Planar Polymer Optical Waveguide with Metal-Organic Framework Coating for Carbon Dioxide Sensing

were obtained. The sensing device also exhibits rapid response to the target sensing gas, with adsorption time of 6 s and desorption time of 16 s, respectively. Sets of cycling experiments demonstrate the reversibility and reproducibility of the sensing performance of the proposed planar polymer waveguide sensor. This proof-of-concept study enables a more flexible fabrication of polymer waveguides, simplifies the introduction of MOF-based optical devices and thus opens up new application poten-tials of this material combination for selective gas detection and sensing. The achievable sensitivity of the approach and the cross-sensitivities to other gases and environmental parameters will be studied in further steps.


Introduction
The increasing attention paid on environment protection and monitoring makes the selective and sensitive sensing of hazardous and greenhouse gases an important issue in environ mental and industrial fields. [1] One of the main greenhouse gases that are largely produced in environment is carbon dioxide (CO 2 ). In the past decades, many different types of CO 2 sensors have been developed with regard to the operation conditions and application fields. Based on sensing mechanisms, CO 2 sensors can be mainly classified into electrochemical gas waveguide with a coated zeolite imidazole framework-8 (ZIF-8) thin film for CO 2 detection and sensing at the near infrared wavelength of 850 nm. ZIF-8 belongs to the group of zeoliteanalogue MOFs and possesses much higher thermal and chemical stability than a lot of other MOFs. [23] It has a large surface area of nearly 2000 m 2 g −1 , a narrow pore aperture diameter of 3.4 Å, as well as high thermal and chemical stability. [10,23] This enables the efficient adsorption of different gases into the pore area without damage of the framework. With regard to the sorption properties of CO 2 in ZIF-8, a study from McEwan et al. has demonstrated the preferred adsorption of CO 2 over nitrogen (N 2 ) in ZIF-8 at 25 °C and the CO 2 isotherm of ZIF-8 is linear in the range from 0 to 1 bar, [24] which is in good agreement with other publications. [25] Since the zinc ions are shielded by the tetrahedral coordination of the nitrogen atoms of the linker molecules, CO 2 can interact much better with the linker molecules than with the IBUs. [23] The proposed polymer waveguides can be produced through different techniques, such as direct laser writing by two-photon polymerization, [26] hot embossing replication, [27] imprint lithography, [28] and projection photolithography. [29] Here, we utilized hot embossing, which is a low-cost and rapid approach toward the fabrication of flexible polymer microstructures, for the replication of proposed optical waveguides made of polymethylmethacrylate (PMMA). A thin ZIF-8 film was grown on top of the produced PMMA waveguides subsequently. Nitrogen and CO 2 were used as the reference gas and the target sensing gas, respectively. An experimental setup was established for the sensing performance characterization. The optical response of the proposed planar waveguide sensor when exposed to pure CO 2 and CO 2 with different concentrations was studied, respectively. The proposed waveguide sensor exhibits a rapid response to CO 2 , with the adsorption time of 6 s and desorption time of 16 s. The cycling sensing results show that the proposed waveguide sensor is able to sense CO 2 reliably and reproducibly.

Materials
In this work, PMMA is employed as the waveguide material. It is a transparent and rigid thermoplastic material, which has a refractive index of ≈1.49 and hence offers high light transmittance. A PMMA sheet (Plexiglas XT 99524, ThyssenKrupp, Germany) was used in the hot embossing process for the fabrication of polymer waveguides. Zinc nitrate-hexahydrate (98%, Sigma-Aldrich), 2-methylimidazole (99%, Sigma-Aldrich) and methanol (99.5%, Roth) were used without further purification for the growth of MOF film on a structured sample.

Fabrication of Polymer Waveguides
Hot embossing allows rapid and large-scale replication of structures with dimensions from microscale to centimeter scale. The fabrication procedures are illustrated in detail in Figure 1.
A structured stamp and a PMMA sheet were first placed onto the upper and lower stages, respectively. They were then heated to 140 °C, which is higher than the glass transition temperature and lower than the melting temperature of PMMA. At this temperature level, the PMMA sheet was in the rubber state and suitable for the subsequent embossing of structure patterns. In the embossing process, the structured stamp was pressed into the PMMA sheet with an embossing force of 4 kN for ≈140 s. A subsequent cooling process followed to cool the stamp and PMMA sheet while maintaining the embossing force. When the temperature reached the demolding temperature (50 °C), the embossing force was released. The stamp and the PMMA sheet could be manually separated. The waveguide sample was then cut into pieces of 20 × 15 mm and cleaned by intensive rinsing with methanol before growing the MOF films on it.

