Reversible MOF-Based Sensors for the Electrical Detection of Iodine Gas

Iodine detection is crucial for nuclear waste clean-up and first responder activities. For ease of use and durability of response, robust active materials that enable the direct electrical detection of I2 are needed. Herein, a large reversible electrical response is demonstrated as I2 is controllably and repeatedly adsorbed and desorbed from a series of metal–organic frameworks (MOFs) MFM-300(X), each possessing a different metal center (X = Al, Fe, In, or Sc) bridged by biphenyl-3,3′,5,5′-tetracarboxylate linkers. Impedance spectroscopy is used to evaluate how the different metal centers influence the electrical response upon cycling of I2 gas, ranging from 10× to 106× decrease in resistance upon I2 adsorption in air. This large variation in electrical response is attributed not only to the differing structural characteristics of the MOFs but also to the differing MOF morphologies and how this influences the degree of reversibility of I2 adsorption. Interestingly, MFM-300(Al) and MFM-300(In) displayed the largest changes in resistance (up to 106×) yet lost much of their adsorption capacity after five I2 adsorption cycles in air. On the other hand, MFM-300(Fe) and MFM-300(Sc) revealed more moderate changes in resistance (10–100×), maintaining most of their original adsorption capacity after five cycles. This work demonstrates how changes in MOFs can profoundly affect the magnitude and reversibility of the electrical response of sensor materials. Tuning both the intrinsic (resistivity and adsorption capacity) and extrinsic (surface area and particle morphology) properties is necessary to develop highly reversible, large signal-generating MOF materials for direct electrical readout for I2 sensing.


■ INTRODUCTION
Prompt, reliable fission gas detection is paramount for the safety of the public and first responders during both nuclear accidents and industrial nuclear fuel reprocessing. 1 In particular, iodine gas detection is of great concern. The isotopes of iodine have immediate impact to human health and long-term impact to the environment, including 131 I isotope (half life ∼8 days) and 129 I isotope (half life ∼17 million years). Commercial I 2 sensors do exist, though they can possess significant drawbacks. Common fuel-cell type I 2 sensors have relatively short lifetimes coupled with susceptibility to fouling. Other solid-oxide-type sensors require temperatures >200°C for interaction between the gas and oxide surface, limiting applications. Ideally, a compact device easily integrated with modern electronics is desired. For simplicity of design, information output, and device reliability, a direct electrical readout is advantageous.
In an effort to identify materials that enable a compact, reliable, and robust sensor, impedance spectroscopy (IS) is used to monitor series of related materials. IS is a technique that is useful for understanding charge movement in complex materials systems. A small sinusoidal voltage is applied to the sample, and the resulting current is measured, with "impedance" being defined as the ratio of complex voltage to complex current. 2 This measurement is then repeated over a broad frequency range, such as 1 MHz to 10 mHz, to best understand the different components of the system. The use of small sinusoidal voltage about a chosen dc potential enables measurement without necessarily driving a reaction or forcing a net flow of current. IS has been successfully utilized across a variety of systems to understand charge movement in solar cells, 3,4 ion-conducting solid electrolytes, 5,6 and structures at solid−liquid interfaces. 7,8 To develop a sensor based on IS but with high response to I 2 , nanoporous materials with the ability to reversibly adsorb I 2 from the environment are necessary. To date, a variety of zeolites and metal−organic framework (MOF) materials have been shown to selectively adsorb I 2 from complex gas streams, 9 − 1 2 including ZIF-8, 1 3 − 1 6 HKUST-1, 1 7 Zr 6 O 4 (OH) 4 (sdc) 6 , 18 Zr 6 O 4 (OH) 4 (peb) 6 , 18 micro-Cu 4 I 4 -MOF, 19 Zn 3 (DL-lac) 2 (pybz) 2 , 20 Co(bdc) 1.5 (H 2 bpz)0.5I 2 · DMF, 21 TbCu 4 I 4 (ina) 3 (DMF), 22 SBMOF-1 (Ca(sdb)), 23 silica zeolites, 24 and silver-containing zeolites (e.g., silvermordenite). 25−27 Early success was achieved with a thin film of ZIF-8 as the adsorbent layer on a sensor bed of Ptinterdigitated electrodes (IDE). 28 While a large (>10 5 ×) change in impedance was observed, the response was largely irreversible. 28 Subsequent investigation of silver-mordenite, a widely used zeolite in industry for I 2 capture, revealed a structure stable to gas adsorption and temperature but with only a modest <5× change in resistance. 29 In an effort to retain the large electrical response of ZIF-8 MOF but with the stability of silver-mordenite zeolites, sensors are reported herein from a family of thermally and chemically robust porous materials MFM-300(X) (X = Al, Fe, In, Sc) which show exceptional reversible I 2 adsorption capacity. 30 This high capacity is due to the strong iodine binding sites (metal hydroxide site and a site between the ligand phenyl rings). Furthermore, the triply charged metal cations are coordinated to a biphenyl-3,3′,5,5′-tetracarboxylate linker to form a porous "wine-rack" structure through which I 2 molecules can diffuse along the c-axis of the crystal, forming tightly packed triple helices of I 2 within the pores. Previous reports on MFM-300(Sc) have shown this family to exhibit highly reversible adsorption of I 2 . 30 By coupling IS with robust, highly I 2 adsorbing MFM-300(X) materials, a series of prototype sensors were fabricated by dropcasting a different MFM-300(X) (X = Al, Fe, In, Sc) material onto each IDE. The prototype sensors were dried and exposed repeatedly to I 2 followed by desorption via vacuum drying. Herein, the variation and reversibility of electrical response, adsorption capacity, and structure are characterized throughout these I 2 loading and unloading cycles, and large, reversible responses (×10 6 ) are demonstrated. The materials studied herein confirm that tuning both the intrinsic (resistivity and adsorption capacity) and extrinsic (surface area and particle morphology) properties is necessary to create a highly reversible, large signal generating response targeted for direct electrical readout I 2 sensors.

