All chemicals were obtained from commercial sources and used without further purification. Copper(II) acetate (99.999%) and 2,3,6,7,10,11-hexahydroxytriphenylene (HHTP) ligand (95%) were purchased from Alfa Aesar and Acros Organics, respectively.

Figure 1 depicts the fabrication process and a structural schematic of the MOF-derived MO_x/C composite sensor prepared by laser irradiation. Initially, a Cu₃HHTP₂ MOF was formed by a LbL process on a Si/SiO₂ substrate with interdigitated electrodes (13 μm width and 7 μm gap). An optimized LbL process was conducted through successive immersion of a functionalized substrate in an ethanolic solution of metal cations and the HHTP ligand under ambient conditions. The trigonal HHTP ligand coordinates with square planar Cu²⁺ ions, forming an extended two-dimensional hexagonal layer in the ab plane [19]. This liquid-phase epitaxial process allows for precise control over the MOF film thickness compared to other coating techniques, such as slurry coating, solvothermal growth, or drop-casting [2,20,21]. The fabrication approach without the transfer process is also free from contamination issues, preventing the incorporation of impurity particles that can cause thermal damage during laser irradiation [22]. Moreover, Cu₃HHTP₂ is a well-established two-dimensional semiconducting MOF capable of gas sensing even in the pristine state [23]. Therefore, Cu₃HHTP₂ was selected as a proof of concept because it easily facilitated a straightforward comparison of gas sensing performance with that of the laser-irradiated MO_x/C composite. After the formation of the MOF, the Cu₃HHTP₂ film was irradiated by a focused 532 nm continuous-wave laser to form a composite of metal oxide and carbon, which was controlled in the X-Y-Z directions using a precision stage. This laser writing process enabled the programmable mask-free patterning of CuO/C with a minimum line width of 2 μm, which was difficult to achieve with the previous solution-based process. Finally, the nonirradiated MOF region was removed by immersion in a basic solution. Additionally, the produced copper oxide is a p-type oxide semiconductor. Compared to n-type oxide semiconductors (such as ZnO, SnO₂, and TiO₂), which were typically used in previous oxide gas sensor studies, it exhibits unique catalytic activity and offers advantages in selectively detecting gaseous molecules through high redox activity. For this reason, contributing to the study of relatively underresearched p-type oxides (which account for only 12.06% of all gas sensor papers) will benefit the future expansion of the gas sensing library [24,25].

To demonstrate the successful growth of the MOF and the formation of CuO/C through the laser process, we implemented imaging analysis and analytical characterization. Figure 2a displays the SEM images of the Cu₃HHTP₂ MOF nanocrystals. The FT-IR spectroscopy spectrum and XRD pattern show successful growth of Cu₃HHTP₂, in good agreement with previous studies (Fig. S1) [26,27]. After the laser calcination process, CuO/C nanoparticles were patterned on the substrates without delamination, perforation, or crack issues, as shown in Fig. 2b. Furthermore, the laser-induced accelerated heating and cooling of the MOF can prevent the formation of large and irregular particles during film formation. Thus, it is possible to retain porosity, resulting in enhanced sensing performance, including high sensitivity and fast response. The TEM image of CuO/C reveals distinct lattice fringes with interplanar distances of 0.27 nm in the (110) crystal plane of CuO, as shown in Fig. 2c [28]. The cross-sectional EDS results for CuO/C show uniform distributions of C, O, and Cu, confirming the formation of metal oxide and carbon compounds (Fig. 2d-g). Additionally, ToF-SIMS was introduced to show the programmable and selective laser patterning process (Fig. 2h, i). It shows the mapping of Cu and C along the patterned symbol of our institute.