Growth of MOF Films
ZIF-8 films were grown on samples with the cycled direct growth method. [30] To grow ZIF-8 thin films, stock solutions of 25 × 10 −3 m Zn(NO 3 ) 2 •6 H 2 O and 50 × 10 −3 m 2-methylimidazole in methanol were first prepared, respectively. Afterward, the cleaned substrate was immersed in a mixture of the two pre-prepared stock solutions (4 mL of each precursor solution, later called synthesis solution) for 30 min at room temperature and rinsed subsequently using pure methanol. The rinsed sample was then incubated into a freshly prepared synthesis solution again to start the next MOF film growth cycle without a drying step in between. This growth process was repeated for three cycles in order to obtain a desired thicker ZIF-8 film (≈300 nm [30] ). After the last cycle, the sample was dried in air and was then ready for the sensing performance investigation.

Characterization Methods
X-ray diffraction (XRD) measurements on ZIF-8 thin films were carried out with an X-ray diffractometer from STOE working in Bragg-Brentano geometry. An Iso-Debyeflex 3003 was used for the generation of X-rays at 40 kV and 30 mA, delivering CuKα 1 radiation. Measurements were performed between 5° and 30° 2θ with a step size of 0.01° 2θ and a measuring time of 20 s per step. A Sentech SE850 spectroscopic ellipsometer was used to measure the refractive index of ZIF-8. All measurements were performed at angles of 50°, 60°, and 70°. Ellipsometry data were fitted with a Cauchy dispersion model using the software SpectraRay/3. The ellipsometry measurements with variable CO 2 concentrations were performed using mass-flow controllers (El Flow Select, Bronkhorst) for the mixtures of the two gases N 2 and CO 2 in different ratios ranging from 0 to 100 vol% CO 2 . These measurements were performed at ambient pressure and temperature. A total flow rate of 0.5 L min −1 for both gases was used.
Optical microscopy images of MOF-coated polymer waveguides were acquired using a digital microscope from Keyence (VHX7000) and a 3D optical profiler from Sensofar (S neox).
Scanning electron microscopy (SEM, Joel JSM-6610L V) was also employed to observe prepared samples. For this  characterization, samples were fixed with carbon tape on SEM metal substrate holders. Electrical contact of the thin films was ensured by the use of silver conductive lacquer and by sputtering the samples with a thin layer of gold. SEM images were taken at 10 kV with a working distance of 10 mm.

Experimental Sensing Setup
A schematic diagram of the established system used for the investigation of gas sensing performance in this study is illustrated in Figure 2. A four-channel fiber-coupled laser (Thorlabs Inc.) was used as the light source. The output light was first split by a 1 × 2 (50:50) coupler. 50% of the light from the laser source was coupled into a power meter (Thorlabs Inc. S151C) as the reference signal. The other 50% was coupled into the MOFcoated polymer waveguide through a single-mode fiber (SM fiber), which was placed inside a self-designed and 3D printed gas cell with a volume of ≈38 mL. The gas cell was connected to the N 2 and CO 2 gas cylinders, while allowing for the light incoupling to and outcoupling from the waveguide. The transmitted light was coupled into a multi-mode fiber (MM fiber), which was connected to the power meter. Both the reference and sensing data were collected by computers through USB power and energy meter interfaces (Thorlabs Inc. PM100USB), respectively. The gas flow was controlled by flow controllers (FC) with a flow range of 0-1 L min −1 .

Statistical Analysis
The normalized transmission data were calculated by taking the output intensity at the stable state when N 2 is purged as the reference gas and T = 100%. Data were presented as mean ± SD. Analysis and measurements of the size of the inhomogeneity sizes were performed using the software ImageJ.   thick film was grown on a silicon wafer for the measurement. The result is shown in Figure 3b. ZIF-8 has a refractive index of 1.36 at the wavelength of 850 nm. As a comparison, the RI of PMMA, which is the material of the waveguide core, is also plotted in the figure. It can be seen that the RI of ZIF-8 is lower than that of the PMMA, which supports the confinement of light modes within the PMMA waveguide core.

Characterization of Materials and Structures
The waveguide structures before and after ZIF-8 growth were observed using both optical microscopy (see Figure 4a,b) and scanning electron microscopy (SEM, see Figure 4c,d), respectively. The PMMA waveguides shown in Figure 4 have dimensions of ≈100 µm. They exhibit green color after the coating of ZIF-8, as can be seen in Figure 4b. This color is caused by interference effects, while ZIF-8 itself is colorless. [30] ZIF-8 grows in a dense layer on top of the waveguide which is very important for the sensing experiments. Only directly next to the waveguide, a small gap can be seen in Figure S1 (Supporting Information). Further characterization of the coated waveguide surface was performed using AFM and SEM. The AFM topographical images show a densely coated rough surface (see Figure S2, Supporting Information). Analysis of various positions on different waveguides gives an average root-mean-squared roughness of 24.08 ± 3.80 nm. The inhomogeneities on the waveguides surface were determined by measuring a number of inhomogeneity sizes in an SEM image (see Figure S3, Supporting Information) together with subsequent average of those measured data. An average size of 354 ± 96 nm was obtained for these inhomogeneities. The 3D view and profile of the obtained waveguide are shown in Figure 4f,g, demonstrating good structure formation. The measured height of the waveguide from the profile is 25 µm. The dimensions of the waveguide are determined by the master tool used for replication and can be varied over a wide range, from multimode to single mode waveguides.

