Electronic Chemical Sensors Based on Conductive Framework Materials

Abstract

The development of portable electronic chemical sensors is key to solving a number of challenges, including monitoring environmental and industrial hazards, as well as understanding and improving human health. Reticular materials possess several desirable characteristics that make them well suited for electroanalytical applications, including their high surface area, their atomically precise distribution of active sites, and their tunable properties that can be leveraged through modular reticular chemistry. This review highlights the emergence of conductive framework materials as active components in electrically-transduced chemical sensors, including the development of new materials for the detection of a wide variety of analytes in both gas and liquid phase. The efforts to gain fundamental understanding of the molecular interactions and sensing mechanisms between framework materials and analytes are described, along with applications of these materials on portable and flexible substrates. The article suggests areas for further study, including the study of material-analyte interactions at the molecular level, and the continued development of scalable methods for the integration of framework materials into low-power, portable sensing devices.

Graphical Abstract

Sensors that provide chemical information about the environment are highly useful in a number of contexts, including environmental monitoring, industrial safety, threat agent detection, and medical diagnostics.^1–3 The utility of these sensors is contingent upon their sensitivity and selectivity, that is, their ability to distinguish a specific analyte from interferent chemical species at relevant concentrations for the desired application and transduce signals accordingly.^2,4 Chemical sensors that use electronic methods of signal transduction can offer remote, continuous monitoring, low power transduction, and access to non-line-of-sight detection, while being easily integrated into existing systems already developed for physical sensors.^3,5 These electronic chemical sensors typically rely on a sensing material, which facilitates a material–analyte interaction between the target analyte and the sensor, and a transducer, which transforms the sensing event into a readable signal output, such as change in current or potential.³ The development of conductive sensing materials that can be easily integrated into compact and portable electronic device architectures represents a significant driving force in continued innovation within the realm of chemical sensors.³

Conductive, layered two dimensional (2D) metal–organic frameworks (MOFs) and covalent organic frameworks (COFs) constitute an emergent class of materials for chemical sensing.^2,3,6 Intrinsically conductive framework materials offer at least five advantages for use in electroanalysis. First, they are molecularly precise, leading to control of the material at a molecular scale that can be achieved through bottom-up synthesis.^2,6 Second, they have a high surface area and porosity, which increase the availability of potential active sites and allow for the diffusion of analytes through the material.^2,3 Third, such frameworks are highly modular and have properties that can be tuned through the strategic selection of the constituent building blocks, allowing for the design of materials optimized for interactions with specific target analytes.⁷ Fourth, the intrinsic conductivity of these materials is well suited for use in electronic chemical sensors and amenable for the transduction of electronic signals.⁸ Fifth, these frameworks can also have multi-functional capabilities, allowing for simultaneous detection, capture, and detoxification of hazardous analytes.^6,9 Recent developments in the use of layered, conductive framework materials have established the promise these materials in new sensor configurations with broad applicability to a variety of analytes, while seminal studies of fundamental structure–property relationships and material–analyte interactions are beginning to pave the way to strategic implementation of these scaffolds in electroanalysis.

This review highlights the use of conductive framework materials as active components in electronic chemical sensors, focusing on electroanalytical applications, and the electronic transduction of chemical signals. The reviewed examples include the use of conductive frameworks in amperometric, voltametric, and potentiometric sensing architectures, as well as chemical field effect transistors. Many, but not all, of the developments in framework-based electronic sensors in the past decade have utilized layered, conductive MOFs and COFs, whose intrinsic conductivity make them well suited for electrically-transduced sensing architectures. This review focuses on the development and implementation of intrinsically conductive framework materials for electronic chemical detection and the efforts to understand structure–property relationships and their impact on host–guest interactions between framework materials and analytes. We highlight recent progress in these areas, including the design of framework materials for sensing applications, the methods of fabrication of framework-based sensors in portable devices and on flexible substrates, and the insights into the fundamental understanding of sensing interactions between frameworks and analytes. This review does not cover sensors based on non-conductive frameworks, MOF/COF composites with materials such as metal oxides or metal nanoparticles, or the use of frameworks in optical sensors, as these have been covered in other recent reviews.^10–14 We seek to highlight the most promising uses of layered, conductive MOFs and COFs as electronic chemical sensors and suggest future directions for research in this area, including in situ and in silico studies of framework-analyte interactions and the development of sensor arrays.

SENSING ARCHITECTURES EMPLOYING LAYERED, CONDUCTIVE FRAMEWORK MATERIALS

The development of layered, conductive framework materials has been pivotal for enabling the use of MOFs and COFs in electronic chemical sensors. Prior to the development of conductive framework materials, framework-based chemical sensors mainly relied on optical detection methods,^1,15 or electro-mechanical sensor architectures, such as quartz crystal micro balance (QCM) electrodes and surface acoustic wave (SAW) devices.^1,11 Following several seminal reports of triphenylene-based conductive MOFs,^16–19 the use of framework materials as active components in the electronic transduction of chemical signals has blossomed.^{2–4,7,20–22} Intrinsically conductive layered framework materials have been employed in sensing architectures that exploit their many valuable properties, including conductivity, porosity, and molecular precision. The types of sensing architectures that have employed layered, conductive MOFs and COFs are described below.

Chemiresistive Gas Sensors:

Chemiresistive sensors constitute the simplest sensing architecture employed for the detection of gases with layered, conductive framework materials.^{17,18,23–35} Chemiresistive sensors have been used since the 1960s,^36,37 employing a variety of materials, including metal oxides,^38,39 carbon nanotubes,⁴⁰ and nanocomposite materials.⁴¹ A chemiresistor consists of a semiconducting material bridging two electrodes, in which sensor response is based on a change in conductivity of the material as a result of gas binding (See Fig 1a). In this type of device, the response of the sensor (S) is typically calculated as a percent change from the baseline resistance of the device (see eq. 1).

Figure 1:

Modes of electronic chemical detection that can employ conductive framework materials. (a) Chemiresistive sensing architecture for detecting gaseous analytes. (b) Voltammetric sensing architecture for liquid-phase detection. (c) Potentiometric sensing architecture for ion detection. (d) Sensing architecture for a chemical field effect transistor (chemFET)- based sensor. Abbreviations: WE= working electrode, RE= reference electrode, CE= counter electrode, ISM = ion-selective membrane.

$$S=\frac{\mathrm{\Delta }R}{R_0}=\frac{R_{\mathit{analyte}}-R_0}{R_0}\times 100\%$$ (1)

Based on the relationship of Ohm’s law (V = IR) , the response can also be calculated in an analogous expression of change in current (eq. 2).

$$S=\frac{\mathrm{\Delta }I}{I_0}=\frac{I_{\mathit{analyte}}-I_0}{I_0}\times 100\%$$ (2)

The change in current in the material is directly proportional to the change in conductance (ΔG) of the device, and normalized chemiresistive sensing data is often presented in terms of the expression $\frac{-\mathrm{\Delta }\mathrm{G}}{G_0}$.

In comparing chemiresistive sensors, two relevant metrics are frequently extracted: response at saturation and initial rate of response. The response at saturation can refer to the total change in conductance over the time of exposure, or the ΔR or ΔI value at which the sensor response no longer changes during gas exposure. Initial rate of response refers to the change in conductance in a given time period at the beginning of exposure, where the response appears to be approximately linear as a function of time. Comparing these metrics for different gases and concentrations allows for the calibration of sensors and differentiation between gases.²⁰ Chemiresistive sensors can exhibit various levels of reversibility in response, which are often correlated to the nature of intermolecular interactions between the material and the analyte. Typical options include: 1) a reversible response, in which the resistance returns to a baseline value when removed from the reactive gas; 2) an irreversible or dosimetric response, in which the device resistance is permanently altered by gas exposure and does not return to baseline following removal from analyte; and 3) a partially reversible response where some or all of the responses show a partial permanent change in resistance coupled with a fully reversible change upon exposure to the analyte.²¹ The nature of the response is typically determined by material-analyte interactions: if the analyte binds permanently and/or initiates an irreversible change to the material structure or electronic properties, the response is likely to be irreversible, whereas reversible binding tends to lead to a reversible sensing response.²²

Voltammetric and Amperometric Sensors:

In the liquid phase, voltammetric and amperometric sensing architectures are frequently employed, especially for redox-active analytes.^42–45 A voltammetric sensor uses a three-electrode architecture: (1) a working electrode, which has the active sensing material, (2) a counter electrode, usually an inert metal such as platinum, and (3) a reference electrode, often a silver/silver chloride or calomel electrode (see Fig. 1b). The potential of the working electrode can be varied relative to the reference electrode, and increases in current can be correlated to the oxidation and reduction of analytes at the electrode surface. Cyclic voltammetry is a basic electrochemical technique that is frequently employed in the development of chemical sensors. This technique, in which the potential of a working electrode is ramped linearly versus time in cyclical phases, can detect and differentiate redox active analytes, and can yield vital information about electron transfer rates at electrode surfaces.⁴⁶ This kind of analysis is a key method for evaluating and comparing the performance of novel sensing materials, including frameworks.^42,47 Elgrishi et al. have published an excellent primer on cyclic voltammetry for further details.⁴⁶

Pulsed voltammetry techniques, such as differential pulse voltammetry (DPV), are frequently employed for quantitative detection of analytes and determination of sensor sensitivity and selectivity.^42,44,45 Pulse voltammetry techniques remove background due to charging current, allowing for greater signal to noise ratios and higher sensitivity to low concentrations of analytes.⁴⁸ A typical DPV experiment features sequential additions of analyte followed by a DPV scan, from which a calibration curve correlating analyte concentration and maximum response can be generated. Such curves can be used to determine the sensitivity of a sensor, the limit of detection (LOD), and the limit of linearity (LOL). Constant potential amperometry (CPA) can also be used for quantitative detection. In this technique, the working electrode is held at a constant voltage and the change in current is recorded in response to the addition of analyte and subsequent oxidation or reduction at the electrode surface. This technique generates a similar calibration curve to DPV experiments, and is known to have particularly high sensitivity to low concentrations of analyte.

Potentiometric Sensors:

Potentiometric sensing architectures are frequently employed for the detection of ions. Such sensors employ a two-electrode architecture, measuring the potential difference between a working electrode and a reference electrode, typically in zero current conditions (Fig. 1c).⁴⁹ The working electrode, which contains a material engineered to bind with a specific ion, changes potential according to the concentration of those ions present in the sensing medium.⁵⁰ The measured potential between the electrodes (E_M) is defined as:

$$E_M=E_{PB}+E_D+\mathit{constant}$$ (3)

Where E_PB is the potential at the sample/working electrode boundary, E_D is the diffusion potential inside the electrode membrane or film, and the constant is usually arbitrarily set to zero. In the simplest model of potentiometric sensing, electrochemical equilibrium is assumed, so E_D is also assumed to be zero. Thus, the measured potential is defined by the Nernst equation:

$$E_{PB}=E_i^0+\frac{RT}{z_{i}F}ln\frac{c_i}{\overline{c_i}}$$ (4)

Where $E_i^0$ is the potential of the cell at standard conditions, c_i and ${\bar{c}}_i$ are the concentrations of the ion in the sample and ion membrane respectively, z_i is the ionic charge, and R, T, and F are the universal gas constant, the absolute temperature, and Faraday’s constant, respectively.⁵⁰ In the idealized case represented by equations (3) and (4), the working electrode membrane only allows the passage of the target ion, and excludes other interfering species, including water and other ions. Thus the development of ion selective materials and membranes (ISMs) is a key aspect of research into potentiometric sensing.⁵¹ Many ionophores and ISMs are organic supra-molecular host molecules, sometimes incorporated into conductive polymer films.^52,53 Conductive MOFs have been employed directly as an ion-binding material,⁵⁴ and as a current collecting material underneath an ISM layer.^55,56

Chemical Field Effect Transistors (ChemFET):

Sensors based on chemical field effect transistors were first developed in the 1980s, featuring primarily organic semiconductors and metal oxides.^57,58 This sensing architecture, which can be used in the gas phase or in the liquid phase, features a source and drain connected by a semiconducting material, and a gate connected to the drain and source via an electrolyte medium (Fig. 1d). Field effect transistors have many applications in electronics, but in a sensing context the electrolyte connecting the gate to the drain and source serves as the sensing medium. As the potential gap between either the source and the drain, or the source and the gate is changed, the measured current varies depending on the presence of analytes. When analytes bind to the semiconducting material connecting the drain and the source, they impact the flow of current across the material, and these changes can be correlated to analyte concentrations.⁵⁷ This sensing architecture is highly amenable to miniaturization, and organic semiconductor materials have been implemented into flexible FET-based biosensors.⁵⁷ Despite their wide use with other materials, chemFETs are an emerging architecture for MOF-based sensors, with only a few examples of MOF-based chemFETs reported thus far.^59–62

Other Sensing Architectures:

Select examples have been reported incorporating layered conductive MOFs and COFs into other architectures, including photoelectrochemical and electrochemiluminescence sensors. In photoelectrochemical devices, the interaction of light with the framework material initiates electronic transitions that enable the sensing response of the device.^63–65 Electrochemiluminescence (ECL) sensors rely on luminescence generated from electrochemical processes, which can be modulated by the presence of the analyte.^66,67 These types of devices are emerging architectures, with very limited reports to date.

