. Copyright 2003 American Chemical Society. (c) AC-modulation of molecular junctions. Reprinted from ref 12. Copyright 2011 American Chemical Society.

As soon as single molecule STM-BJ conductance data were available, it was apparent that Eq. [1] was insufficient to describe these data. A molecular junction is a hybrid system that exhibits a conductance, G, determined by both the molecule and the molecule–electrode contacts:

$$G=\hspace{0.17em}\frac{n2e^2}{h}\frac{{\Gamma }^2}{{\Delta }^2+{\Gamma }^2}\hspace{1em}n=1,2,3\dots$$ [2]

where Γ and Δ= E – ε are the coupling between molecule and electrode, and the energy difference between the conduction orbital and the Fermi level of the metal electrodes respectively.¹⁶² To account for differences in the electrode–molecule contact and to confirm the formation of Au–amine and Au–thiol bonds, a follow-up study was carried out that simultaneously measured the conductance and breakdown forces of the molecule junctions.³⁹ The breakdown force of the Au–thiol contact was found to be ca. 1.5 nN (Figure 6b) whereas that of the Au–amine contact was ca. 0.7 nN, in agreement with theoretical calculations.

Soon after the discovery of STM-BJ, two seminal publications based on this technique were published from other research groups. The first, by Reddy et al., reported the measurement of single-molecule Seebeck coefficients S_junction for 1,4-benzenedithiol (BDT), 4,4′-dibenzenedithiol, and 4,4′-tribenzenedithiol, revealing a molecular length dependence on the value of S_junction.¹⁶⁴ The second, by Venkataraman et al., reported an investigation of seven biphenyl molecules with varying ring substitutions that imparted a twist angle to these molecules.¹⁶⁵ The single molecule conductance of these molecules decreased as the twist angle increased,¹⁶⁵ quantitively in accord with theory.¹⁶⁶ These two investigations enabled by STM-BJ are cornerstones in the development of the field of molecular electronics.^167,168 Subsequent development of STM-BJ by the Tao group enabled the capture of fast events (~1 ns) by using high-speed electronics (Figure 6c),¹² observations of the effects of anchoring groups,¹⁶⁹ the introduction of target-molecule-specific binding sites, assessments of the quantitative influence of molecular geometry,¹⁷⁰ and the elucidation of the origin of some junction instabilities,⁴⁰ making STM-BJ a comprehensive experimental platform in measuring various single-molecule transport properties.

A review of NJ’s expansive contributions to molecular electronics is beyond the scope of this Perspective, but several especially impactful contributions are described. One of these is the gating of electron transport in molecular junctions using an electrochemical gate, which contributed to an understanding of electron transport mechanisms in molecular junctions and enabled device applications. In fact, a number of gating modalities were explored (Figure 7).

Figure 7.

Schematic illustration of controlling electron transfer in molecular junctions by mechanical, electrochemical, thermal, and magnetic gating.

One such gating modality involved mechanical control of the STM tip–surface spacing. The influence of tip position on the apparent conductance was noticed by Haiss et al. who reported that the conductance of a molecule located in the gap between an STM tip and the surface oscillated between two discrete values as the molecule spontaneously connected and disconnected from the STM circuit.^171,172 NJ termed this behavior “blinking.”¹⁷³ Blinking did not occur when the tip–surface distance exceeded the length of the molecule under examination, and this parameter could also be used to control the angle of a molecule within the gap, thereby affecting the conductivity.^171,172 With ac modulation on the tip–surface distance, NJ was able to distinguish in real time between an open gap, which produced large current modulations, and a molecular junction, which showed damped current modulations. Simultaneous measurement of the conductance of the gap was facilitated as well.¹⁷³ A “blink and pull” method¹⁷⁴ took this experiment one step further, by probing the ac conductance while the tip–sample distance was increased, causing the tilt of a single molecule to be reduced. This technique enabled precise measurement of conductance as a function of molecular tilt,¹⁷⁴ exceeding what had been possible in previous studies. These papers established that precise mechanical control of single-molecule conductance was achievable and understandable at a fundamental level.^173,174