To further investigate the nature and chemical environment of MOF-derived CuO/C, spectroscopic analyses were performed. Figure 3a shows the high-resolution Raman spectra of pristine Cu₃HHTP₂ (upper) and CuO/C (bottom). In pristine Cu₃HHTP₂, the C-H in-plane bending modes (1,270 and 1,179 cm⁻¹) and the stretching of aromatic C-C bonds (1,400, 1,468, and 1,547 cm⁻¹) were observed, in addition to the G and D bands resulting from the graphitic platform of the aromatic triphenylene ligand [27]. After laser irradiation, the vibration peaks attributed to organic ligands disappeared, and only the D and G bands arising from graphitic carbon were observed. Additionally, the 2D band was preserved, which is the fingerprint signal of graphene and the second-order overtone of the D band [29]. However, the 2D band exhibited a lower intensity than the broad D + D’ peak, indicating that the resulting carbonaceous component is a multilayered and amorphous sp² carbon structure. Similarly, several recent studies involving the annealing of MOFs have reported the coating of graphitic carbon on the surface of the resulting products [11,12,30]. During laser irradiation, localized heating caused the pyrolysis of organic ligands, resulting in the formation of Cu²⁺ ions. These ions then reacted with oxygen species in the atmosphere to form copper oxide [31]. Then, upon instant cooling, the thermally decomposed carbonaceous materials from the organic ligands (such as C₂H₂, CH₄, and CO) formed a graphitic carbon layer on the surface of the oxide [30]. Additionally, high-resolution Raman mapping was performed to show the miniaturization of the laser patterning process. Figure 3b displays the spatial mapping of the peaks to baselines of graphitic G bands of CuO/C patterned with a 2 µm minimum line width. It should be noted that the patternable size of CuO/C in the laser writing process has the potential to be further reduced. Such miniaturization can be realized by optimizing parameters affecting the laser beam size, such as laser fluence, beam shaping, and the choice of objective lens.

The chemical environment and valence states of MOF-derived CuO/C were characterized by XPS. In the O 1s spectra, pristine Cu₃HHTP₂ can be deconvoluted into the main C=O peak (531.6 eV), the C-O peak (532.9 eV), and the O-Cu peak (530.5 eV), as shown in Fig. 3c. After laser irradiation, the ratio of C=O and C-O decreased due to the calcination of the organic ligand, resulting in the clear distinction of the O-Cu peak (Fig. 3d). In Fig. 3e, the Cu 2p_3/2 peak of Cu₃HHTP₂ exhibited an asymmetric shape because the redox-active HHTP ligand capable of having multiple oxidation states exists in the semiquinonate and catecholate states, resulting in Cu²⁺ and Cu⁺ mixed-valency metal states [20,27]. In contrast, laser-processed CuO/C clearly showed a Cu(II) valence state, which was characterized by a binding energy at 933.5 eV and a strong Cu²⁺ satellite, reflecting the influence of the intensive laser fluence during the patterning process (Fig. 3f) [32,33].

The sensing performance of MOF-derived CuO/C was assessed for ethanol gas detection at room temperature. Due to the dual threats of explosion hazards and physiological implications associated with ethanol, there is a critical need for rapid monitoring of ethanol concentrations over a broad range [1]. In Fig. 4a, the ethanol sensing performance of pristine Cu₃HHTP₂ and laser-calcined CuO/C sensors was compared. Upon exposure to 170 to 3,400 ppm (10 min exposure—50 min recovery cycle), the CuO/C sensor could rapidly detect ethanol gas over a wide range that spans the permissible exposure limits for the general industry set by the Occupational Safety and Health Administration (1,000 ppm) and 10% of the lower explosion limit (3,300 ppm) for explosion alarm systems. In particular, CuO/C displayed a much faster response and recovery time (105 and 18 s, respectively) than the pristine Cu₃HHTP₂ MOF (Fig. 4b). In addition to exhibiting fast recovery and a stable baseline signal, it was free from the baseline-shift issue common in previous chemiresistor-type sensors [34,35]. These superior properties ensure reliable operation that can distinguish gas concentrations even during prolonged device operation. The response at adsorption-desorption equilibrium was plotted against the concentration. It revealed a linear relationship between ln(response) and ln(concentration) (Fig. 4c), a typical feature of the Langmuir-Freundlich chemical adsorption model [36]. This linear relationship allows for the accurate and straightforward readout of gas concentrations. The theoretical limit of detection (LoD) was also calculated using two methods. The first was obtained by setting the concentration value corresponding to a 10% response in a linear approximation, which is 5.464 ppm [19]. The second method, based on IUPAC recommendations, calculates it as 3 × (root mean square of noise/slope of regression), resulting in an LoD of 1.109 ppm [27].