TYPES OF FRAMEWORKS USED FOR ELECTRICALLY TRANSDUCED SENSING APPLICATIONS

2D layered conductive frameworks, especially MOFs, have been the conductive framework materials most frequently employed in electronic chemical sensors thus far. The design of these materials features a layered structure that leverages π-conjugated linkers connected by metal nodes to form extended-π-d conjugated sheets, which exhibit ‘through-bond’ conductivity.⁸ These materials have been incorporated into electronic chemical sensors because they offer intrinsic conductivity and metal-centered host-sites in a porous framework, allowing for selective interactions with analytes in a high surface-area material.¹⁰ 2D layered frameworks are also highly modular, with a wide variety of ligands and metals incorporated into a variety of structures.²² The 2D conductive MOFs and COFs that have been employed in electronic chemical sensors are described below.

Metal–Organic Frameworks:

Layered, conductive MOFs can encompass several topologies and include a variety of metals and linkers. Conductive MOFs based on hexatopic triphenylene ligands were the first reported materials in this class,¹⁶ and also the first framework materials used as chemiresistive sensors.¹⁷ MOFs with triphenylene-based linkers can have one of several heteroatoms (See Fig. 2a–i,ii): oxygen— hexahydroxytriphenylene (HHTP), nitrogen— hexaiminotriphenylene (HITP); or sulfur— hexathioltriphenylene (HTTP). MOFs with benzene-based linkers featuring the same heteroatoms have also been reported.^23–25,68 A mixed-linker MOF featuring both benzene-based and triphenylene-based units has been employed for chemiresistive sensing²⁶ and voltammetric sensing.⁶⁹ Another mixed-linker MOF, in which a Cu₃(HHTP)₂ MOF was ‘doped’ with a small amount of HITP linkers, was reported as a chemiresistive sensor for volatile organic compounds.³⁰ Mixed-metal HHTP MOFs, with combinations of Ni, Cu, and Co nodes, have also been reported as voltammetric^43,44 and chemFET sensors⁵⁹, but the specific placement of the metal nodes in these materials is uncertain, as is their precise elemental composition. Triphenylene- and benzene-based materials have a few key advantages over other conductive MOF materials in chemical sensing applications, including the structural modularity described above, with many possible combinations of metal nodes and linker heteroatoms.⁷⁰ They are also synthetically accessible, featuring many ligands that are commercially available, and others that can be synthesized from commercially available precursors in only a few steps.²⁸ Control over MOF nanoscale morphology is another key advantage, with several strategies available to control the sizes and shapes of MOF particles.^71,72

Figure 2:

Examples of layered conductive (a) metal–organic frameworks and (b) covalent organic frameworks employed as electronic chemical sensors. Molecular components and space-filling models of framework structures are shown, as well as the metal-centered host sites contained in each framework. Space-filling model for NiPc-CoTAA COF reproduced with permission from “Conductive Metallophthalocyanine Framework Films with High Carrier Mobility as Efficient Chemiresistors,” Y. Yue, P. Cai, X. Xu, H. Li, H. Chen, H.-C. Zhou, N. Huang, Angew. Chem Int. Ed., Vol. 60, No. 19, (Ref. 34) Copyright 2021 John Wiley and Sons. Space-filling model for M-TPCOF reproduced with permission from “Porphyrin-Based COF 2D Materials: Variable Modification of Sensing Performances by Post-Metallization,” M. Liu, Y.-J. Chen, X. Huang, L.-Z. Dong, M. Lu, C. Guo, D. Yuan, Y. Chen, G. Xu, S.-L. Li, Y.-Q. Lan, Angew. Chem Int. Ed., Vol. 61, No. 12, (Ref. 35) Copyright 2022 John Wiley and Sons.

An alternative MOF topology can be created through the incorporation of metallophthalocyanine linkers (See Fig. 2a–iii) into a framework.⁷³ These MOFs have the potential for a regularly repeating bi-metallic structure, in which the metal center of the metallophthalocyanine is different from the metal nodes linking the metallophthalocyanine units.³¹ This structure allows for the presence of multiple types of metal-centered host sites that can each interact differently with analytes of interest.⁷⁴ These active sites can even work cooperatively to enhance material function.^75,76 The metallophthalocyanine motif is also bio-inspired, mimicking the porphyrin based metal-center that is featured in many proteins which interact with gases in biological systems.⁷⁷ A key limitation for these metal-lophthalocyanine-based materials is synthetic accessibility, as many of the octa-functionalized metallophthalocyanines required for MOF formation are not currently commercially available.

Thus far, the metal nodes reported for these MOFs have largely been limited to 1^st row transition metals. However, the incorporation of larger metal nodes, such as 6^th row elements, allows access to more diverse coordination geometries, unlocking novel framework topologies, and thus potentially new sensing modalities.^29,78 Lanthanide-based HHTP MOFs have been reported, but have not yet been reported as electronic chemical sensors.⁷⁸ Our group recently reported a new, HHTP-based coordination polymer with bismuth nodes, which had a novel topology and showed promising chemiresistive sensing performance (Fig. 2a–ii).²⁹ Figure 2a shows molecular components, structures and metal-centered host sites for the metal–organic frameworks described in this review.

Covalent Organic Frameworks:

The scope of reported layered, conductive COFs, and thus their uses in electronic chemical sensors, is more limited than that of MOFs. However, several examples of COF-based chemiresistors have been reported. Layered, conductive COFs have primarily featured metallophthalocyanines and metalloporphyrins, as the metal centers incorporated into these materials are hypothesized to function as gas-binding sites and/or catalytic active sites.⁷⁹ Other non-conducting COF materials have been used in composites with conductive substrates for electronic COF-based sensors.² In 2019, we reported a layered, conductive COF based on Nickel-phthalocyanine subunits connected via pyrazine linkages (Fig. 2b–i). This material had good intrinsic conductivity, and proved to be an excellent chemiresistive sensing material for NH₃, H₂S, NO, and NO₂.³³ A recent report followed a similar approach to generate a metallophthalocyanine-based COF by condensing octahydroxy- and octafluoro-functionalized metallophthalocyanine monomers to form a dibenzo-p-dioxin linked COF. By varying the metal center in each type of functionalized phthalocyanine, a bi-metallic COF with alternating copper and nickel centers was formed.⁸⁰ Another approach to incorporate multiple metal centers into a conductive COF from metallophthalocyanine subunits has been reported, in which nickel phthalocyanine monomers are covalently connected through cobalt tetraazaanulene complexes (See Fig. 2b–i).³⁴ Compared with their MOF analogs, metal-lophthalocyanine-based COFs have superior chemical and thermal stability, resisting degradation under high temperatures and treatments with strong acid and base,² and their limits of detection also surpass that of MOFs for a variety of analytes.³³ Conductive COFs with porphyrin-based metal centers can also be synthesized through post-metalation of a porphyrinic COF, H₂-TPCOF, to create cobalt and copper containing chemiresistor materials (Fig. 2b–ii).³⁵ The enhanced stability and sensitivity of layered conductive COF materials indicates their potential to form highly robust chemical sensors. Figure 2b shows molecular components, structures, and metal-centered host sites for the metal–organic frameworks described in this review.

RECENT PROGRESS IN CHEMICAL SENSING WITH LAYERED CONDUCTIVE FRAMEWORK MATERIALS

Gas-Phase Analytes:

The detection of hazardous gases in air is of vital importance to protecting human health and safety in areas where such threat agents are common. Toxic gases such as ammonia, nitric oxide, hydrogen sulfide, and carbon monoxide are common byproducts in industries such as mining and oil refining, and are hazards produced more generally by automobile exhaust.⁹⁰ Other gases, including volatile organic compounds (VOCs), such as formaldehyde and acetone, are also common pollutants produced by a variety of industrial processes.⁹¹ At the same time, gasotransmitters (NO, CO, H₂S) and VOCs are produced by the human body and their presence in human breath can be indicative of certain disease states.^92,93 Both industrial safety concerns and the need for improved healthcare diagnostics have motivated the development of framework materials as gas sensors for these analytes. Reactive gases such as NH₃, NO, and H₂S can also serve as spectroscopic probe molecules to elucidate properties of the framework, including surface chemistry,⁹⁴ material analyte interactions,^82,94,95 and charge transfer mechanisms.⁹⁶ Table 1 details the reported example of conductive framework-based gas sensors, as well as relevant performance metrics.

Table 1:

Detection of Gas Phase Analytes with Layered Conductive Framework Materials

Analyte	Framework Material(s)	Sensing Architecture	LOD	Detection Range	Details	Ref.
NH3	Cu3(HITP)2 on Au IDE	Chemiresistive (0.1 V)	0.5 ppma	0.5 – 10 ppm	Detection in dry N2, air, and up to 60% RH, reversible response	17
	Ni3(HHTP)2 & Cu3(HHTP)2 on graphite electrodes on shrinkable polymer films	Chemiresistive (1.0 V)	Not reported	2.5 – 80 ppm	Detection in dry N2, device performance unaffected by 1000 ppm water vapor, 2-material array for detection of NH3, NO, and H2S, reversible response (Cu3(HHTP)2)	27
	Cu3(HHTP)2 thin film on Si substrate with patterned Au electrodes	Chemiresistive (5.0 V)	0.5 ppmb	1 – 100 ppm	Detection in dry air, less than 20% response to 100 ppm benzene, toluene, ethylbenzene, H2, CO, n-hexane, methane, methanol, reversible response	81
	NiPc-O8-Ni & Ni-Pc-O8-Cu on Au IDEs (5 μm gap)	Chemiresistive (0.1 V)	0.31 ppmc (Ni) 0.33 ppmc (Cu)	2 – 80 ppm	Detection in dry and humid (5000 ppm H2O) N2, 4-material array for detection of NH3, H2S and NO, partially reversible response (Ni-linked) reversible response (Cu-linked)	31
	Ni-COF-DC-8 on Au IDEs (5 μm gap)	Chemiresistive (1.0 V)	70 ppbc	2 – 10 ppm	Detection in dry N2, partially reversible response	33
	Cu3(HHB)2, Cu3(HHB)(HHTP), Cu3(HHTP)2 nanorods and nanosheets on Ag/Pd IDEs	Chemiresistive (5.0 V)	0.45 ppmb (Cu3(HHB)2)	1 – 100 ppm	Detection in dry air, reversible response (Cu3(HHB)(HHTP), Cu3(HHTP)2 nanorods and nanosheets), partially reversible response (Cu3(HHB)2)	23
	Cu3(HHB)(HHTP) on Au IDEs	Chemiresistive (5.0 V)	0.35 ppmc	1 ppm – 1 %	Detection in dry air, selective against ethanol and acetone, reversible response	26
	Cu3(BHT)2 thin film on Si/SiO2 substrate with patterned Au electrodes	Chemiresistive (0.01 V)	0.21 ppmc	1 – 100 ppm	Detection in dry and humid air up to 80% RH, minimal response to acetone, toluene, THF, DMC, hexane, H2O, EAC, reversible response, detection on flexible PET substrate	25
	Bi(HHTP) on Au IDEs, (10 μm gap)	Chemiresistive (1.0 V)	0.29 ppmc	5 – 40 ppm	Detection in dry N2, irreversible response	29
	Cu3(HHB)2 on Au IDEs (5 μm gap)	Chemiresistive (1.0 V)	14 ppbc	1 – 40 ppm	Detection in dry N2, reversible response, saturation within 1 minute	68
NO	Ni3(HHTP)2 & Cu3(HHTP)2on graphite electrodes on shrinkable polymer films	Chemiresistive (1.0 V)	Not reported	2.5 – 80 ppm	Detection in dry N2, device performance unaffected by 1000 ppm water vapor, 2-material array for detection of NH3, NO, and H2S, irreversible response	27
	Ni3(HHTP)2 & Ni3(HITP)2 on cotton textile	Chemiresistive (1.0 V)	0.16 ppm (Ni3(HITP)2) 1.4 ppm (Ni3(HHTP)2)	0.1 – 80 ppm	Detection in dry N2, device performance unaffected by 5000 ppm water vapor, 2-material array for detection of NO and H2S, filtration capability	28
	NiPc-O8-Ni & Ni-Pc-O8-Cu on Au IDEs (5 μm gap)	Chemiresistive (0.1 V)	1.06 ppbc (Ni) 1.00 ppbc (Cu)	20 ppb – 1 ppm	Detection in dry and humid (5000 ppm H2O) N2, 4-material array for detection of NH3, H2S and NO, partially reversible response (Ni-linked) irreversible response (Cu-linked)	31
	Ni-COF-DC-8 on Au IDEs (5 μm gap)	Chemiresistive (1.0 V)	5 ppbc	20 ppb – 40 ppm	Detection in dry N2, partially reversible response	33
	Cu3(HHTP)2 on cotton textile	Chemiresistive (1.0 V)	1 ppma	1 – 20 ppm	Detection in dry N2, air and humidity (5000 ppm) irreversible response, filtration & decontamination capability	82
	Bi(HHTP) on Au IDEs, (10 μm gap)	Chemiresistive (1.0 V)	0.15 ppmc	5 – 40 ppm	Detection in dry N2, reversible response	29
	Cu3(HHB)2 on Au IDEs (5 μm gap)	Chemiresistive (1.0 V)	13 ppbc	1 – 20 ppm	Detection in dry N2, partially reversible response	68
NO2	Ni-COF-DC-8 on Au IDEs (5 μm gap)	Chemiresistive (1.0 V)	16 ppbc	2 – 40 ppm	Detection in dry N2, partially reversible response	33
	NiPc-CoTAA COF on Au IDE	Chemiresistive (5.0 V)	Not reported	1 – 40 ppm	Detection in dry N2, selectivity against NO, NH3, H2S, and H2, reversible response	34
	Co-TPCOF on IDE	Chemiresistive (5.0 V)	6.8 ppb	5 – 100 ppm	Detection in dry air, no sensor response to NH3, CO, H2S, SO2, CH4, methanol, ethanol, benzene, toluene, and methylamine, reversible response	35
	COF-CuNiPc on Ti/Au IDE	Chemiresistive (3.0 V)	5.4 ppbc	50 ppb – 10 ppm	Detection in dry air, 20 – 80% RH, selective against H, NH3, acetone, formaldehyde, chloroform, ethanol, EAC, n-hexanal, full device recovery with UV light	80
	Ni3(HHTP)2 on Au electrodes on flexible PI substrate	ChemFET (Vds 0.1 V, Vgs 0 V)	56 ppbc	50 ppb – 10 ppm	Detection in air, selective against CO2, NH3, N2O, H2S, SO2, ethanol, benzene, acetone, triethylamine, formaldehyde, sensing performance maintained in bent device, green light-assisted device recovery	83
H2S	Ni3(HHTP)2 & Cu3(HHTP)2 on graphite electrodes on shrinkable polymer films	Chemiresistive (1.0 V)	Not reported	2.5 – 80 ppm	Detection in dry N2, device performance unaffected by 1000 ppm water vapor, 2-material array for detection of NH3, NO, and H2S, reversible response (Cu3(HHTP)2), partially reversible response (Ni3(HHTP)2)	27
	Ni3(HHTP)2 & Ni3(HITP)2 on cotton textile	Chemiresistive (1.0 V)	0.52 ppm (Ni3(HITP)2) 0.23 ppm (Ni3(HHTP)2)	0.1 – 80 ppm	Detection in dry N2, device performance unaffected by 5000 ppm water vapor, 2-material array for detection of NO and H2S, filtration capability	28
	NiPc-O8-Ni & Ni-Pc-O8-Cu on Au IDEs (5 μm gap)	Chemiresistive (0.1 V)	32 ppbc (Ni) 19 ppbc (Cu)	0.2 – 20 ppm	Detection in dry and humid (5000 ppm H2O) N2, 4-material array for detection of NH3, H2S and NO, irreversible responses	31
	Ni-COF-DC-8 on Au IDEs (5 μm gap)	Chemiresistive (1.0 V)	204 ppbc	5 – 80 ppm	Detection in dry N2, partially reversible response	33
	Cu3(HHTP)2 on cotton textile	Chemiresistive (1.0 V)	1 ppma	1 – 20 ppm	Detection in dry N2, air and humidity (5000 ppm) irreversible response, filtration & decontamination capability	82
	Cu3(HHB)2 on Au IDEs (5 μm gap)	Chemiresistive (1.0 V)	25 ppbc	1 – 40 ppm	Detection in dry N2, partially reversible response	68
	Ni3(HITP)2 thin film on alumina substrate with patterned Au IDE (150 μm gap)	Chemiresistive	3 ppbc	0.1 – 5 ppm	Detection in dry N2 and humid air, dosimetric response, detection on flexible substrate, selectivity against NO, HCHO, CO, CH4, C2H6, H2, C2H2	84
SO2	Ni3(HHTP)2 SiO2 substrate with Cr/Au electrodes	ChemFET (Vds 0.5 V)	625 ppb	625 – 1125 ppb	Detection in dry air, selectivity against CO, NO2, CH4, C2H2	85
CO	CoPc-O8-Cu & NiPc-O8-Cu on Au IDEs (5 μm gap)	Chemiresistive (0.1 V)	0.53 ppm (CoPc) 3.0 ppm (NiPc)	10 – 80 ppm	Detection in dry N2, dry air, and humid N2, reversible responses, selective against NO2, CO2	74
	Zn3(HHTP)2 on B-doped Si with Au electrodes	Chemiresistive and ChemFET	3.96 ppmc	10 – 100 ppm	Detection in dry and humid air, selective against SO2 and NH3	86
CO2	Cu3(HIB)2	Chemiresistive (0.8 V)	Not reported	400 – 2500 ppm	Detection from 0 – 80% RH	87
H2O	Ni3(HHTP)2 on PAN nanofiber membrane	Chemiresistive	Not reported	20 – 70 % RH	Used for monitoring of human breath and PM filtration	88
	CuPc-(NH2)8-Ni thin film on SiO2 substrate with patterned Au electrodes	Chemiresistive (1.0 V)	10 ppm	50 – 600 ppm	Response and recovery time modulated by coating MOF film with alkylsilanes	32
	Ni3(HIB)2 on Si substrate with Au/Ti electrodes	Chemiresistive (1.0 V)	Not reported	20 – 70 % RH	Detection in N2, selective against ethanol and acetone	24
	Cu3(HIB)2 thin film on SiO2 substrate with patterned Au electrodes	Chemiresistive (1.0 V)	Not reported	200 – 1000 ppm	Detection in N2, reversible response	89
VOCs	Cu3(HHTP)2, Cu3(HITP)2, & Ni3(HITP)2 on Au IDEs	Chemiresistive (0.1 V)	Not reported	200 ppm for all VOCs	Detection of 16 VOCs- alcohols, ketones, ethers, amines, aliphatics, aromatics, 3-material array for differentiating among functional groups	18
	Bi(HHTP) on Au IDEs, (10 μm gap)	Chemiresistive (1.0 V)	41.2 ppmc (acetone) 185 ppmc (ethanol) 278 ppmc (methanol) 50.2 ppmc isopropanol	280 – 670 ppm acetone 347 – 2097 ppm ethanol 604 – 2623 ppm methanol 89.7 – 822 ppm isopropanol	Detection in dry N2, reversible responses	29
	CuPc-(NH2)8-Ni thin film on SiO2 substrate with patterned Au electrodes	Chemiresistive (1.0 V)	Not reported	100 – 500 ppm methanol, ethanol, acetone, isopropanol	Response and recovery time modulated by coating MOF film with alkylsilanes	32
	Cu3(HHTP)2 thin film doped with HITP on SiO2 substrate with Au electrodes	Chemiresistive (5.0 V)	0.024 ppmb benzene	1 – 100 ppm benzene	Detection in dry air, response and recovery times modulated with degree of HITP doping, selectivity against acetone, butanone, H2, ethylbenzene, methane, triethylamine, NH3	30

Abbreviations: LOD = limit of detection, IDE = interdigitated electrode, RH = relative humidity, THF = tetrahydrofuran, DMC = dimethyl carbonate, EAC = ethyl acetate, PI = polyimide, PAN = polyacrylonitrile, PM = particulate matter.

^aExperimental LOD based on lowest concentration delivered/detected.

^bEstimated LOD based on 10% max response.

^cTheoretical limit of detection based on S/N = 3.

Ammonia:

Triphenylene-based MOFs were the first framework materials to be used as chemiresistive sensors, beginning in 2015 with the use of Cu₃(HITP)₂ to detect ammonia. This demonstration from Dincă and coworkers was the first report of Cu₃(HITP)₂, and showed a 0.5 ppm limit of detection for the MOF in response to NH₃.¹⁷ (The OSHA permissible exposure limit (PEL) for ammonia is 50 ppm over a 10 hour workday).⁹⁷ Following this initial report, other triphenylene-based MOFs, such as Cu₃(HHTP)₂, were shown to be effective chemiresistive sensors for ammonia.⁸¹ In 2016, we reported the use of hexahydroxytriphenylene MOFs (Ni₃(HHTP)₂ & Cu₃(HHTP)₂) for the detection of ammonia and other gases (Fig. 3a–c) and demonstrated the potential for direct self-assembly of MOFs into chemiresistive sensors. Because Ni₃(HHTP)₂ did not respond to ammonia at all, the 2-material sensor array successfully differentiated ammonia from NO and H₂S. In addition, this report demonstrated the potential for direct-from-solution fabrication of MOF-based sensors.²⁷

Figure 3.

(a-c) Chemiresistive sensing response of HHTP-based MOFs to ammonia. (a) Response vs. concentration for Ni₃(HHTP)₂ (blue) and Cu₃(HHTP)₂ (orange). (b-c) Chemiresistive sensing trances of repeated exposures to NH₃ for (b) Cu₃(HHTP)₂ and (c) Ni₃(HHTP)₂. (a-c) Reproduced from M. K. Smith, K. E. Jensen, P. A. Pivak, K. A. Mirica, “Direct Self-Assembly of Conductive Nanorods of Metal–Organic Frameworks into Chemiresistive Devices on Shrinkable Polymer Films.” Chem. Mater. 2016, 28, 5264–5268 (Ref. 27) Copyright 2016 American Chemical Society. (d-e) Intermolecular interactions between M₃(HXTP)₂ MOFs and NH₃. (d) Potential Lewis acid sites (LAS) and Brønsted acid sites (BAS) for the MOFs. (e) In situ DRIFTS spectra of triphenylene based MOFs after exposure to 10% NH₃ and subsequent purging. Spectra of Cu₃(HITP)₂, Ni₃(HITP)₂, Cu₃(HHTP)₂, Ni₃(HHTP)₂ are shown, as well as a Ru-TiO₂ control. Bands corresponding to Lewis acid sites (LAS) and Bronsted acid sites (BAS) for NH₃ and NH₄⁺ are labeled accordingly. (d-e) Reproduced from R. M. Stolz, A. Mahdavi-Shakib, B. G. Frederick, K. A. Mirica, “Host–Guest Interactions and Redox Activity in Layered Conductive Metal–Organic Frameworks” Chem. Mater., 2020, 32 7639-7652 (Ref. 94) Copyright 2020 American Chemical Society. (f-h) Structure-property relationships of Cu-based MOFs in chemiresistive gas sensing of ammonia (f) Structures of MOFs tested. (g) PXRD traces for the MOFs tested, including Cu₃(HHTP)₂ nanorods (NRs) and nano-sheets (NSs) showing the shifts in (100) and (002) peak corresponding to changes in pore size and interlayer distance, respectively. (h) Overlaid, normalized chemiresistive sensing traces for the MOFs in response to 100 ppm NH₃, showing the differences in response time (t_res) and recovery time (t_rec). (f-h) Reproduced from M.-S. Yao, P. Wang, Y.-F. Gu, T. Koganezawa, H. Ashitani, Y. Kubota, Z.-M. Wang, Z.-Y. Fan, K. Otake, S. Kitagawa, “A comparative study of honeycomb-like 2D π-conjugated metal–organic framework chemiresistors: conductivity and channels” Dalton Trans. 2021, 50, 13236-13245, (Ref. 23) with permission of the Royal Society of Chemistry. (i-n) Chemiresistive sensing of H₂S with a Cu₃(HHTP)₂ textile device. (i-j) SEM images of Cu₃(HHTP)₂ nanowires grown from deposited Cu⁰ on cotton textile. (k) Chemiresistive sensing traces of Cu₃(HHTP)₂ with 1-80 ppm H₂S in dry nitrogen. (l) Sensor response at saturation for H₂S in dry and humid environments. (m) In situ DRIFTS spectra of Cu₃(HHTP)₂ powder exposed to 1% H₂S. (n) Chemiresistive response of Cu₃(HHTP)₂ textile to 80 ppm H₂S following chemical and physical stresses. Reproduced from A. M. Eagleton, M. Ko, R. M. Stolz, N. Vereshchuk, Z. Meng, L. Mendecki, A. M. Levenson, C. Huang, K. C. MacVeagh, A. Mahdavi-Shakib, J. J. Mahle, G. W. Peterson, B. G. Frederick, K. A. Mirica, “Fabrication of Multifunctional Electronic Textiles Using Oxidative Restructuring of Copper into a Cu-Based Metal–Organic Framework” J. Am. Chem. Soc., 2022, 144, 23297-23312 (Ref. ⁸²) Creative Commons CC-BY-NC-ND 4.0.