Although rectification should not be termed “gating”, it is a phenomenon of interest in molecular electronics. Single-molecule rectifiers based upon asymmetric organic molecules were predicted in the 1974 paper by Aviram and Ratner.¹⁶³ The concept was experimentally demonstrated for asymmetric single molecules in self-assembled monolayers by Luping Yu and coworkers.¹⁷⁵ Isolated single molecular diodes were achieved in 2005 by Elbing et al.¹⁷⁶ using a mechanically controllable break junction. However, switching of the forward-biased direction for the diode, dictated by the molecular orientation in the junction, was not possible in this case. Using an asymmetric diblock dipyrimidinyldiphenyl molecule, NJ demonstrated this capability in 2009 using the STM-BJ blinking methodology, which enabled an immobilized molecule to be expelled from the junction and a new molecule (or the same one) reattached with reversal of its polarity.¹⁷⁷

Mechanical gating was also achieved by the application of tensile and compressive stressed to single molecules. This discovery in 2012 involved the modulation of the conductance of 1,4’-benzenedithiol (BDT) molecules by applying either a tensile stress or a compressive stress to the molecule using the STM tip.¹⁷⁸ The conductance of BDT was perturbed by up to an order of magnitude by these forces. Counter to intuition, in these experiments the BDT conductance was enhanced during stretching and depressed during compression. The HOMO of the BDT migrates toward the Fermi level of the Au electrodes during stretching, closer to resonance with the Fermi energy of the Au contacts.¹⁷⁸

Related to transport gating in molecular junctions is the issue of heating.^179,180 NJ and his students studied this issue for 1,8’-octadecanedithiol using the STM-BJ technique.^181,182 These experiments afforded an estimate of the temperature increase induced by current flow of ~30 K for V_bias = 1.0 V, the highest bias for which these junctions were stable.¹⁸² Subsequently, more detailed STM-BJ studies of C6, C8, and C10 alkanedithiols revealed that heating of these molecules increased with decreasing molecular length; these junctions showed complex bias dependences in which maximum heating of these molecular junctions was seen at V_bias ≈ 0.80 V, because at higher V_bias electron–electron cooling occurs in accord with the recently proposed hydrodynamic theory proposed by D’Agosta and Di Ventra.¹⁸³

Electrochemical gating of molecular junctions was demonstrated by NJ and his group in 2005, in a flurry of papers describing an electrochemically gated single-molecule FET,¹⁸⁴ electromechanically controlled conductance switching,¹⁸⁵ and electromechanically controlled negative differential resistance-like (NDR-like) behavior.¹⁸⁶ But there would be many more publications from NJ and his students exploring this phenomenon, reporting redox-gated electron transport in electrically wired ferrocene molecules,¹⁸⁷ and electrochemical control of electron transport in graphene¹⁸⁸ and in coronenes.¹⁸⁹ Furthermore, electrochemically gated single-molecular switches¹⁹⁰ and ambipolar single-molecule FETs¹⁹¹ were successfully demonstrated, pushing single-molecular device applications a step forward. Using temperature control as the thermal gating, Tao and his group also investigated electron transport in single redox molecules,¹⁹² the electron–phonon interactions in single-molecular junctions,^193,194 tunneling-to-hopping transitions in single-molecular junctions,¹⁹⁵ low-bias rectification in a single-molecule diode,¹⁹⁶ thermopower and transition voltages,¹⁹⁷ and the nonexponential length dependence of conductance.¹⁹⁸ Lastly, the Tao group measured the electronic spin filtering capability of a single chiral helical peptide by using a ferromagnetic electrode source to inject spin‐polarized electrons.¹⁹⁹

What might NJ have done next in this area? Based upon the collective experience of the authors, most of whom worked with him over many years, one goal that he articulated was to extend the measurement platform to explore and to control chemical reactions at the single-molecule level, including the development of the ability to determine subtle molecular structural information, such as the determination of chirality for two stereoisomers. A second aspiration was to integrate optical imaging techniques with STM-BJ to explore the coupling between electronic transport and light in single-molecule junctions, work that would have impacted both molecular electronics and molecular plasmonics.