Sensing Performance Investigation
The sensing performance of the MOF-coated polymer waveguide was investigated at room temperature and a pressure of 1 bar without any pretreatments such as heating or evacuation under vacuum. The gas cell was first purged with the reference gas with a gas flow rate of 0.2 L min −1 for half an hour. After that, the gas was switched from reference gas to CO 2 with a gas flow rate of 0.5 L min −1 . The time interval for gas switching, which also indicates the time duration for gas purging, is denoted as Δt. Different concentrations of CO 2 were achieved by controlling the flow rates of N 2 and CO 2 . The sum of the flow rates for both gases equals 0.5 L min −1 , which is the previously stated flow rate for single gas purging.
Here, sensing performance tests of the proposed device with air as the reference gas were carried out first. The optical responses toward pure CO 2 with different gas changing time interval (Δt = 1 min and Δt = 30 s, respectively) and different CO 2 concentrations are shown in Figure S4 (Supporting Information). It can be seen that the optical sensing device exhibits stable and reproducible sensitivity of CO 2 when air is used as the reference gas. These results indicate that the MOF-coated waveguide sensor performs well in the presence of air.
Nevertheless, an inert gas such as nitrogen or argon is usually employed as the reference gas instead of oxygen or air. This is because oxygen might lead to oxidation of the devices. Air consists of different gas compositions including CO 2 . This will affect the CO 2 concentration that has been set and makes the experiments not controllable. Additionally, the humidity of air may not be stable when surrounding air is employed as the reference gas. In this work, N 2 is used as the reference gas in the following sensing experiments.
With reference gas N 2 , the optical sensing response of a ZIF-8-coated polymer waveguide toward CO 2 was investigated and  characterized. Pure N 2 gas was first purged into the gas cell and was then swtiched to CO 2 , the stable signal of the former representing the base signal (T = 100%). The optical power out of the MOF-coated waveguide was continuously recorded accordingly. At a steady state, the optical power out of the waveguide is ≈186 µW in the presence of N 2 . For a reference, the response of a bare PMMA waveguide toward both gases was investigated, which did not exhibit any signal variation (see Figure S5, Supporting Information). The transmission of optical signals out of the MOF-coated waveguide with a gas switching time interval of Δt = 1 min was recorded as shown in Figure 5a. Multiple repeated cycles were implemented. It can be clearly seen that the transmission signal changed repeatedly and reproducibly with the alternative purging of N 2 and CO 2 . When CO 2 was purged into the gas cell, the transmission decreased quickly and stayed in a relatively stable level. The repeated cycling experiments further confirm the sensing functionality and stability of the developed MOF-based polymer waveguides. With the purpose of investigating the optical response of the planar waveguide sensor when exposed to CO 2 in shorter gas switching periods, an additional experiment with Δt = 30 s was performed. The result is shown in Figure 5b. Also here, the optical signal coming out of the waveguide varies periodically with the alternative purging of N 2 and CO 2 . It can also be seen that the high power level when N 2 is purged does not exhibit an obvious steady state due to the short gas switching time. Nevertheless, the multicycling results verify the high reversibility and reproducibility of the sensing performance of proposed planar polymer waveguides.
The response time of a gas sensor to target substances is of great utility for the evaluation of gas diffusion into the MOF sensing layer. The adsorption and desorption time are defined as the time period from the response onset time to 90% of the maximum response in the steady state. [1] As shown in Figure 5c, the MOF-coated PMMA waveguide in this work has the CO 2 adsorption time of ≈6 s and desorption time of ≈16 s, respectively. These results indicate that the diffusion of gases from the surrounding media into the MOF sensing layer is very rapid. The fast adsorption and desorption can be explained by the physical nature of the interaction between ZIF-8 and CO 2 . The CO 2 molecules are not chemically bound to the inner MOF surface which allows the guests to leave the pores when the purging of CO 2 from the environment is decreased to a lower concentration or completely stopped. This quick response is a significant advantage for real-world applications in gas sensing. In addition, it can be noticed that the CO 2 desorption time is much longer than the adsorption time. It is well-known from work on polycrystalline membranes that ZIF-8 has a higher selectivity for adsorbing CO 2 in preference over N 2 , a result which was recently verified on single-crystal membranes. [31] According to theoretical studies, this is due to the fact that CO 2 molecules exert stronger interactions with the framework than N 2 molecules do. [32] A further experiment was also carried out to investigate the gas