■ EXPERIMENTAL SECTION
Sensor Fabrication. Platinum IDEs on a glass substrate were obtained from DropSens (product G-IDEPT10). These electrodes contain 125 pairs of platinum lines 250 nm thick and 10 μm wide with a spacing of 10 μm between lines. The IDEs were rinsed with acetone (HPLC grade, Sigma-Aldrich, ≥99.9%), dried under N 2 , heated to 70°C in air for 30 min, and cooled to room temperature. IDEs with impedances of less than 2 × 10 10 Ω at 100 mHz were discarded.
The activated MOF materials MFM-300(X) (X = Al, Fe, In, or Sc) were synthesized as reported elsewhere. 30−34 In a 10 mL glass vial, 25 mg of MFM-300(X) and 2.5 mL acetone were added. The mixture was sealed and stirred vigorously for 30 minutes, after which 25 μL was pipetted onto the active area of the IDE. The IDE was allowed to dry at room temperature for 10 min, followed by vacuum drying at 175°C and <1 mTorr for 8 h. To determine the deposition amount, the mass of the IDE was monitored after each step in this process using a calibrated, high-resolution balance (Mettler Toledo XS105). Across all samples, an average of 1.52 ± 0.21 mg (N = 16) MFM-300(X) was deposited onto an IDE.
Iodine Exposure. The MOF-coated IDE and 100 mg of I 2 (ultradry, 10 mesh, Alfa Aesar, 99.998%) were placed into a 100 mL glass vial with a ground glass joint and top. The vial was closed in air and placed in an oven at 70°C for 3 h (I 2 vapor pressure = 1.18 kPa), 35 after which the sensor was promptly removed and allowed to cool to room temperature in air. I 2 exposures were run in triplicate.
To remove adsorbed I 2 , sensors were vacuum-dried at <1 mTorr and 175°C for 8 h. The reversibility of all samples was evaluated by sequentially loading with I 2 and vacuum drying five times.
Electrical Testing. Impedance spectra were recorded after every I 2 adsorption and subsequent desorption using a Solartron 1260 frequency response analyzer connected in series with Solartron 1296 Dielectric Interface, utilizing the internal reference capacitors for measurements. The high input impedance of this system enables measurement of impedances in excess of 10 14 Ω. Impedance spectra were recorded at 0 V dc and 100 mV (rms) ac over 1 MHz to 10 mHz. Sensors were placed on a 5 mm thick alumina plate inside a Faraday cage and contacted with tungsten probes. All measurements were taken at room temperature in air. Data were analyzed, and equivalent circuits fitted using the complex nonlinear least squares algorithm in the Z-Plot (Scribner Associates, North Pines, NC, USA).
Materials Characterization. Crystallinity was evaluated using Xray diffraction (XRD) with a Bruker D2 PHASER system set in traditional Bragg−Brentano geometry using Cu Kα radiation. The morphology and phase assemblage were characterized using a Zeiss GeminiSEM scanning electron microscope operating with an accelerating voltage of 3 kV and a working distance of 5.8 mm. A layer of Au−Pd was sputtered onto samples to minimize the effects of sample charging. Energy-dispersive X-ray spectroscopy (EDS) was performed during scanning electron microscopy (SEM) analysis, using an accelerating voltage of 10 kV and working distance of 10.9 mm and a Bruker XFlash 6 | 60 EDS detector.
The MOF films were removed from the IDE and loaded into polyimide capillaries for synchrotron-based structure analysis. Highenergy X-ray scattering data suitable for PDF and diffraction analysis were collected at beamline 11-ID-B of the Advanced Photon Source at Argonne National Laboratory using an X-ray wavelength of 0.2113 Å. Two dimensional X-ray scattering images were calibrated and reduced to one-dimensional diffraction patterns using GSAS-II. X-ray PDFs were obtained within the xPDFsuite.