CHARGE TRANSPORT IN SINGLE DNA MOLECULES

Beyond its role as the carrier of genetic information, DNA has become an important material in nanoscience and nanotechnology.²⁰⁰ The properties of DNA, including self-assembly, programmable structural control, rich chemistry provided by the polymer backbone with varying side groups, and controllable helical structure, have led to a wide variety of innovations utilizing DNA as a structural, mechanical, lithographic, or biotechnology element.^201–205 However, the utilization of DNA as a circuit element has proven difficult due to its complex electronic structure and the strong interplay between the electronic and structural degrees of freedom. Much of our understanding about the utility of DNA as an electronic element has come from Tao’s work, and the work he inspired.

After the development of the STM-BJ methodology in 2003,²⁰⁶ NJ’s attention returned to a topic that had interested him since his Ph.D. studies: the properties and dynamics of DNA.^76,77,207 This time, however, he placed the emphasis on the electronic properties of DNA. This was an opportune time for such studies as there was significant conflict in the field about the inherent charge transport properties of DNA. Reports emerged that ran the gamut of possible electronic capabilities for DNA including insulating,^208,209 semiconducting,²¹⁰ conducting,²¹¹ and proximity-induced superconductivity.²¹² The incongruity in conductance studies was mirrored by the significant disparity in charge-transfer properties seen in photochemical and electrochemical measurements.^213–216

Tao’s background in electrochemical STM²¹⁷ and the advent of the new break junction system²⁰⁶ gave a unique opportunity to begin addressing these issues: The DNA duplexes could be measured in aqueous environments that could help preserve the inherent DNA structure, which may be lost if measuring the DNA in ambient conditions or vacuum as had been done in some other experiments. Initial experiments on short DNA duplexes (< 18 bp) showed that G:C repeat sequences yielded a conductance proportional to the length of the sequence, indicating hopping, whereas A:T pairs inserted into the stack yielded a weak exponential, suggesting that the charges tunneled through the A:T barriers.²¹⁸ This work placed conductance measurements for short sequences in line with photochemical measurements with similar outcomes.^{50,214,219,220} Interestingly, however, the temperature dependence could not be determined within the temperature window of the experiment (between the liquid freezing temperature and the DNA melting temperature), which left the opportunity open for more complex transport behavior.^45,221

Over the next 15 years, a variety of work from Tao and many others placed intensive focus on understanding the native electronic behavior of DNA duplexes,^222–229 RNA:DNA hybrids,²³⁰ and G-quadruplexes.^231–233 NJ’s group focused on understanding the thermal, mechanical, and thermoelectric properties of these systems, and how these properties relate to the electronic structure.^234–237 The thermoelectric measurements demonstrated conclusively that DNA transport was hole dominated: hopping-dominated in G:C rich systems, and tunneling dominated through short A:T bridges, but it again transitioned to hopping if the A:T bridge became long enough.²³⁷ The addition of the Seebeck coefficient to the conductance measurements for these systems provided a second methodology to verify previous results from conductance and the resulting implications about the transport. These measurements heralded new opportunities for quantifying contact properties and understanding alignment of energy levels between bases in the stack.

Tao’s group made important advances in understanding the contact and coupling properties within DNA and their influence on conductance by utilizing the pulling capabilities of the STM-BJ system on duplexes of various designs.^234–236 These results demonstrated that when pulling on the molecule, conductance breaks down once the hydrogen bonds in the first one or two bases break, which means the forces must be carefully controlled in DNA to observe conductance behavior. This result helps rationalize the wide variety of initial conductance values observed in DNA-based systems. An additional mechanical approach was to apply a small AC modulation to the tip (~1–2 Å) at a higher frequency (2 kHz).^173,234,238 This approach enables more direct information to be determined about the coupling between bases, especially purines, in the base-stack. This advance led to understanding of the effects of coupling between bases in the stack when placed in interstrand or intrastrand configurations, which has helped lead to designs for improving the conductance of DNA-based structures.