Other triphenylene- and benzene-based frameworks have been employed for chemiresistive detection of ammonia. Kitagawa and coworkers synthesized a copper-based mixed-ligand MOF, featuring alternating HHTP and HHB ligands. A chemiresistive gas sensor incorporating this Cu₃(HHB)(HHTP) MOF had a limit of detection of 0.35 ppm for ammonia. This approach showcases the tunable nature of MOF properties such as porosity and conductivity through the choices of metals and linkers, and the potential impact of this modulation on sensing performace.²⁶ Chen et al. developed a spin-coating based method for synthesizing thin films of Cu-BHT on a silica substrate. They fabricated a chemiresistive sensor by pattering gold electrodes directly onto the Cu-BHT film, and the device proved to be effective at sensing ammonia with an LOD of 0.23 ppm (see Table 1). This approach demonstrated the potential of MOFs to be directly self-assembled into sensor architectures, enabling fast and facile device fabrication.²⁵

The development of metallophthalocyanine-based framework materials offered an alternative approach for NH₃ detection. NiPc-O₈-Ni and NiPc-O₈-Cu MOFs had 0.31 and 0.33 ppm limits of detection for ammonia, respectively.³¹ The NiPc-O₈-Cu MOF had a fully reversible response and maintained sensing performance upon repeated ammonia exposure and recorvery.³¹ The NiPc-based COF, COF-DC-8, offered improved sensing performance, with a 70 ppb LOD (see Table 1 for limits of detection).³³ EPR and XPS studies of the COF surface showed only minimal changes after NH₃ exposure, suggesting that ammonia sensing with the material is facilitated by a weak binding interaction, which leads to a reversible response.³³

While conclusions can be drawn about the molecular mechanisms of sensing based on chemiresistive sensing experiments, it is can be difficult to untangle the molecular phenomena from the impact of material structure and morphology. Martí-Gastaldo and coworkers undertook an investigation of the structural changes in Cu₃(HHTP)₂ that drive chemiresistive sensing responses.⁹⁵ Fabrication of extremely thin films (30 nm) of the MOF allowed for the minimization of the effects of nanoscale morphology, while making molecular scale changes more prominent. Monitoring the changes in conductivity of the MOF film first in response to vacuum, dry nitrogen, and ambient conditions, showed the important influence of the surrounding atmosphere, even with non-reactive gasses, upon the measured resistance.⁹⁵ However, the authors noted the greatest change in resistance of the MOF film occurred in response to ammonia, consistent with previous uses of Cu₃(HHTP)₂ in chemiresistive sensing.⁹⁵ They rationalized their experimental data with DFT calculations, which showed that the MOF layers distort significantly as a result of the binding of water and ammonia, and that changes in inter-layer distances impact the conductivity of the MOF.⁹⁵ This distortion was proposed to be driven by changes in coordination geometry of the copper centers upon the binding of NH₃ and H₂O, in contrast with N₂, which exhibited a weak interaction with the MOF.⁹⁵ Infrared reflection absorbance spectroscopy (IRRAS) of the films after NH₃ exposure confirmed the binding of ammonia to the copper (II) centers in the MOF.⁹⁵

Comparisons between a series of related framework materials can help to discern the impact that structural features, such as metal center and linker heteroatom, have on material–analyte interactions and the sensing properties of these materials. We sought to compare multiple M₃(HXTP)₂ MOFs and their interactions with ammonia in order to further understand the role of metal center and surface chemistry in chemiresistive sensing interactions.⁹⁴ The use of in situ Diffuse Reflectance Infrared Spectroscopy (DRIFTS) showed both the binding of ammonia to Cu- and Ni- HHTP and HITP at Lewis acid and Bronsted acid sites, as well as changes to the broad electronic absorbance that indicated redox shifts driven by NH₃ binding (Fig. 3d–e). These changes were reversible for the copper-based materials, and irreversible for the nickel-based materials. Electron paramagnetic resonance (EPR) and x-ray photoelectron spectroscopy (XPS) showed that the Cu-based MOFs exhibited a shift in the Cu⁺/Cu²⁺ toward Cu²⁺ upon NH₃ binding, while the Ni-based MOFs did not undergo any redox changes in response to ammonia.⁹⁴ The use of in situ spectroscopic methods allowed for the study and comparison of both reversible and irreversible interactions of analytes with framework materials.

Properties, such as porosity and conductivity, can greatly influence the chemiresistive sensing performance of framework materials, but it is often difficult to untangle the contributions of individual properties to the overall sensing response. To understand the impact of pore size on chemiresistive sensing, Yao et al. synthesized a series of copper-based MOFs, incorporating HHB and HHTP with Cu²⁺ metal centers (Fig. 3f–g). The authors found that the MOFs decreased in conductivity as pore diameter decreased (Cu₃(HHTP)₂> Cu₃(HHTP)(HHB)> Cu₃(THQ)₂) and that correspondingly, the least conductive MOF had the greatest % response in the chemiresistive detection of NH₃. However, the response/recovery time was negatively impacted by smaller pores, with Cu₃(HHTP)₂ responding and recovering faster than Cu₃(HHTP)(THQ) and Cu₃(THQ)₂. The authors also compared Cu₃(HHTP)₂ nanorods and nanosheets, and found that nanorods had a slower response time as well, suggesting the impact of the length and diameter of MOF pores on gas diffusion through the framework (Fig. 3h).²³ The elucidation of these types of structure–property relationships for existing MOFs and COFs is expected to promote the rational design of new chemiresistive sensors and the strategic development of new framework materials to target specific analytes.

Hydrogen Sulfide:

Hydrogen sulfide is a highly toxic gas with an OSHA PEL of 25 ppm over a 10 hour workday.⁹⁷ H₂S is also a gasotransmitter used as a physiological signaling molecule in functions including vasorelaxation and nerve function.^93,98 In 2016, our group reported the first use of Cu₃(HHTP)₂ & Ni₃(HHTP)₂ for the chemiresistive detection of H₂S with MOFs grown on shrinkable polymer films.²⁷ We subsequently demonstrated detection of H₂S with Ni₃(HHTP)₂ and Ni₃(HITP)₂ grown on cotton textile.²⁸ Ni₃(HITP)₂ and Ni₃(HHTP)₂ detected H₂S with LODs of 0.52 ppm and 0.23 ppm respectively (see Table 1), and demonstrated filtration capability in addition to sensing.²⁸ We recently reported a Cu₃(HHTP)₂ textile sensor that exhibited similar multifunctional properties. Cu₃(HHTP)₂ grown on cotton textile had uniform nanowire morphology (Fig. 3i–j) and detected H₂S from 1-80 ppm with a dosimetric (irreversible) response that was enhanced in the presence of humidity (Fig. 3k–l). DRIFTS, XPS, and EPR analysis of bulk MOF powder before and after H₂S exposure provided insights into the interactions between H₂S and the framework (Fig. 3m), showing irreversible changes corresponding to the generation of elemental sulfur and copper sulfide species on the surface of the MOF.⁸² Taken together, these spectroscopic investigations provide insight into the potential mechanisms of both the sensing response and the uptake and decontamination of H₂S. The Cu₃(HHTP)₂ e-textile also maintained its H₂S sensing performance in response to physical and chemical stresses, such as washing, heating, and abrading (Fig. 3n), demonstrating the stability of the MOF on the textile.⁸² Taken together, these reports showed that layered conductive MOFs have tremendous potential as multifunctional materials that can simultaneously sense and protect against hazardous agents, such as H₂S.

As with ammonia, metallophthalocyanine-based framework materials offer significant improvements to limits of detection for H₂S detection. NiPc-O₈-Cu and NiPc-O₈-Ni exhibited ppb levels of detection for H₂S, 19 ppb and 32 ppb for NiPc-O₈-Cu and NiPc-O₈-Ni, respectively.³¹ Ni-COF-DC-8 also had a highly sensitive response to H₂S, with an LOD of 204 ppb.³³ These sensors are approaching the levels at which the human nose detects H₂S, and surpass other chemiresistive H₂S sensors based on metal oxides or reduced graphene oxides. Spectroscopic studies of the intermolecular interactions between these materials and H₂S suggest the sensitivity is driven by open metal coordination sites on the metallophthalocyanine sites on the surface of the framework.

Sulfur Dioxide:

Sulfur dioxide (SO₂) is a highly toxic gas (OSHA PEL 5 ppm over a 10 hour workday)⁹⁷ and an environmental pollutant whose accumulation in the atmosphere leads to acid rain and other adverse environmental effects.⁹⁹ Non- conductive MOFs, COFs, and porous metal-organic cages have frequently been employed for SO₂ capture, often relying on uncoordinated metal sites or hydrogen bonding interactions to form adducts with SO₂.¹⁰⁰ Some of these non-conductive MOFs have been employed in capacitative sensing architectures for SO₂ detection,¹⁰¹ or combined with conductive materials, such as carbon nanotubes, to form chemiresistive sensors for SO₂.¹⁰² Ingle et al. have also employed Ni₃(HHTP)₂ in a ChemFET sensor to detect SO₂ at levels as low as 625 ppb.⁸⁵ Computational analysis suggests that HITP-based MOFs may also be effective sensing materials for SO₂,⁸⁶ but experimental demonstrations using these materials have not yet been reported, and remain a fruitful area for further research.

Nitric Oxide/NO_x:

Nitric oxide and nitrogen dioxide are toxic gases present in the air of urban environments, produced by a variety of industrial processes.¹⁰³ The OSHA PEL for NO is 25 ppm over a 10 hour workday; the OSHA PEL for NO₂ is a ceiling of 5 ppm.⁹⁷ Our research group reported the first use of layered conductive MOFs as sensors for NO in 2016. Triphenylene-based MOFs (Ni₃(HHTP)₂, Cu₃(HHTP)₂, and Ni₃(HITP)₂) demonstrated ppm limits of detection for NO when incorporated into chemiresistive sensors, both on shrinkable polymer films²⁷ and textile substrates.^28,82 MOF-on-textile sensors demonstrated the ability to detect NO and withstand real-world environments, maintaining their conductivity and sensing performance following stresses such as bending, twisting, abrasion, and washing.^28,82 Textiles functionalized with Cu₃(HHTP)₂ also showed the ability to detoxify NO, with XPS and IR spectroscopy indicating the formation of NO_x species on the textile.⁸² These reports demonstrate the multifunctional potential of layered, conductive framework materials to serve as active components in a variety of wearable electronic devices.⁹

Metallophthalocyanine-based framework materials offer significant improvements to limits of detection for NO. NiPc-based MOFs linked with nickel and copper both exhibited dosimetric, irreversible responses to ppb levels of NO (Fig. 4a–b). Both MOFs exhibited sub-ppb LODs for NO, 0.13 ppb and 0.06 ppb for NiPc-O₈-Cu and NiPc-O₈-Ni, respectively. Both the decrease in resistance and the decrease in EPR signal at g =1.988 (Fig. 4c) indicated that NO acted as an oxidizing gas for NiPc-O₈-Ni, lowering the concentration of ligand-centered radicals in the MOF, and increasing the concentration of C=O bonds (determined by XPS). NiPc-O₈-Cu also showed a decrease in resistance in response to NO, but the EPR signal at g = 2.021 increased post-NO exposure, suggesting that the EPR-silent Cu⁺ observed in XPS spectra of the pristine MOF was oxidized to Cu²⁺ by irreversible binding with NO (Fig. 4c).³¹ A NiPc-based COF, Ni-COF-DC-8, also exhibited a dosimetric response to NO and NO₂. The COF-based chemiresistor sensed NO and NO₂ over the wide ranges of 0.02 – 40 ppm (NO) and 2 – 40 ppm (NO₂) and exhibited ppb-level LODs, 5 ppb and 16 ppb for NO and NO₂, respectively. XPS and EPR spectra indicated that the surface adsorption and subsequent charge transfer associated with these gases led the to the strong chemiresistive response.³³

Figure 4.