■ RESULTS AND DISCUSSION
The materials MFM-300(X) (X = Al, Fe, In, Sc) were synthesized as powders and individually dropcast onto platinum IDE, forming a series of sensors capable of detecting I 2 electrically. IS was used to define the broad electrical frequency response of these materials. The reversibility of the electrical response upon repeated I 2 loading (70°C, 3 h, air) and unloading (175°C, 8 h, <1 mTorr) revealed wide variation between MFM-300(X) as a function of different metal centers X. These observed changes in electrical properties were correlated with changes in adsorption capacity, structural integrity, and morphology and are detailed in depth below. Importantly, in this work, only the MOF metal centers were varied in an effort to minimize outside factors interfering with the impedance of the electrode, such as particle and grain sizes.
It should be noted that all MFM-300(X) materials exhibited a similar impedance response, though the magnitude of the change was variable (see Figures S1−S4). Typical impedance spectra are provided in Figure 1 for MFM-300(In). The assynthesized and dried MFM-300(In) displays a high capacitive response, with |Z| increasing with decreasing frequency, and the phase angle is nearly −90°except for the lowest frequencies. After I 2 is adsorbed, the low-frequency impedance (|Z|) levels out below 1 Hz, and the phase angle transitions from −90°to 0°, indicative of a change in response from capacitive to resistive. Desorption of I 2 from MFM-300(In) leads to the impedance magnitude and phase angle returning to nearly their original values. Thus, the reversible adsorption and desorption of I 2 from MFM-300(In) is accompanied by a reversible change in impedance response.

Research Article
To better quantify the changes in impedance, an equivalent circuit model previously developed for I 2 adsorption in ZIF-8 was used. 28 This circuit, depicted in Figure 2, consists of a series resistance, R s , and two parallel resistor−capacitor networks linked in parallel. The series resistance is dominated by the platinum lines on the IDEs and is typically 400−450 Ω. The first resistor−capacitor network relates the glass substrate resistance, R g (≈10 12 Ω), and capacitance, C g (≈40 pF). Actual values for R s , R g , and C g were measured on each blank IDE before dropcasting with MFM-300(X). These values were then fixed for subsequent analysis of MFM-300(X) resistance, R MFM , and capacitance, CPE MFM . Here, a constant phase element is used to describe the inhomogeneity of the response of MFM-300(X). 36 This equivalent circuit analysis was then performed as IDEs containing MFM-300(X) and was cycled through five I 2 loading and unloading cycles.
The resulting changes in I 2 adsorption capacity (mass change) and resistance of MFM-300(X) (R MFM ) are plotted in Figure 3. Two distinct sets of responses are observed. First, MFM-300(Al) and MFM-300(In) both show a gradual decrease in adsorption capacity, accompanied by an increasing amount of mass retained after desorption. R MFM was consistently high (10 11 Ω) across all readings, with MFM-300(In) displaying the overall highest resistance of all tested materials. However, the resistance of I 2 -loaded MFM-300(Al) and MFM-300(In) both decreased as cycling continued.
Similar to the two previous members of the series, MFM-300(Fe) and MFM-300(Sc) both display large and consistent I 2 adsorption capacities across all five loading cycles. After three cycles, however, a mass increase of 13.6 and 12.0 wt % for MFM-300(Fe) and MFM-300(Sc), respectively, was noted. In contrast to MFM-300(Al) and MFM-300(In), the large, highly reversible adsorption capacities were accompanied by a relatively small change in R MFM (10× and 100×, respectively). MFM-300(Fe) revealed very consistent changes in R MFM from 550 GΩ unloaded to 55 GΩ when loaded with I 2 . MFM-300(Sc) showed similar differences in R MFM , though the unloaded resistance slowly increased with the cycle number.
From these results, it is concluded that a consistent electrical response is indicative of a consistent I 2 adsorption amount. However, large I 2 adsorption capacity is not necessary to create a large change in R MFM . By cycle five, the MFM-300(Al) and (In) devices only adsorbed an additional 28 and 15 wt % I 2 , respectively, yet they exhibited changes in R MFM of ×9 × 10 3 and ×6 × 10 6 . The fact that R MFM continuously decreases for the MFM-300(Al) and MFM-300(In) devices suggests that either (i) a chemical or structural change was occurring to the MOF or (ii) relatively small amounts of I 2 were being adsorbed irreversibly into preferential sites in the MOF pore system, decreasing the resistance.
To verify the structural stability of MFM-300(X) materials, XRD was performed on the sensors after each I 2 cycle. Figure 4 compares diffraction patterns for the blank IDE to that loaded