The experiments described above indicate that careful design of DNA sequence and structure could lead to unique effects, but they also indicate that obtaining significant changes in the overall conductance or modulating the transport properties of DNA will require additional changes to its chemical and electronic properties. In 2015, Tao’s group demonstrated that in G:C rich hopping systems, the coherence length of the charge carriers (holes) could extend beyond a single guanine base.²³⁹ This finding was followed with another study in collaboration with Beratan’s group that began to provide design rules for engineering the coherence and coupling between bases.²⁴⁰ These studies were quickly followed by other groups demonstrating longer coherence lengths in DNA, RNA, and peptide nucleic acid (PNA) systems than initially expected from length-dependent conductance measurements,^{44,230,241–243} which may in the long term provide a basis for the rational design of DNA-based devices.

Developing devices is one of the ultimate goals for DNA transport studies. For native structures, the devices will likely be sensor systems, and Tao’s group conducted initial studies to show the impact of single-base mismatches²⁴⁴ and epigenetic effects²⁴⁵ on conductance, which have been expanded on by other groups.^246–248 Beyond these native structures and sequences, focus has been given to other device designs as well, such as the creation of a G-quadraplex-based current-splitting element,²³² which may give way to 3-terminal devices or advanced wiring concepts for electronic devices (Figure 8). Inclusion of non-native bases,²⁴⁹ intercalators,²⁵⁰ and metal ions²⁵¹ could be used to create DNA-based wires,²⁵² transistors,²⁴⁹ diodes,²⁵³ or photoactive elements.

Figure 8.

Design and testing of a G-quadraplex-based current splitter. (A) Schematic of the current-splitting element placed between two electrodes. The G4 element (black) plus 3 DNA duplexes are held between an Au tip and an Au substrate in a break-junction configuration. (B) Conductance histograms of the quadruplex 3 with six layers of guanines. The different conductance histograms represent whether the strands are arranged in an antiparallel or parallel configuration and which of the three duplex terminals (T1, T2, T3) have thiols to bind to the electrodes. The conductance values (x-axis) are similar for all the DNA structures. For clarity, each histogram was shifted vertically to avoid overlap. Reprinted with permission from ref²³². Copyright 2018 Springer Nature.

Single-molecule transport experiments have dramatically advanced our understanding of charge transport in DNA. These data and the deeper understanding enabled by it have powered the nascent field of DNA-based device design, spawning proof-of-principle systems for sensing and electronic circuit elements, including current-splitting and potential 3-terminal elements,²³² diodes,²⁵⁴ and systems that support long-range transport over tens of nanometers.²³¹ Moving forward, the initial experiments on the length coherence of carriers in systems may lead to new designs that harness the quantum properties of DNA’s electronic structure. These innovations could result in both new sensing paradigms where changes in sequence result in modifications of coherence and new devices that engineer interference effects or modifications to the electronic structure for device applications.

Much is yet to be learned, including developing methods to engineer and to control the coherence length in these materials, developing new avenues for integrating these devices into larger scale structures and the outside world, fundamental knowledge about how to align and to manipulate the energy levels in DNA in a robust and reliable way, and understanding how fluctuations and the inherent flexibility in these systems contribute to the electronic properties and how these properties can be utilized for improvement. The knowledge gained from Tao’s experimental insights over the years built an important foundation for exploring these areas, impacting not only the field of DNA-based electronics but research across nanosystems, where emerging technologies such as DNA-based memories and computing,²⁵⁵ origami-enabled electronic design,²⁵⁶ and long-range transport in protein-based systems²⁵⁷ continue to gain momentum.

OPTICAL MICROSCOPY AND DETECTION TECHNOLOGIES

Optical microscopy is one of the most common and powerful tools used in materials and life sciences research labs. Advances in microscopic imaging technologies have played critical roles in biomedical research and clinical applications.^258–260 A large portion of NJ Tao’s research career was devoted to the development of new functional imaging tools for chemical and biological applications. His group invented two groundbreaking microscopy technologies for functional imaging of chemical and biological targets.