We further investigated selectivity, which is one of the crucial requirements for gas sensor operation in the real world. As shown in Fig. 4d, the response value under 900 ppm ethanol exposure was 213%, showing the greatest response compared to the same concentration of other VOCs and odorant molecules in the air (acetone, isopropanol, geraniol, 1-heptanol, 2-ethylfenchol, D-limonene, octanal, eugenol, cis-3-hexenol, and 2,3,5-trimethylpyrazine). The excellent response performance of CuO/C could be elucidated through the energy band diagram in Fig. 4e. When copper(II) oxide, which has a bandgap energy of 1.2-1.5 eV and a work function of 5.0-5.1 eV, forms a junction with carbon, a built-in potential results [4,37,38]. Although pinpointing the energy structure of the carbon product from the laser calcination process can be challenging, the product is inferred to have a small bandgap and a work function of 4.6 eV. This is based on energy structure studies of laser-induced graphene and reduced graphene oxides, which exhibit similar few-layered sp² carbon structures (Fig. S2) [27,39,40]. Hence, numerous p-p junctions are formed in CuO/C nanoparticles. During gas adsorption, as the Fermi level of carbon decreases, this leads to a reduction in the energy barrier. Consequently, there is an increase in the conduction of holes, which are the primary carriers of copper oxides. This results in a sensitive conversion into electrical signals.

The strong interaction between CuO/C and ethanol can be attributed to the influence of the surface graphitic carbon layer formed from the localized rapid heating and cooling in the laser process. The graphitic carbon, which can be observed in Raman spectra (Fig. 3a) and reported in recent studies on laser-irradiated metal-organic frameworks [30,31], offers a high density of defect sites and functional groups. To investigate the chemical environment of the formed carbon, we analyzed the XPS C 1s spectrum of MOF-derived CuO/C, as shown in Fig. S3. Interestingly, besides the C-C peak, which is the basal plane of carbon in the CuO/C composite, strong intensities of C-O and C=O peaks were observed. These peaks imply the presence of a large number of functional groups on the carbon surface that can form hydrogen bonds with the -OH of ethanol, such as hydroxyl (-OH) and carboxyl groups (-COOH). Furthermore, the reason for the high sensor selectivity toward ethanol, even among other alcohols, is that the α-C-H bond in ethanol is weaker than the α-C-H bond in secondary alcohols like isopropyl alcohol (IPA), leading to stronger interactions (hydrogen bonding) with the functional groups on the carbon surface [41]. Additionally, non-alcoholic VOCs like acetone and complex aromatic molecules have relatively high activation energies. Thus, the CuO/C sensor can enhance selectivity toward ethanol at room temperature.

The thermal stability of gas sensing materials not only enables their use in a wide range of applications, such as in factories or the automotive industry, but also ensures compatibility when integrated with other sensors or during subsequent fabrication stages [42]. Figure 4f displays the resistance changes after thermal heating for both pristine MOF and MOF-derived CuO/C. With calcination of the MOF ligand by a laser process, CuO/C showed little change in electrical resistance even when the temperature changed from room temperature to 200 °C. On the other hand, in the case of pristine Cu₃HHTP₂, ligand pyrolysis occurred at 160 °C, as indicated in the thermogravimetric analysis (TGA) plot (Fig. 4g), thereby restricting its application in environments with demanding temperature endurance requirements. Furthermore, as shown in Fig. S4, we demonstrated the long-term stability and reusability by examining the changes in the ethanol sensing performance four weeks after the initial experiment.