(a-c) Chemiresistive NO sensing with layered conductive frameworks NiPc-O₈-Ni, NiPc-O₈-Cu, and Ni-COF-DC-8. (a) Chemiresistive sensing traces of framework materials exposed to 1 ppm NO for 30 minutes show irreversible or partially irreversible responses. (b) Maximum response of each framework after 30 min exposure to 1 ppm NO. (c) EPR spectra of pristine NiPc-O₈-Ni, NiPc-O₈-Cu, and Ni-COF-DC-8 (dotted lines) and framework materials following exposure to 1 ppm NO for 30 minutes (solid lines). (a-c) Adapted from Z. Meng, A. Aykanat, K. A. Mirica, “Welding Metallophthalocyanines into Bimetallic Molecular Meshes for Ultrasensitive, Low-Power Chemiresistive Detection of Gases,” J. Am. Chem. Soc., 2019, 141, 2046-2053 (Ref. 31) Copyright 2019 American Chemical Society and Z. Meng, R. M. Stolz, K. A. Mirica, “Two-Dimensional Chemiresistive Covalent Organic Framework with High Intrinsic Conductivity,” J. Am. Chem. Soc., 2019, 141, 11929-11937 (Ref. 33) Copyright 2019 American Chemical Society. Conductive MOF-based chemiresistive sensing of carbon monoxide and carbon dioxide. (e-g) Carbon monoxide sensing with bimetallic, metallophthalocyanine-based MOFs. (e) Chemiresistive sensing traces for CoPc-O₈-Cu detect CO from 10-80 ppm. (f) Chemiresistive sensing response of CoPc-O8-Cu (blue) and NiPc-O8-Cu (green) to 80 ppm CO in dry nitrogen, air, and humidified nitrogen. (g) In situ DRIFTS experiment exposing CoPc-O₈-Cu MOF to 1% CO. (e-g) Reproduced with permission from “Bimetallic Two-Dimensional Metal–Organic Frameworks for the Chemiresistive Detection of Carbon Monoxide” A. Aykanat, Z. Meng, R. M. Stoz, C. T. Morrell, K. A. Mirica, Angew. Chem. Int. Ed. Vol. 61, No. 6 (Ref. 74) Copyright 2021 John Wiley and Sons. (h-i) Chemiresistive sensing of CO₂ with a Cu₃(HIB)₂-based sensor. (h) Chemiresistive sensing traces of the MOF exposed to 400 −2500 ppm CO₂ at 0% - 80% relative humidity. (i) In situ DRIFTS of Cu₃(HIB)₂ exposed to 0 – 10000 ppm CO₂ under air at 50 % RH. Blue line indicates the free carrier absorbance (FCA). (h-i) Reproduced from I. Stassen, J. H. Dou, C. Hendon, M. Dinca, “Chemiresistive Sensing of Ambient CO₂ by an Autogenously Hydrated Cu₃(hexaiminobenzene)₂ Framework” ACS Cent. Sci., 2019, 5, 1425-1431 (Ref. 87) ACS Author Choice Open Access.

Several other layered, conductive COFs have demonstrated sensitivity to NO and NO₂ in chemiresistive sensing architectures. Yue et al. reported a NiPc-based COF, covalently linked with CoTAA linkages (see Fig. 2b–i), which responded to NO₂ from 1 – 40 ppm. The COF also responded to NO, but with a smaller magnitude response.³⁴ Another metallophthalocyanine-based COF, which combined Ni- and CuPc monomers demonstrated ppb level detection of NO₂, and could be irradiated with UV light to rapidly desorb bound NO₂ from the COF.⁸⁰ Liu et al. reported two metalloporphyrin-based COFs, with cobalt and copper metal centers generated through post-metallation of the metal-free COF (see Fig. 2b–ii for COF structure). The Co-TPCOF was a particularly sensitive NO₂ sensor, with an LOD of 6.8 ppb (see Table 1). DRIFTS spectroscopy confirmed the adsorption of NO₂ to the COF, and the generation of nitrite species on the framework surface. DRIFTS also showed the presence of peaks corresponding to Co-NO₂ compounds, confirming the cobalt metal center as the NO₂ binding site. The ultra-low LODs demonstrated by metal-lophthalocyanine and metalloporphyrin-based frameworks in these electrically transduced sensing architectures suggest their utility as gas sensors for a variety of environmental and health applications. These studies also demonstrate the power of spectroscopic techniques to elucidate the nature of the intermolecular interactions between analyte molecules and framework materials.

Carbon Monoxide:

Carbon monoxide is an odorless, toxic gas whose buildup poses significant health hazards to humans (OSHA PEL 50 ppm over a 10 hour workday),⁹⁷ particularly as a result of incomplete combustion of heating fuels in buildings. Existing, commercially-available CO sensors rely on metal oxide-based chemiresistors, which require high temperatures.¹⁰⁴ Layered conductive frameworks offer the advantage of a more portable, low-power sensing device. We recently reported the use of two copper-linked metallophthalocyanine MOFs, CoPc-O₈-Cu and NiPc-O₈-Cu, as chemiresistive sensors for CO,⁷⁴ inspired by previous work in our group showing the efficacy of these materials for electrochemical CO₂ reduction.⁷⁵ Both the NiPc and CoPc-based materials responded to CO, and the CoPc-O₈-Cu MOF showed particularly sensitive CO detection, with an LOD of 0.53 ppm (Fig. 4d, Table 1). The MOF maintained its CO sensing capability in the presence of air and humidity (Fig. 4e), and had a minimal response to CO₂, a potential interferent. DRIFTS studies combined with DFT calculations indicated that the copper nodes were the dominant CO binding site and that main role of the NiPc and CoPc metal centers is to tune the electronics of the Cu host site (Fig. 4f).⁷⁴ Another recent example of MOF-based CO sensing used Zn₃(HHTP)₂ in both chemiresistive and ChemFET sensing architectures.⁸⁶ In this demonstration, Zn₃(HHTP)₂-based chemiresistive sensors effectively differentiated CO from SO₂ and NH₃, and had an LOD for CO of 3.96 ppm.⁸⁶ The authors determined that the sensor maintained its performance in the presence of up to 40% relative humidity.

Carbon Dioxide:

Carbon dioxide, while not acutely toxic to humans, nonetheless poses dangers to humans as a primary driver of climate change.¹⁰⁵ CO₂ sensors can function as an indicator of human activity in buildings or monitor overall air quality,¹⁰⁶ and potentially measure the efficacy of CO₂-capture technologies.¹⁰⁷ In 2019, Dinca and coworkers successfully employed Cu₃(HIB)₂ in a chemiresistive sensor to detect CO₂ from 400-2500 ppm, a range corresponding to the lower limit of outdoor CO₂ concentrations to the upper limit typically found indoors (Fig. 4h). Cu₃(HIB)₂ was selected based on the ability of the protons in the bis-diimmine MOF linkages to form hydrogen bonds to CO₂, and the capacity of the –NH moieties to generate acid-base adducts— these material-analyte interactions were expected to generate enhanced sensitivity and selectivity in the sensing response for CO₂. DRIFTS and near-IR diffuse reflectance spectroscopy were used to probe these intermolecular interactions, and demonstrated that the sensing response is driven by the modulation of charge carrier concentration upon CO₂ binding to the MOF (Fig. 4i). The Cu₃(HIB)₂-functionalized sensor maintained its performance over a wide humidity range (0-80% RH) and displayed excellent long-term stability.⁸⁷

H₂O/Humidity:

In the development of chemiresistive sensors for reactive/hazardous gases, the presence of water is often tested as a potential interferent that will impact the sensor performance and require calibration. However, the response of layered conductive framework materials to humidity can also be leveraged to create chemiresistive devices that monitor human breath. In 2020, Lee and Jeon reported a facemask consisting of Ni₃(HHTP)₂ hydrothermally grown on electrospun polyacrylonitrile fibers. This multifunctional material demonstrated the ability to filter particulate matter (PM_2.5 and PM_2.5-10) from air, as well as chemiresistive sensing responses to humidity. When the Ni₃(HHTP)₂-PAN composite was incorporated into a facemask (Fig. 5a), the mask showed changes in resistance that corresponded to the breathing of the user as a result of humidity in exhaled air (Fig. 5b).⁸⁸ Demonstrations such as this show the potential utility of framework materials integrated into smart textile materials.⁹

Figure 5.

Conductive MOFs employed as humidity sensors. (a-b) Humidity sensing with a conductive MOF-based face mask. (a) Ni₃(HHTP)₂-polyacrylonitrile composite material incorporated into a humidity sensing facemask. (b) Chemiresistive sensing trace of the Ni₃(HHTP)₂ facemask showing response to the user’s breath. (a-b) adapted from H. Lee, S. Jeon, “Polyacrylonitrile Nanofiber Membranes Modified with Ni-Based Conductive Metal Organic Frameworks for Air Filtration and Respiration Monitoring,” ACS Appl. Nano Mater. 2020, 3 8192-8198 (Ref. ⁸⁸) Copyright 2020 American Chemical Society. (c-d) Humidity sensing with modified Ni₃(HIB)₂ MOF. (c) Proposed modified, ‘missing linker’ structure of Ni₃(HAB)₂ employed for humidity detection. (d) Calibration curve comparing the humidity sensing performance of the missing linker MOF with the unmodified, crystalline MOF. (c-d) reproduced from C. Liu, Y. Gu, C. Liu, S. Liu, X. Li, J. Ma, M. Ding, ACS Sens. 2021, 6, 429-438 (Ref. 24) Copyright 2021 American Chemical Society. (e-g) Chemiresistive humidity sensor based on oriented thin film of Cu₃(HIB)₂. (e) Schematic of the sensor. (f) Comparison of the sensing response of the thin film sensor with the sensing response of Cu₃(HIB)₂ powder. (g) HRTEM image of Cu₃(HIB)₂ thin film, showing oriented, hexagonal pores. (e-g) reproduced from S. Park, Z. Zhang, H. Qi, B. Liang, J. Mahmood, H.-J. Noh, M. Hambsch, M. Wang, M. Wang, K. H. Ly, Z. Wang, I. M. Weidinger, S. Zhou, J.-B. Baek, U. Kaiser, S. C. B. Mannsfeld, X. Feng, R. Dong, “In-Plane Oriented Two-Dimensional Conjugated Metal–Organic Framework Films for High-Performance Humidity Sensing,” ACS Materials Lett., 2022, 4, 1146-1153 (Ref. ⁸⁹) Copyright 2022 American Chemical Society.

Post-synthetic modifications can be used to alter the humidity sensing performance of framework materials. Wang et al. examined the impact of functionalizing CuPc-(NH)₈-Ni and NiPc-(NH)₈-Ni with various alkylsilanes. Octadecyltrimethoxysilane (OTMS) grafted onto the MOF made the MOF film hydrophobic, as shown by an increase in water contact angle from 55° to 138°. The hydrophobic MOF/OTMS film had a lower response to humidity (5% vs. 2% change), but a faster recovery time (<10s).³² Liu et al. aimproved the humidity sensing performance of Ni-HIB by modifying the framework structure. By oxidizing the hexaaminobenzene prior to MOF synthesis, an amorphous structure with defect sites was produced (Fig. 5c). The Material with the missing linker sites was much more responsive to humidity in a chemiresistive sensing architecture, and had a detection range of 20-90% relative humidity vs. 60-90% for the crystalline Ni-HIB analog (Fig. 5d). The authors attributed this enhancement in performance to hydroxylated defect sites and their ability to hydrogen bond with water molecules in the chemiresistive sensing experiments.²⁴ In an alternative approach, Park et al. created a chemiresistive sensor for humidity featuring an oriented thin film of Cu₃(HIB)₂ deposited on an SiO₂ substrate (Fig. 5e).⁸⁹ This sensor detected humidity from 200 - 1000 ppm with enhanced sensitivity and response times over the MOF in bulk powder form (Fig. 5f), which the authors attributed to the short and highly oriented nature of the pores (Fig. 5g) and the correspondingly rapid diffusion of gas into the material.⁸⁹

Volatile Organic Compounds (VOCs):

Volatile organic compounds (VOCs) are environmental hazards generated by automobiles, oil production, and other industrial processes.¹⁰⁸ In low concentrations, VOCs are also endogenously produced and found in human breath at ppb levels, and specific VOCs can serve as biomarkers for diseases such as cancer.¹⁰⁹ Dincă and coworkers first demonstrated the potential of conductive MOFs to detect VOCs in 2015. The researchers created an array of three MOFs, Cu₃(HHTP)₂, Cu₃(HITP)₂, and Ni₃(HITP)₂, which responded to a large number of airborne organic compounds upon exposure to 200 ppm of each VOC vapor. Principal component analysis showed that the array could differentiate between functional groups, with alcohols, ketones, aromatics, and aliphatics having distinct responses from one another. The distinctions between functional groups were driven by consistent directionality of responses for specific MOF-functional group pairings (i.e. decrease in resistance for Cu₃(HHTP)₂ with alcohols, etc.).¹⁸ This report demonstrated the power of MOF-based arrays, in which only three materials provided reliable distinction between functional groups.