The first imaging technology is a multifunctional plasmonic-based electrochemical impedance microscopy (P-EIM),^23,261,262 integrating surface plasmon resonance microscopy (SPRM)^28,263,264 and electrochemical impedance microscopy (EIM)²⁵ into a conventional inverted microscope (Figure 9). Unlike conventional electrochemical impedance spectroscopy,²⁶⁵ which measures alternate current amplitudes, P-EIM exploits the more sensitive optical detection of surface plasmonic signals for metals on the surface charge density,²⁵ thus enabling fast (sub-ms) and noninvasive imaging of impedance at near diffraction-limited resolution without an electrode array or a scanning probe. In addition, P-EIM maps the local electrical polarizability and conductivity, which reflects local changes in the cellular structure and ionic distribution. Compared to conventional optical imaging techniques, the plasmonics-based approach can image not only the morphology, but also molecular binding events and local chemical information taking place at fast time scales. Important benefits of this approach include fast, noninvasive, and quantitative imaging of optical mass change, electrochemical current, and impedance of various samples, from small molecules to whole cells.

Figure 9.

Principles and applications of plasmonic-based electrochemical impedance microscopy (P-EIM) microscopy. (A) Schematic of P-EIM imaging principle: Surface plasmon waves are created optically with a laser or light-emitting diode beam incident on the metal film via a prism or high-numerical-aperture objective at a resonant angle, and the reflected beam creates a surface plasmon resonance microscopy (SPRM) image. The resonant angle is sensitive to changes in refractive index near the metal surface, which is widely used to study molecular binding processes. In addition, the resonance angle is also sensitive to surface charge density, which produce the electrochemical impedance image when an AC voltage is applied to the surface. (B) optical, SPRM and P-EIM images of a protein microarray, viral nanoparticles (NPs), and a mammalian cell.

A variety of applications for the P-EIM technology have been demonstrated, including trace chemical analyses (Figure 10);²³ catalytic reactions^266–270 and thermal diffusion²⁷¹ of nanomaterials; charge distributions in 2D materials;⁶³ detection of single DNA,⁶⁹ exosomes,²⁷² or single virus (Figure 11);²⁸ tracking single organelle transportation in cells;²⁷³ time-resolved digital immunoassay;²⁷⁴ nano-oscillators for detection of charge,²⁷⁵ molecules,²⁷⁶ and biochemical reactions;²⁷⁷ ligand–membrane protein interactions in intact cells;^{261,278–283} tracking bacterial metabolic changes for antimicrobial susceptibility tests;²⁸⁴ and mapping action potentials in single neurons (Figure 12).²⁸⁵ With further improvement of this technology, particularly in signal-to-noise ratio, P-EIM could lead to label-free functional detection of single proteins. This purpose was a goal of NJ’s since 2000, when he calculated that, with sufficient light, SPR was capable of detecting single protein molecules (as noted during discussions with Dr. Shaopeng Wang, then his postdoc at FIU).

Figure 10.

Detection of TNT traces on a fingerprint using the electrochemical current imaging technique. (A) Surface plasmon resonance image of a fingerprint. (B1–B5) Five snapshots recorded while sweeping the potential negatively from 0 to −1.0 V at a rate of 0.05 V/s. The appearance of the “spots” is due to the reduction of TNT particulates. (C) Local voltammograms of the regions with (blue and red dots) and without (black dots) TNT particulates. The electrolyte is 0.5 M KCl. Reprinted with permission from ref²³. Copyright 2010 American Association for the Advancement of Science.

Figure 11.

(A) Surface plasmon resonance microscopy (SPRM) images of H1N1 influenza A virus and three different sized silica nanoparticles in phosphate-buffered saline (PBS) buffer. (Insets) Nanoparticle images generated by numerical simulation. (B and C) The SPR intensity profiles of selected particles along X and Y directions (indicated by dashed lines in A, respectively. (Insets) Corresponding profiles from simulated images. Reprinted with permission from ref²⁸. Copyright 2010 Wang et al.

Figure 12.