ACS Applied Materials & Interfaces
Research Article with MFM-300(X) (X = Al, Fe, In, Sc) as dried, after the first I 2 adsorption, after the first desorption, and after the fifth desorption. The blank IDEs display strong Pt(111) reflection near 40°2θ, and a broad peak near 22°, attributed to the glass substrate. The addition of the MFM-300(X) film creates a characteristic set of diffraction peaks, all of which are suppressed once I 2 is adsorbed. This behavior is consistent with previous reports, where scattering from adsorbed I 2 was implicated in the decrease in peak intensity, although the crystal structure of MFM-300(X) was retained. 30 After desorption of I 2 , the diffraction peaks return but with lower intensities. By the fifth desorption, significant differences in peak intensity are observed between the different MFM-300(X) analogues.
These differences are attributed to the amount of I 2 retained in the MOF pores after the fifth desorption cycle. For example, MFM-300(Fe) retained the least I 2 (10.7 ± 0.4 wt %) and showed the most intense XRD pattern. No additional phases were observed during XRD analysis, consistent with MFM-300 phase stability. Thus, it is concluded that changes in R MFM are largely not a result of degradation of the MFM-300 crystal structure. In fact, high energy synchrotron diffraction data shows that the bulk of the MOF film remains crystalline and intact after cycling (crystalline powder). Visual observation of the powder reveals a partial degradation of the top layer of the film into a glassy phase (a minor component of the sample); it is this top layer that is primarily probed in laboratory-based diffraction measurements due to a limited penetration of Xrays into the sample.
Large I 2 absorption capacity is not necessary to create a large change in R MFM . If the electrical conduction mechanism is primarily through networks of I 2 (or generically polyiodides), then, the resistance will not linearly increase with the I 2 content. At some critical amounts of I 2 , a percolating network will form, and the system-level resistance will dramatically decrease. For example, the MFM-300(Al) shows a large response with only ∼15 wt % change after multiple cycles. It is postulated that though a relatively small amount of I 2 is being adsorbed after cycling, I 2 was residing on preferential sites in the MOF pore system. Without enough I 2 to build its extended network inside the MFM-300 pores, the resultant overall electrical resistance decreased for the system.
To understand how particle morphology changes upon repeated I 2 loadings, all sensor films were evaluated in SEM. Characteristic micrographs of MFM-300(X) powders before I 2 exposure and after cycling with I 2 five times are shown in Figure 5. While significant differences in morphology are observed between the different MFM-300(X) analogues, no appreciable changes in morphology were observed upon cycling with I 2 . MFM-300(Al) and MFM-300(In) revealed particles of extraordinary uniformity: for MFM-300(Al) riceshaped particles, ∼1.2 μm in length are observed, while for MFM-300(In) octagonal rods, ∼4.5 μm in length are formed. In contrast, MFM-300(Fe) and MFM-300(Sc) exhibited a wide distribution of orthorhombic particles less than 0.5−25 μm in length, displaying highly variable aspect ratios varying from rods to plates.
EDS was recorded at multiple sites for all powders shown in Figure 5. No significant changes in EDS spectra were observed before versus after I 2 cycling, as shown in Figures S5−S8. For powders before I 2 exposure ( Figure 5A,C,E,G) carbon, oxygen, and the relevant metals were observed. After the fifth I 2 desorption cycle, all powders (Figure 5 B,D,F,H) displayed the same framework elements by EDS but also confirmed the presence of additional iodine. This is consistent with the residual (added) masses observed upon desorption (Figure 3). In conjunction with the previous XRD and SEM data, it is concluded that the residual mass upon I 2 unloading (Figure 3) is predominantly iodine retained in the MFM-300(X) framework and not a secondary species from the environment (e.g., O 2 , H 2 O, etc.).