New layered conductive framework materials have also demonstrated potential for VOC sensing. We reported a Bi(HHTP) coordination polymer that proved to be an effective sensing material for acetone, ethanol, methanol and isopropanol. Bi(HHTP) had unique responses to these analytes, with methanol inducing an increase in resistance, and ethanol and isopropanol inducing decreases in resistance. These responses are likely driven by differences in factors such as pKa, dielectric constant, and dipole moment among the four analytes. The limits of detection for acetone, ethanol, methanol and isopropanol were 41.2 ppm, 278 ppm, 50.2 ppm and 185 ppm respectively,²⁹ values similar to those seen with metal oxide and reduced graphene oxide-based sensors. Another triphenylene-based material, a mixed ligand Cu(HHTP)(HITP) MOF, exhibited selective chemiresistive detection of benzene with an LOD of 0.024 ppm. Increasing the level of HITP doping into the framework enhanced the selectivity of the MOF sensor for benzene over NH₃.³⁰ These new materials show the modular utility framework materials, in which the choice of specific ligands and metal nodes can tune the performance of the material for specific material-analyte interactions.

Liquid-Phase Analytes:

The development of electrically transduced chemical sensors for the detection of liquid-phase analytes is largely centered around the area of healthcare diagnostics, and typical analytes are those relevant to human health and biological signaling mechanisms. In particular, the development of flexible, miniaturizable sensors that provide real-time chemical information about human health have great potential in wearable or insertable electronics.^110,111 The detection of neurochemicals, other signaling molecules, and biologically relevant ions presents an opportunity for gaining insight into physiological processes and their implications for human disease.^112–114 Other analytes are health-adjacent, such as glucose, which can be detected in physiological contexts (i.e., blood) or in food.^115–117 Another area of analytes are potential water contaminants that are harmful to humans, such as heavy metals.^118,119 A summary of the reported examples of liquid-phase sensing with layered conductive frameworks can be found in Table 2.

Table 2:

Detection of Liquid Phase Analytes with Layered Conductive Framework Materials

Analyte	Framework Material(s)	Sensing Methoda	LOD	Detection Range	Additional Details	Ref.
Dopamine (DA)	Ni3(HHTP)2 rods @GCE	Voltammetric (DPV)	63 ± 11 nMb	63 nM – 200 μM	Detection in 0.1 M PBS (pH 7.4) Simultaneous DA and 5-HT detection DA detection in simulated urine	42
	Ni3(HHTP)2 nanosheets @GCE	Voltammetric (DPV)	9.9 ± 2.2 nMb	9.9 nM – 6 μM	Detection in 0.1 M PBS (pH 7.4) Detection in the presence of DOPAC and simulated CSF	72
	Ni3(HHTP)2@ molecular wire @carbon fiber microelectrode	Voltammetric (DPV)	1 nMb	1 – 400 nM	Detection in simulated CSF, Selectivity against DOPAC, AA, UA, 5-HT, in vivo detection in mouse brain	121
	Oriented Cu3(HHTP)2 thin film @glass substrate with Au electrodes	ChemFET (SGTFT, VDS 20 mV, VGS 40 mV)	Not reported	100 nM – 50 μM	Detection in 0.1 M CaCl2, selectivity against AA, UA, and glucose	61
Serotonin (5-HT)	Ni3(HHTP)2 rods @ GCE	Voltammetric (DPV)	40 ± 17 nMb	40 nM – 200 μM	Detection in 0.1 M PBS (pH 7.4) Simultaneous DA and 5-HT detection DA detection in simulated urine	42
Gluconic acid	Ni3(HITP)2@SiO2 substrate with Au electrodes	ChemFET (liquid gated, Vds −0.1 V)	Not reported	10−6 – 10−3 g/mL	Detection in PBS	60
Glucose	GOD- GA@Ni/Cu(HITP)@Si O2 substrate with Au electrodes	ChemFET (liquid gated, Vds −0.1 V, Vgs −0.1V)	0.51 μMc	1 μM – 20 mM	Detection in PBS (pH 7.4), No sensor response to sucrose, fructose, galactose, lactose, UA, AA	59
	Ni3(HHTP)2 (no electrode support material listed)	Amperometric (CPA, 0.55 V vs. Ag/AgCl)	0.66 μMc	1 μM – 0.5 mM	Detection in 0.1 M NaOH, no sensor response to 1 mM glucose, 2 mM AA, urea, Cl−, L-cysteine, lactose, fructose, UA, DA	122
	Ni3(HITP)2 nanosheets @carbon cloth	Amperometric (CPA, −0.6 V vs. Ag/AgCl)	0.57 μMc	1 μM – 7 mM	Detection in 0.1 M NaOH, no sensor response to 1 mM glucose, cytosine, fructose, UA, AA, urea, Cl−, SO42−	123
	Ni3(HITP)2@GCE	Voltammetric (CV)	Not reported	1 – 10 mM	Detection in 0.1 M KOH, response unaffected by 1 mM AA, UA	124
	Ni/Co(HHTP) @carbon cloth	Amperometric (CPA, 0.5 V vs. Ag/AgCl)	100 nMc	0.3 μM – 2.3 mM	Detection in 0.1 M NaOH, no sensor response to urea, NaCl, fructose, lactose, UA, l-cysteine, Detection in spiked human serum, tea, and orange juice	43
	Ni3(HHTP)2@SPCE	Voltammetric (CV)	1.30 μMc	25 – 300 μM	Sensing in 0.1 M NaOH, dual detection of H2O2, detection in spiked human serum	125
	Co/Cu(HHTP) @carbon cloth	Amperometric (CPA, 0.58 V vs. Ag/AgCl)	0.27 μMb	0.27 μM – 2.4 mM	Sensing in 0.1 M NaOH, no sensor response to lactose, fructose, l-cysteine, urea, AA, DA, UA, NaCl	126
Ascorbic acid	Ni3(HITP)2@SPCE	Amperometric (CPA, 0.5 V)	0.1 μMc	0.5 – 100 μM	Sensing in 0.1 M NaOH, selectivity against 1 μM Urea, glucose, NaCl, UA, DA, detection of sweat with portable device	127
Paracetamol	Ni/Cu(HHTP)/Nafion @GCE	Voltammetric (DPV)	5 μMb	5 – 190 μM	sensing in PBS buffer (pH 6.5), selectivity against K+, Cd2+, Cu2+, Pb2+, Fe3+, Al3+, SO42−, Cl−	44
Nifedipine	CoPc-O8-Cu@GCE	Voltammetric (DPV)	6 nMc	10 nM – 93 μM	Sensing in PBS buffer (pH 7.0) No response to 500x SO42−, Fe3+, 300x glucose, sucrose, 200x K+, 100x Ca2+, Na+, Cl−, 50x starch, dextrin, glycine, L-cysteine, Zn2+, Cu2+, NH3+, CO32−, NO3−, 10x AA	128
Malachite Green (MG)	Cu3(HHTP)2@CPE	Voltammetric (DPV)	1.34 nMc	5 nM – 1 μM	Sensing in 0.1 M PBS, (pH 7.0) Sensor response unaffected by 0.5 μM K+, Na+, Mg2+, Ca2+, Zn2+ and 1 μM glucose, serine, L-cysteine, UA, XA, CAP, TET, OTC, ERY, detection of MG in fish	129
Paroxon	AChE@Cu3(HHB)2@GCE	Amperometric (CPA,	0.37 ng/mLc	1 -1000 ng/mL	Sensing based on inhibition of AChE conversion of ATCh, sensing in PBS (50 mM, pH 7.4), and real water and soil samples	130
Bisphenol A (BPA)	Cu3(HHB)(HHTP) @rGO@GCE	Voltammetric (DPV)	52 nMb	50 nM - 100 μM	sensing in PBS buffer (pH 7)	131
	Cu3(HHTP)2 @rGO@GCE	Voltammetric	49 nMb	50 nM - 100 μM	selective against 10 mM K2CO3, MgCl2, NaNO3, (NH4)2SO4	69
H2O2	Cu3(BHT)2 thin film@SPE	Amperometric (CPA, −0.75 V vs. Ag/AgCl)	16.5 nMc	100 nM – 50 mM	Selectivity against NaCl, KCl, methanol, ethanol, glucose, sucrose, AA, DA, O2	132
	Cu3(HHTP)2 nanosheets/Au-NPs	Amperometric (CPA, −0.6 V)	5.6 nMc	50 nM – 16.4 mM	Sensing in 0.1 M PBS (pH 7.4) detection of H2O2 release from live cells	133
	Ni3(HHTP)2@SPCE	Amperometric (CPA, 0.4 V)	2.13 μMc	50 μM – 5 mM	Sensing in 0.1 M NaOH, dual detection of glucose, detection in spiked human serum	125
K+	K+-ISM-I @Ni3(HHTP)2@GCE	Potentiometric	5.01 ± 0.01 x 10−7 Mb	5.01 x 10−7 – 10−1.5 M	Selectivity against Ca2+: −5.46 ± 0.09, Na2+: −4.10 ± 0.01, NH4+: −2.12 ± 0.05	55
NO3−	NO3−-ISM @Ni3(HHTP)2@GCE	Potentiometric	6.31 ± 0.01 x 10−7 Mb	6.31 x 10−7 – 10−1.5 M	Selectivity against SO42−: −3.15 ± 0.01, Cl−: −1.82 ± 0.01, Br−: −2.12 ± 0.05	55
Li2+	ISM@Ni3(NIB)2@GCE	Potentiometric	0.99 μMb	1 x 10−6 – 1 M	Selectivity against Na+: −2.21, K+: −2.63, Mg2+: −3.43	56
Na+	Nafion/Oriented Ni3(HHTP)2 rods @CNTF	Potentiometric	Not reported	1 x 10−5 – 1 x 10−1 M	Incorporated into a wearable sweat-sensing device	134
Cu2+	Cu3(HHTP)2@GCE	Potentiometric	10−6.5 Mb	10−7 M – 10−2 M	ISM@Cu3(HHTP)2 had lower LOD, slower response time	54
Hg2+	Ni3(HITP)2 nanosheets with grafted GA-DNA @SiO2 substrate with Ag electrodes	ChemFET (liquid gated, VDS −0.1 V, VGS −0.1 V)	10 pMb	10 pM – 100 nM	Detection in PBS Selective against 10 mM Cu2+, Cd2+, Pb2+, Zn2+, Mg2+, Co2+	62
NO2−	Nafion@NiPc-(NH2)8-Ni@GCE	Voltammetric (DPV)	2.3 μMc	10 μM -11.5 M	Detection in 1 M PBS buffer (pH 7.0) Selective against DA, AA, UA, glucose, NO3−	45

^aListed technique indicates the method used to calculate detection range and LOD.

^bExperimental limit of detection, based on lowest measured concentration where S/N > 3.

^cTheoretical limit of detection, based on calculated concentration where S/N = 3.

Abbreviations: LOD = limit of detection, GCE = glassy carbon electrode, DPV = differential pulse voltammetry, PBS = phosphate buffered saline, DOPAC = 3,4-dihydroxyphenylacetic acid, AA = ascorbic acid, UA = uric acid, CSF = cerebrospinal fluid, SGTFT = solution-gated thin film transistor, GOD = glucose oxidase, GA = glutaraldehyde, CV = cyclic voltammetry, CPE = carbon paste electrode, XA = xanthine, HXA = hypoxanthine, CAP = chloramphenicol, TET = tetracycline, OTC = oxytetracycline, ERY = erythromycin, AChE = acetylcholinesterase, ATCh = acetylthiocholine, rGO = reduced graphene oxide, SPE = screen-printed electrode, CPA = constant potential amperometry, SPCE = screen printed carbon electrode, ISM = ion selective membrane, CNTF = carbon nanotube fiber.