Plasmonic imaging of action potential in single neurons. a) Bright-field and b) plasmonic images of a hippocampal neuron, where the dashed lines mark the edge of a neuron. c) Patch clamp recording of action potential and d) simultaneous plasmonic-based electrochemical impedance microscopy (P-EIM) recording of action potential of the whole cell (frame rate of 10000 fps). To clearly illustrate the data, we normalized the intensity change DI with mean intensity I, and plotted the plasmonic intensity in negative intensity change for fair comparison. e–h) Snapshot P-EIM images of action potential at the moments marked by the gray spots in (d). The P-EIM images were averaged over 90 cycles of repeated action potential firing at 23 Hz repeat rate to reduce random noise. Reprinted with permission from ref²⁸⁵.

The second imaging technology developed in the Tao group is mechanically amplified detection of molecular interactions (MADMI).^286,287 Measuring molecular binding to membrane proteins is critical for understanding cellular functions, validating biomarkers, and screening drugs.^288,289 Despite its importance, measuring small molecule binding to membrane proteins in their native cellular environment remains a challenge.^276,278,279 However, MADMI can measure the mechanical deformation in cell membranes that accompanies the binding of molecules with the membrane proteins on the cell surface. The laws of thermodynamics predict that when molecules bind to a surface, the surface tension changes. According to thermodynamics, the surface concentration (Γ) of molecules bound on the membrane surface is given by:²⁸⁶

$$\mathrm{\Gamma }=-\frac{d\gamma }{RTd\left(\text{ln\hspace{0.17em}}c\right)},$$ [3]

which applies to ideal solutions at a constant temperature and pressure, where γ is the surface tension, R is the gas constant, T is temperature, and c is the bulk analyte concentration. RTd(ln c) is the chemical potential of the molecules. From Eq. (3), molecular binding is proportional to the surface tension change. Because cells are soft, a small change in the surface tension can induce a substantial mechanical deformation of the cell, leading to signal amplification. Therefore, the interactions of molecules with membrane proteins on a cell can be measured by determining the mechanical deformation of the cell membrane. Because the signal does not scale with the size of analyte molecules, MADMI is particularly suitable for quantifying the binding kinetics between small molecule drugs and membrane protein targets in their native environment. Quantification of binding kinetics between protein, peptide, and small-molecule ligands on different classes of membrane proteins on adherent²⁸⁶ and floating cells²⁸⁷ has been demonstrated. Further development of MADMI for improved sensitivity, higher multiplexing capabilities, automated deformation-tracking algorithms, and testing in wider ranges of cells and ligands will lead to label-free biomarkers and drug-discovery tools for cell-based membrane protein binding kinetic studies.

To detect minute, nanometer-scale membrane deformations accurately, Tao’s lab developed a differential detection algorithm for tracking the edge movement of a cell using optical imaging (Figure 13). The method nulls common-mode noise in the optical system, thus achieving superior detection limits. This image-based differential tracking method not only works for detection of molecular interactions, but has also been adapted for broader applications, including tracking fast cellular membrane dynamics;²⁹⁰ tracking nanometer-scale cellular membrane deformation associated with single vesicle release;²⁹¹ determining electrochemical surface stress of single nanowires;²⁹² performing charge-sensitive optical detection of molecular interactions;^62,64 and measuring subnanometer mechanical motion accompanying action potential in neurons, where the detection limit was further improved via signal modulation and digital filtering.²⁹³ This differential tracking method is extremely general and is expected to have a range of applications in phenomena for which nanoscale mechanical deformations are important.

Figure 13:

Tracking sub-nm membrane displacement accompanying action potential in single mammalian neurons. a) Experimental setup showing a hippocampal neuron cultured on a glass slide mounted on an inverted optical microscope, which is studied simultaneously with the patch clamp configuration for electrical recording of the action potential, and optical imaging of membrane displacement associated with the action potential. b) Imaging and quantification of membrane displacement with a differential detection algorithm that tracks the edge movement of a neuron. c) Bright-field image of a HEK293T cell and a micropipette used to change the membrane potential. d) Difference image of the entire cell showing cell edge displacement and local variability (intensities for the orange and purple bands) of the displacement. e) Membrane displacement map obtained from the differential detection algorithm. Scale bar in (c, d, e): 10 μm. Reprinted from ref²⁹³. Copyright 2018 American Chemical Society.