Research Article
Differences in MFM-300(X) particle morphologies affect bulk adsorption and transport of I 2 gas throughout the pores. Furthermore, the differences in electrical responses in MFM-300(X) are due to the differences in the polarization of I 2 in the pores, which is likely controlled by the amount and concentration of adsorbed I 2 , and by the depth of penetration of adsorption in each crystallite. All of these are controlled to a large part by crystallite morphology and nanopore size opening. The crystal structure of MFM-300(X) resembles a "wine rack", where I 2 molecules enter the c-face and diffuse along the c-axis forming a dense network of triple helices when completely packed. 30 Specific differences in adsorption properties between the MFM-300(X) analogues are summarized in Table 1.
Both MFM-300(Fe) and MFM-300(Sc) display wide distribution particle sizes and orientations. Many of these particles are <1 μm, creating a relatively high surface area and encouraging prompt, reversible I 2 diffusion. MFM-300(Al) has particles that are highly oriented and, on average, smaller than those of MFM-300(Fe) and MFM-300(Sc). As the c-axis of the MOF lies along the long axis of the particles, the effective surface area and pore opening for I 2 to enter the MOF is thus much smaller for MFM-300(Al) than for MFM-300(Fe) and MFM-300(Sc).
This effect is more pronounced for MFM-300(In), where the c-axis is a relatively smaller surface area on the large, long crystallites. The adsorption pathway for I 2 is via the small octagonal face with diffusion down the long axis of the crystallite. As MFM-300(In) is progressively cycled, I 2 may become trapped or hindered at a given point or defect in the channel, creating a blockage in a particular channel and impeding I 2 transport for the entire channel. Surprisingly, as decreasing amount of I 2 molecules are adsorbed on subsequent cycles (Figure 3), the MOF resistance, R MFM , continues to decrease.
It is proposed that I 2 adsorbs at key binding sites in the MFM-300(In) structure, enabling the formation of a conductive network. The additional time and temperature provided by subsequent loading/unloading cycles may optimize these adsorption locations in the crystal structure. This hypothesis is consistent with crystallographic data of the preferential adsorption site of I 2 at low loadings in other MOFs. 26 Upon desorption, enough of the I 2 network may be removed so as to destroy the conduction pathway and create a large increase in resistance, enabling a large sensor response with relatively small addition of I 2 .
■ CONCLUSIONS Development of robust active sensors for the direct electrical detection of I 2 will enable reliable, rapid detection of I 2 in critical environments. Herein, the performance of a family of MOFs as reversible selective adsorption materials in an IS-based sensor has been detailed. Interestingly, the responses of these MFM-300(X) materials can be controlled by the choice of X; MFM-300(Al) and MFM-300(In) displayed large (up to ×10 6 ) changes in resistance upon adsorption of I 2 . While the adsorption capacity of MFM-300(Al) and MFM-300(In) decreased upon successive I 2 cycling, the change in resistance increased. On the other hand, MFM-300(Fe) exhibited stable I 2 adsorption capacities coupled with relatively stable changes in resistance of about ×23. Differences in electrical properties were attributed to variations in morphology of the MFM-300(X) materials and how this influences I 2 diffusion into the crystallites. For MFM-300(In) especially, it is noted that the addition of only a relatively small amount (∼15 wt %) of I 2 is necessary to generate a large change in resistance. This behavior is attributed to the addition of I 2 upon cycling to key adsorption sites, enabling the formation of a conductive I 2 network within the pores of the MOF.
More broadly, this work demonstrates how changes in the metal centers of the MOF can profoundly affect the magnitude and reversibility of the electrical response of the sensor materials. Nanoscale tuning of both the intrinsic (resistivity and adsorption capacity) and extrinsic (surface area and particle morphology) properties is necessary to affect the conductivity of the adsorbed I 2 gas molecules. This in turn is necessary to produce the conductivity pathway necessary for an electrical response due to gas adsorption. Attention to the nanoscale enabled the development of reversible, highlyspecific large signal generation of direct electrical readout I 2 sensors. Ongoing research is focused into the enhanced durability of these sensors and the development of related sensors for the targeting of high-impact industrial gases.

Notes
The authors declare no competing financial interest. a "Observed I 2 uptake" relates the I 2 adsorption during the first adsorption cycle (in air), while "max I 2 uptake" describes the maximum observed I 2 uptake for the material in ref 30 (in N 2

■ ACKNOWLEDGMENTS
The authors thank B. Mckenzie for help in acquiring SEM images. The authors acknowledge partial funding from EPSRC (EP/I011870), the ERC (AdG 742041), and the University of Manchester. This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under contract no. DE-AC02-06CH11357. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-NA0003525. This paper describes objective technical results and analysis. Any subjective views or opinions that might be expressed in the paper do not necessarily represent the views of the U.S. Department of Energy or the United States Government.