Neurochemicals:

The class of neurotransmitters that includes dopamine (DA) and serotonin (5-HT) are known as catecholamines. These redox-active compounds participate in neurochemical signaling and can be oxidized through proton-coupled electron transfer (PCET) processes (Fig. 6a).¹²⁰ The oxidation of these neurochemicals can be detected in voltammetric sensing architectures. In 2020, we reported the first use of conductive MOFs as working electrode materials for voltammetric detection of dopamine and serotonin with Cu₃(HHTP)₂, Ni₃(HHTP)₂, Cu₃(HITP)₂ and Ni₃(HITP)₂. Ni₃(HHTP)₂ was the best candidate for electrochemical detection neurochemicals both because of its semi-reversible redox kinetics for electrochemical probes and a lack of intrinsic redox activity in the potential window relevant for DA and 5-HT detection. We found Ni₃(HHTP)₂ to have a limit of detection of 63 ± 11 nM and 40 ± 17 nM for dopamine and serotonin respectively (see Table 2). This material also showed the ability to differentiate dopamine and serotonin in the same solution.⁴²

Figure 6.

Voltammetric detection of dopamine with MOF-based electrodes. (a) Dopamine is oxidized through PCET. (b-d) Detection of dopamine with oriented Ni₃(HHTP)₂ films. The edge facet of Ni₃(HHTP)₃ (c) had a lower response to dopamine than the basal plane (d) in CV studies. (b-d) Adapted from R. M. Stolz, A. F. Kolln, B. C. Rocha, A. Brinks, A. M. Eagleton, L. Mendecki, H. Vashisth, K. A. Mirica, “Epitaxial Self-Assembly of Interfaces of 2D Metal–Organic Frameworks for Electroanalytical Detection of Neurotransmitters” ACS Nano, 2022, 16, 13869-13883 (Ref. 72) Creative Commons CC-BY-ND-NC 4.0. (e-g) In vivo detection of dopamine in mouse model with Ni₃(HHTP)₂ based microelectrode. (e) Microelectrode design featuring Ni₃(HHTP)₂ connected to a gold surface with a molecular wire. (f) The microelectrodes were implanted in three locations in mouse brains to monitor dopamine levels in a highly localized fashion. (g) Changes in dopamine concentrations in different brain regions over the course of seven days post-injection of a mitochondrial toxin that induces acute Parkinson’s disease in the mouse. Adapted from Y. Wang, Y. Qian, L. Zhang, Z. Zhang, S. Chen, J. Liu, X. He, Y. Tian, “Conductive Metal–Organic Framework Microelectrodes Regulated by Conjugated Molecular Wires for Monitoring of Dopamine in the Mouse Brain” J. Am. Chem. Soc., 2023, 145, 2118-2126 (Ref. 121) Copyright 2023 American Chemical Society.

In subsequent work, we investigated the specific electrochemically active surfaces of the MOF that are responsible for electrochemical detection. By growing Co₃(HHTP)₂ and Ni₃(HHTP)₂ in controlled morphologies that favored exposure of either the basal plane (denoted {001}) or edge sites (denoted {100}) to the analytes in solution, we found that the basal planes of these materials were far more sensitive to dopamine than the edge sites (Fig. 6b–d). By contrast, the basal facets of the Co₃(HHTP)₂ and Ni₃(HHTP)₂ showed little to no enhancement for other neurochemicals such as 3,4-dihydroxyphenylacetic acid (DOPAC). As such, the exposed basal plane sites of Ni₃(HHTP)₂ showed nanomolar limits of detection for dopamine, LOD_DA =48 nM and 381 nM in the presence of 5 μM and 50 μM DOPAC, respectively (See Table 2).⁷² Langmuir isotherm studies showed that adsorption of dopamine to the surface of Ni₃(HHTP)₂ was more favorable (more negative Gibbs free energy of adsorption) for basal plan than for the edge facets, indicating that the enhanced oxidative current for dopamine was driven by the ability of dopamine to adsorb to the MOF {001} surface.⁷² This approach clearly shows the potential selectivity offered by preferential orientation of surfaces within an atomically precise nanomaterial, such as a MOF, and suggests that other highly-selective MOF-based sensors for a variety of organic analytes can be fabricated through careful control of MOF morphology and epitaxial orientation within devices.⁵⁶

Subsequent reports have explored alternative sensing configurations with triphenylene-based MOFs for dopamine detection. Song et al. fabricated a solution-gated transistor device with Cu₃(HHTP)₂ through a layer-by-layer deposition process.⁶¹ The Cu₃(HHTP)₂ films grown through this method did not have the nanowire morphologies typically observed in hydrothermal synthesis, but had smaller, lower-aspect ratio nanocrystals. The film thickness was controlled through the number of layering cycles, and grazing incidence x-ray diffraction analysis suggested that the film contained preferentially oriented nanocrystals.⁶¹ The chemFET sensor fabricated through this method showed an LOD for dopamine of 100 nM, and demonstrated selectivity against ascorbic acid, uric acid, and glucose (See Table 2).⁶¹ The improvements to LOD and selectivity in this report are likely a result of the thin-film transistor architecture, and the highly-oriented nature of the MOF material.

Another recent report advanced MOF-based detection of neurochemicals from highly controlled, in vitro laboratory studies, to real-time in vivo detection in animal models. Wang et al. fabricated a carbon fiber microelectrode functionalized with Ni₃(HHTP)₂ and demonstrated the use of this electrode in mouse brains to study Parkinson’s disease. The microelectrode was created by immersing a gold-coated carbon fiber microelectrode first in a solution with a rigid, conjugated small molecule, a ‘molecular wire’, and then in a dispersion of Ni₃(HHTP)₂ and water (Fig. 6e).¹²¹ Wang et al. evaluated the analytical performance of the microelectrodes, including the sensitivity to dopamine, and the selectivity against other biologically relevant analytes such as epinephrine, norepinephrine, ascorbic acid and uric acid, as well as the electrode stability and biofouling in bovine serum albumen. They found a detection limit of 1 nM DA in an artificial cerebrospinal fluid solution, and found that the signal decreased by less than 2% over the course of 100 cycles.¹²¹ After the in vitro evaluation of the electrodes, they were surgically implanted into three brain regions of laboratory mice (Fig. 6f). Using the implanted microelectrodes, the researchers tracked concentrations of dopamine in the mouses’ hippocampuses, cortices, and striata following an injection of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), a mitochondrial toxin that induces acute Parkinson’s disease. The microelectrode measurements showed a rapid decrease in DA concentration in the striatum following injection, and less dramatic decreases in the cortex and hippocampus (Fig. 6g).¹²¹ These results indicate the potential for incorporating MOF-based sensors into implantable devices, and show the ability of MOF materials to offer sensitive and selective detection in complex environments.

Glucose:

The detection of glucose is a vital part of diabetes management, both in point-of-care diagnostics to measure glucose in blood or interstitial fluid, and in measurements of food and beverages.¹³⁵ Enzyme-based glucose sensors are widely available, but suffer from limited lifetime and stability.¹³⁶ MOF materials have demonstrated potential to serve as non-enzymatic electrochemical glucose sensors. Duan and coworkers developed a chemFET sensor based on Ni₃(HITP)₂. The device was fabricated by growing a Ni₃(HITP)₂ hydrothermally on a Si wafer substrate pre-fabricated with a channel and gold source and drain electrodes. SEM images showed a uniform coverage of Ni₃(HITP)₂ nanosheet structures between the Au electrodes (Fig. 7a). The MOF film was approximately 500 nm thick after 60 minutes of growth (Fig. 7b).⁶⁰ This chemFET device was used to detect gluconic acid, a biproduct of enzymatic glucose oxidation. At a constant gate voltage of −0.1 V and increasing concentrations of glucose (from 10⁻⁶ to 10⁻³ g/mL) the gate-source current increased, and the gatedrain current was decreased (Fig. 7c).⁶⁰ These changes were attributed to the fact that the dominant charge carriers in the Ni₃(HITP)₂ film were holes, and the addition of gluconic acid to the solution reduced the number of holes.⁶⁰ This work demonstrated the potential of Ni₃(HITP)₂ to be incorporated into chemFET sensors for the non-enzymatic detection of glucose.

Figure 7.

Glucose detection with triphenylene-based MOFs. (a-c) A Ni₃(HITP)₂-based ChemFET device for gluconic acid detection. (a) SEM image of a chemFET device incorporating Ni₃(HITP)₂ MOF channel connecting Au electrodes. (b) SEM image of the Ni₃(HITP)₂ film on the Si substrate with 516 nm thickness. (c) I_ds vs. V_gs traces at increasing concentrations of gluconic acid. (a-c) Reproduced from B. Wang, Y. Luo, B. Liu, G. Duan, “Field-Effect Transistor Based on an in Situ Grown Metal–Organic Framework Film as a Liquid-Gated Sensing Device,” ACS Appl. Mater. Interfaces, 2019, 11, 35935-35940 (Ref. 60) Copyright 2019 American Chemical Society. (d-e) SEM images of Ni/Co-HHTP MOF grown on carbon cloth. (f) Detection of glucose with Ni/Co-HHTP@carbon cloth electrodes using CV. (d-f) Reproduced from Food Chemistry, Vol. 349 Z. Xu, Q. Wang, H. Zhangsun, S. Zhao, Y. Zhao, L. Wang, “Carbon cloth-supported nanorod-like conductive Ni/Co bimetal MOF: A stable and high-performance enzyme-free electrochemical sensor for determination of glucose in serum and beverage,” p. 129202 (Ref. 43) Copyright 2021 with permission from Elsevier.

The same authors subsequently reported a chemFET device for glucose detection, fabricated through similar methods, but incorporating a Ni/Cu-HITP bi-metallic MOF which was synthesized by combining NiCl₂ and CuSO₄ with the linker in a hydrothermal synthesis.⁵⁹ Several ratios of nickel to copper were tested, and 7:1 Ni:Cu demonstrated the most enhancement to the film conductivity.⁵⁹ XPS confirmed the presence of both nickel and copper nodes in the MOF, and SEM showed a uniform film of nanoflakes, similar in morphology to the previously reported Ni₃(HITP)₂ film.⁶⁰ The MOF film was functionalized with glutaraldehyde (GA) and glucose oxidase (GOD) to enhance the glucose sensing performance of the device. The GOD-GA-Ni/Cu-MOFs-FET device had a detection limit of 0.51 μM for glucose, and had a linear range up to 20 mM.⁵⁹ The device did not respond to uric acid, ascorbic acid, galactose, or lactose, and maintained its performance in the presence of these interferents.⁵⁹ Despite good initial performance, the sensor had limited shelf-life, loosing current signal over the course of one week. The authors attributed this limitation to the inactivation of GOD on the MOF surface, but highlight the potential for this sensor to serve as a disposable, single-use device.⁵⁹

Another recently reported glucose sensor also featured a bimetallic triphenylene MOF. Xu et al. grew a Ni/Co-HHTP MOF on carbon cloth and employed the MOF@textile for voltammetric detection of glucose. The Co-Ni MOF was grown hydrothermally on the cloth, using a 2:1 molar ratio of nickel to cobalt precursors (Fig. 7d–f).⁴³ Using amperometry (at 0.5 V working potential) the MOF@carbon cloth sensor had an LOD of 100 nM for glucose in 0.1 M NaOH. The sensor also demonstrated selectivity against a variety of interferents, including urea, lactose, fructose, and uric acid.⁴³ The authors also tested their MOF@carbon cloth sensor in simulated human serum, and beverage samples of tea and orange juice and found that the sensor maintained its performance in these liquids.⁴³ Several other recent reports have demonstrated glucose sensing with layered conductive MOFs, primarily with Ni-based materials (see Table 2).^122–125

Other Biologically Relevant Organic Analytes:

Select examples exist employing MOFs for voltammetric detection of other organic analytes of biological relevance (see Table 2).^{127–130,137} In 2020, Wang et al. reported using a mixed metal Ni/Cu-HHTP MOF for the detection of paracetamol (a.k.a. acetaminophen), a redox active analyte.^44,138,139 The MOF was synthesized hydrothermally using a 3:2 ratio of Ni:Cu precursors, and EDX confirmed the presence of both nickel and copper in the prepared MOF powder.⁴⁴ The MOF, dropcast on glassy carbon electrodes, detected paracetamol from 5 −190 μM in DPV experiments. The sensing performance was unaffected by the presence of inorganic ions (including K⁺, Cl⁻, and SO₄²⁻), and the electrode was able to distinguish paracetamol from dopamine and ascorbic acid.⁴⁴