TRANSLATING SCIENCE TO SOCIETAL BENEFIT BY COMMERCIALIZATION

NJ was a successful entrepreneur as well as a scientist. Upon his return to ASU from Florida International University in 2004, he founded Biosensing Instrument Inc. (BII) with Dr. Tianwei Jing and Professor Feimeng Zhou of California State University, Los Angeles. From its inception, NJ was actively engaged in BII, spearheading product design, identifying applications, and even providing customer support. After just one year of development effort, BII sold its first unit to the University of Texas at Austin for gas-phase applications. BII grew quickly and diversified its product line to include fully automated, multichannel SPR instruments and microscope-based variants, along with various analysis modules for diverse fields, including surface chemistry, single-cell analysis, and studies of cancer and neurological disorders, to integration with other analytical techniques for the detection of species of biological, pharmaceutical, and environmental importance. The companies’ record of innovation and commercialization has been recognized by two recent R&D 100 Awards.

As NJ’s research group attained momentum in chemical and biological sensing in the 2003–2008 time frame, the resulting technologies presented further opportunities for commercialization. In 2008, NJ and Erica Forzani co-founded TF Sensors LCC, which soon became TF Health Corp. with the aim of developing sensors for health monitoring. The addition of two scientists to the TF Health team, Drs. Francis Tsow and Xiaojun Xian enabled a series of product developments starting in 2011 with indoor air quality sensors compliant with the Centers for Disease Control and Prevention guidelines (Figure 14). In 2012, a sensor for the analysis of exhaled nitric oxide in asthma patients was introduced. And in 2013, a wireless metabolism tracker, Breezing, was demonstrated (Figure 15). This product, highlighted by the BBC,²⁹⁴ Forbes,²⁹⁵ and Scientific American,²⁹⁶ was a commercial success and received numerous awards and recognitions, including the Materials Research Society’s iMatsci, Consumer Technology Association’s 2017 Baby Tech Awards finalist, the Flinn Foundation Entrepreneur Award, and Arizona Commerce Authority’s Challenge. Refinement of the Breezing platform has culminating in the demonstration of an innovative smart mask device (Breezing Pro) that provides data that dieticians can use to provide precise nutritional and exercise advice to their clients (for more details see: https://breezing.com/resources/).

Figure 14.

Wireless wearable chemical sensor based on tuning fork sensors for selective and real-time detection of hydrocarbons (traffic pollutants) and acid vapors with sensitivity of ppb – ppm range, spatial resolution, and signal processing based on a cell phone app. The device was used to monitor the exposure of individuals to traffic pollutants.

Figure 15.

Evolution of products manufactured by: (A) Biosensing Instrument Inc. (https://biosensingusa.com/), (B) TF Health Co., dba Breezing (https://breezing.com/).

CONCLUSIONS AND OUTLOOK

Electrochemistry, molecular electronics, surface science, and chemical sensing were all impacted by the tremendous reach of NJ’s research over more than 30 years. In spite of its diversity, a common thread running through all of NJ’s science is an emphasis on applying simple physical principles, and asking simple questions based upon these principles, as a means for understanding complex phenomena. He applied this reductionist strategy successfully in all of the areas he investigated.

Now, with his prior work as a foundation, we are left to consider where each of these areas will progress: What are the frontiers in metabolic sensing, molecular electronics, DNA-based electronic devices, and optical microscopy and detection that could be advanced, or simplified using novel physical apparatuses? What questions could be answered by looking across these domains? In this Perspective, the goal has been to highlight some of the most exciting science in each of these areas. The new challenge is to apply the understanding generated by NJ’s work, along with that of many others over the past 25 years, to answer problems in fields other than molecular electronics, surface science, and electrochemistry. This challenge is in the hands of a new generation of scientists, including NJ’s many students and postdocs who trained with him and who will pursue this goal, with his inimitable style.