sensing performance of the proposed PMMA waveguide sensor with respect to CO 2 concentration. The transmission obtained on the ZIF-8-coated optical waveguide sensor with different CO 2 concentrations is shown in Figure 6a. In this experiment, alternative purging of pure N 2 , pure CO 2 and  the mixtures of N 2 and CO 2 with different concentrations was chosen. It can be seen from the eight cycling results that lower transmission is obtained with higher CO 2 concentration in almost all cycles. This indicates higher signal changes of the ZIF-8-coated planar polymer waveguide sensor with the increase of CO 2 concentration and demonstrates the reversibility of this ZIF-8-coated polymer waveguide sensor. Sensing performance of the same waveguide sensor in the case of direct reduction of CO 2 concentration without the purging of N 2 to recover the transmission signal was investigated as well (see Figure 6b). The transmission values were recorded with CO 2 concentrations of 0, 20, 40, 60, 80, and 100 vol%. These reveal a steady decrease with the increase of CO 2 concentration. A linear fit of the transmission at each CO 2 concentration was also carried out (see Figure 6c). This result indicates the quantifiable relations between the transmission and CO 2 concentration. Within this CO 2 concentration range, the sensing device has a sensitivity of ≈2.5 µW/5 vol%, which is estimated based on the linear fit. A set of corresponding cycling experiments was also carried out with the result shown in Figure S6 (Supporting Information).
The higher decrease of the transmission with increasing CO 2 concentrations can be explained by changes of the MOFs refractive index resulting from the varying adsorption of CO 2 molecules from the environment at different CO 2 concentrations (see Figure 6d). The refractive index of ZIF-8 increases approximately linearly with increasing CO 2 concentration. This is also the case for the transmission that decreases with increasing concentration. The evanescent field effect [33] gives a good explanation for this: light propagates in the waveguide Figure 6. a) Dynamic response of the ZIF-8-coated waveguide sensor with respect to different CO 2 concentrations under the alternative purging of pure N 2 and a gas mixture of N 2 and CO 2 . b) Dynamic response of the ZIF-8-coated waveguide sensor with respect to different CO 2 concentrations without the purging of pure N 2 in between the changing of CO 2 concentration. c) The transmission as a function of CO 2 concentration. d) The measured refractive index of ZIF-8 film at different CO 2 concentrations. core material, which has a higher refractive index than the surrounding material, based on internal reflection at the interface between the waveguide (core, PMMA) and the coated layer (ZIF-8). A higher refractive index difference between the core material and coating material leads to stronger reflection at the interface, thus, a higher output signal from the waveguide results. On the other side, a lower refractive index difference between the core and coating materials results in a higher intensity loss. When CO 2 is purged into the MOF sensing layer, the refractive index increase of the MOF lowers the refractive index difference. As a result, a decrease of the transmission output signal of the sensing device was measured. The experimental result shows that there is only tiny transmission change at low CO 2 concentrations (see Figure S7, Supporting Information). This can be explained by the small refractive index increase of ZIF-8 due to CO 2 adsorption at low CO 2 concentrations. The result in Figure S7 (Supporting Information) reveals a limit of detection (LOD) down to 5 vol%, as the transmission variation percentage from the neighboring higher (7.5 vol%, transmission variation ≈1%) or lower (2.5 vol%, transmission variation ≈0.5 vol%)) concentrations to this concentration is detectable.

Conclusion
In this work, we demonstrate a planar polymer waveguide sensor coated with ZIF-8 thin films for CO 2 sensing. A low-cost and efficient hot embossing technique was introduced for the fabrication of the proposed waveguide structure together with an easy-to-implement technique for the realization of the MOF film. Multimode waveguide structures were employed for the investigations. The waveguide sensor shows sensitive optical response to pure CO 2 as well as CO 2 with different concentrations. A sensitivity of ≈2.5 µW/5 vol% and LOD down to 5 vol% were obtained. The sensing device also exhibits rapid response to the target sensing gas, with adsorption time of 6 s and desorption time of 16 s, respectively. Sets of cycling experiments demonstrate the reversibility and reproducibility of the sensing performance of the proposed planar polymer waveguide sensor. This proof-of-concept study enables a more flexible fabrication of polymer waveguides, simplifies the introduction of MOFbased optical devices and thus opens up new application potentials of this material combination for selective gas detection and sensing. The achievable sensitivity of the approach and the cross-sensitivities to other gases and environmental parameters will be studied in further steps.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.