Copper-based conductive MOFs have been used in the detection of bisphenol A (BPA), an endocrine disruptor commonly introduced to water supplies from plastics manufacturing. Chen and coworkers reported two different copper-based MOFs as electrochemical sensors for BPA. Cu₃(HHTP)₂, attached to glassy carbon electrodes with reduced graphene oxide (rGO) had an LOD of 4.9 nM for BPA.⁶⁹ A mixed ligand Cu-HHB/HHTP MOF, anchored to glassy carbon with a similar approach, had an LOD of 5.2 nM for BPA (see Table 2).¹³¹ Both sensors also had selectivity against a variety of interfering ions, and detected BPA in real water samples.^69,131

Ions:

The detection of ions has a number of use cases, including drinking water quality monitoring and the detection of biomarkers in human sweat or serum.¹⁴⁰ Potentiometric detection of ions generally relies on selective host-guest interactions between the target ion and a membrane or solid electrode material.⁵⁰ Ion selective membranes (ISMs) have been developed for a wide variety of ions, including metal cations, and anions such as nitrate and chloride.⁵² Conductive framework materials can function as a solid support for these membranes, serving as an ion-to-electron transducer. Our group demonstrated this approach to create sensors for K⁺ and NO₃⁻ featuring two potassium-selective ISMs and a nitrate selective ISM layered on top of Ni₃(HHTP)₂ electrodes (Fig. 8a). These ISEs demonstrated near-Nernstian responses and had mM-nM working ranges (Fig. 8b–c, Table 2). Conductive MOFs demonstrated a number of important advantages for use in this kind of ion sensor, including low contact resistance between the MOF and ISM, and compatibility with detection of both anions and cations.⁵⁵ Abdollahzahdeh et al. applied this same approach to the detection of lithium ions, using Ni-HAB as the conductive MOF substrate and potassium tetrakis(4-chlorophenyl) borate as the Li⁺-selective ionophore. Their sensor demonstrated near-Nernstian response (57.6 mV/decade) and potential stability over a period of 12 hours, as well as the ability to detect Li⁺ in the presence of Na⁺, K⁺ and Mg²⁺.⁵⁶ Another recent report synthesized oriented Ni₃(HHTP)₂ rods grown on carbon nanotube fibers (CNTF) as the ion-to-electron transducer for sodium detection. This Ni₃(HHTP)₂@GNTF composite was coated with Nafion and incorporated into a wearable device for real-time sweat detection.¹³⁴

Figure 8:

Conductive-MOF-based sensors for ions employing potentiometric and ChemFET-based sensing architectures. (a) HHTP based MOFs used as the signal transducing material underneath non-conductive ion-selective membranes in a potentiometric sensing architecture. (b) A nitrate-sensing electrode with Ni₃(HHTP)₂ coated with either valinomycin (K⁺-ISM-I) or NaTFPB (K⁺-ISM-II) showed a near-Nernstian slopes and had an LODs of 6.31 x 10⁻⁷ M. (c) Potassium-sensing electrodes with Ni₃(HHTP)₂ coated with NO₃⁻-ISM showed a near-Nernstian slope and had LODs of 5.01 x 10⁻⁷ M and 6.76 x 10⁻⁶ M, respectively. (a-c) Reproduced from L. Mendecki, K. A. Mirica, “Conductive Metal–Organic Frameworks as Ion-to-Electron Transducers in Potentiometric Sensors” ACS Appl. Mater. Interfaces, 2018, 10, 19248-19257 (Ref. 55) Copyright 2018 American Chemical Society. (d) A ChemFET sensor used Ni₃(HHTP)₂ as the channel material which was functionalized with DNA for the detection mercury ions. (e) Amperometric sensing of Hg²⁺ ions from picomolar to nanomolar concentrations with the DNA functionalized, MOF-based ChemFET sensor. (f) Amperometric sensing trace demonstrating selectivity of sensor against other divalent ions. (d-f) Reproduced from S. Shen, P. Tan, Y. Tang, G. Duan, Y. Luo, ACS Appl. Electron. Mater., 2022, 4, 622-630 (Ref. 62) Copyright 2022 American Chemical Society.

This approach of using a conductive MOF as a conductive substrate under non-conductive ion selective materials has also been applied to ChemFET sensing architectures. Shen et al. used Ni₃(HITP)₂ as the channel material in a ChemFET sensor for Hg²⁺. The MOF film was functionalized with a DNA probe that served as a selective binder for Hg²⁺ ions (Fig. 8d). The DNA probes were covalently attached to the MOF via a glutaraldehyde (GA) crosslinker, which interacted with the amino groups on the Ni₃(HITP)₂ surface, and the amino groups at the 5’ ends of the DNA strands. The sequence of the DNA probe was rich in thymines, the amino group of which binds to Hg²⁺. The Ni₃(HITP)₂-GA-DNA sensor responded to Hg²⁺ concentrations as low as 10 pM and exhibited minimal responses to other 2+ ions (Fig. 8e–f).⁶²

Conductive MOFs can also be used directly as an ion sensing material. In a recent report, Xu et al. used Cu₃(HHTP)₂ for the detection of Cu²⁺ ions in a potentiometric sensing architecture. They compared the bare Cu₃(HHTP)₂ sensor with a Cu-selective ISM on top of Cu₃(HHTP)₂, and found that the bare Cu₃(HHTP)₂ had a higher response, as well as greater stability. The ISM-free sensor was far more resistant to damage by organic solvent than the ISM@Cu₃(HHTP)₂ sensor. However, the bare Cu₃(HHTP)₂ had limited selectivity and was impacted by the presence of other ions such as K⁺ and Li⁺, which the authors attributed to ion exchange of Cu²⁺ out of the framework.⁵⁴

Nitrite is an anion generated in physiological systems as a biproduct of nitric oxide, generated environmentally as an intermediate in the nitrogen cycle, and also a common water and soil contaminant. Because nitrite (NO₂⁻) is a redox active ion that can be oxidized to nitrate (NO₃⁻), it can be detected voltammetrically. Lu et al. reported the voltammetric detection of nitrite using a nickel-phthalocyanine-based MOF. A dispersion of NiPc-O₈-Ni nanosheets blended with 5 % Nafion was dropcast onto glassy carbon electrodes, and used for the voltammetric detection of nitrite. In DPV studies, the MOF electrode detected nitrite at concentrations ranging as low as 10 μM.⁴⁵ This study constitutes the first reported use of phthalocyanine-based metal–organic frameworks as a voltammetric sensing material.

CONCLUSIONS AND OUTLOOK

The use of layered, conductive framework materials as electrochemical sensors has blossomed over the course of the past decade. The research efforts in this field have resulted in a solid body of knowledge, and critical progress on two fronts. The first is in the area of applied sensing, including integration of frameworks into devices and sensing demonstrations with a wide variety of analytes of interest. New materials have pushed the limits of detection for target analytes lower, in some cases surpassing the performance metrics of other state-of-the-art sensing materials.³³ The second is the fundamental understanding of chemical interactions between analytes and framework materials that lead to sensing responses. The combination of in situ spectroscopic techniques, computational analysis, and the comparison between modular framework components has allowed for in-depth, molecular scale understanding of the interface between framework materials and the chemical environment.^74,87,94,95 This understanding is vital to the development of chemical sensors, and helps realize a key advantage of frameworks over materials for sensing applications: bottom-up self-assembly allows for precise, tunable control over these interfaces and thus the interactions that govern sensing.

We see several promising areas for future work in the field of frameworks as electronic chemical sensors, which will greatly advance the field. One such direction involves deepening understanding of framework material structure–property relationships, e.g. how morphology affects electrochemical performance, and the structural changes that accompany electrochemical sensing responses. A number of factors, including the role of specific crystallographic interfaces, the role of atomic and molecular defects, and the junctions between crystallites of the material, impact host–guest interactions and the sensing performance of the material. Understanding of the mechanisms of charge transport and the electronic structure of framework materials can also be leveraged to understand and improve the sensing response. New mechanistic and structural insights will come from the combination of a variety of spectroscopic techniques that probe chemical interactions between host frameworks and guest molecules. In addition, developing methods for in situ and operando spectroscopic monitoring of materials during sensing processes will lead to key break-throughs in understanding the mechanisms of chemical sensing. Equally important will be the insights gained from computational modeling of framework–analyte interactions, which can help explain the differences in sensing response observed with different analogs of framework materials. The molecular understanding of sensing interactions gained will drive the strategic design and development of new framework materials for more sensing applications.

The development of new framework materials is another vital future direction for the field. As demonstrated in this review, the incorporation of new metal nodes,²⁹ as well new linker motifs and linkage types^31,34 have the potential to offer significant enhancements in sensing performance,³³ and new ability to tune the properties of frameworks.²³ The incorporation of motifs such as metalloporphyrins and metallophthalocyanines into layered conductive MOFs and COFs blends the advantageous properties of framework materials with biologically-inspired molecular recognition sites.¹⁴¹ Continuing to draw inspiration from nature’s carefully honed chemical sensing mechanisms will lead to new breakthroughs in sensitivity and selectivity in framework-based chemical sensors. Additionally, the continued development of conductive COFs will lead to the incorporation of these molecular recognition sites into materials that have superior thermal and chemical stability, in addition to high surface area and conductivity. New layered conductive COF materials, as well as continued understanding and control over COF crystallization and self-assembly will allow for the development of highly robust chemical sensing materials. An alternative approach for the creation of new materials is the combination of layered-conductive framework materials with metal oxides,¹⁴² metal nanoparticles,^133,143 and non-conductive framework materials,¹⁴⁴ which serve to enhance sensitivity or offer additional chemical functionality for detecting a wider variety of analytes.

Another future direction is the continued development of methods for integrating framework materials onto flexible substrates and into miniaturized devices. Thorough understanding of the process of framework self-assembly is required to successfully adapt material synthesis from bulk powder onto porous and/or flexible substrates^25,28,82 and into thin films.^81,89 Interactions between components of the framework and the chemical functionality of the substrate has a large impact on successful integration, as does the identity of the metal and the growth kinetics of the framework material. These processes must be thoroughly understood in order to generalize textile integration/self-assembly methods across multiple materials.

An additional future direction is the expansion of framework-based chemical sensing arrays.¹⁴⁵ Arrays can leverage the modularity of framework materials to generate variable responses that allows for highly selective differentiation between analytes. Previous demonstrations have shown that relatively few framework materials are capable of differentiating between many analytes.^18,28,31,74 Expanding the use of framework-based arrays will include development both of materials but also computational methods to transform complicated, multi-dimensional data sets into useable information about analytes. In tandem with this, a high impact direction will be moving arrays from gas phase detection to liquid phase detection, which has not yet been demonstrated.

Taken together, conductive MOFs and COF have demonstrated promising capabilities for electronic chemical detection of a number of analytes. The future of the field lies in moving sensing demonstrations away from highly controlled, laboratory scale sensing to the detection of analytes in complex mixtures and real-world environments. Some preliminary demonstrations in this area have been reported,^{88,121,127,134} but most studies are still very fundamental. The future directions outlined above will help realize this goal— fundamental studies and new materials will help to maximize the intrinsic sensitivity and selectivity of framework-based sensors, and sensor arrays will further enhance the ability of these sensors to differentiate between highly similar analytes in complex environments. Integrating frameworks onto flexible/textile substrates will allow for the creation of wearable and insertable devices that can provide human beings with new and transformative information about the chemical environment.

Biographies

Emma K. Ambrogi is currently a postdoctoral researcher in the research group of Katherine Mirica at Dartmouth College (Hanover, NH, USA). She obtained her B.A. in Chemistry from Wellesley College (Wellesley, MA, USA) in 2016, and her Ph.D. in Chemistry from Dartmouth College in 2024. Her research is focused on the use conductive MOFs for electrochemical detection of biologically relevant analytes, and on understanding MOF self-assembly and MOF–substrate interactions.

Katherine A. Mirica is an Associate Professor of Chemistry at Dartmouth College. She completed her undergraduate studies at Boston College, followed by graduate studies at Harvard University with George M. Whitesides, and postdoctoral training at the Massachusetts Institute of Technology with Timothy M. Swager. Her research interests include molecular engineering of multifunctional materials, crystal engineering, and self-assembly.