Eye-Visible Oxygen Sensing via In-Situ Synthesizing Blue-Emitting Cu(I) Cluster in Red-Emitting COF: Characterization and Performance

Covalent organic frameworks (COFs) have shown virtues of well-defined and uniform pores with structural diversity, including the shape, size and even chemical nature of pores. These features are excellent for the application of O2 gas optical sensors. In this paper, two oxygen probes based on halogen-bridged Cu cluster were in-situ synthesized in the micropores of COFs, to allow a uniform distribution. The resulting composite samples were characterized in detail to confirm the successful probe loading. The doping level was determined as ~22%. The halogen-bridged Cu clusters showed blue emission peaking at ~440 nm, while COF host showed red emission peaking at 630 nm. These halogen-bridged Cu clusters had long emissive lifetime of ~6.7 μs and high emission quantum yield of 0.30 in pure N2 atmosphere. Given pure O2 atmosphere, lifetime and quantum yield were quenched to 2.5 μs and 0.11, showing oxygen-sensing possibility. A linear oxygen-sensing calibration curve was observed, with sensitivity of 12.25, response time of 13 s and recovery time of 38 s. Sample emission color was changed from blue to red when testing atmosphere was changed from pure N2 to pure O2, which was detectable by eyes.


Introduction
Porous materials are a class of important host to support functional component in the fields of catalysis, optoelectronics, drug storage/transportation and sensors, which makes the development for porous materials always an attractive topic [1][2][3][4]. As a class of attractive porous crystalline polymers, covalent organic frameworks (COFs) have shown virtues of well-defined and uniform pores [5][6][7]. Their organic building components allow structural diversity, including the shape, size and even chemical nature of pores [8]. These features are excellent for the application of O 2 gas optical sensors. It is well known that O 2 is an important life-supporting gas and its quantification is always an important task in the field of medical treatment, industry, manufacturing and food preservation.
To construct an O 2 gas optical sensor with linear sensing response, photosensitizers should be uniformly loaded into these COF pores, so that the microenvironment around each photosensitizer molecule is the exactly the same to the others [9,10]. Zhang and coworkers have demonstrated a doping method based on ionic exchange for MOF (metal-organic-framework) materials with improved linearity of calibration curves [11,12]. However, the backbone of COF materials is usually neutral, which denies the possibility of dopant loading via an ionic exchange reaction. Some alterative doping methods make the dopant loading generally an inhomogeneous one, leading to a down-bending calibration curve, compromising sensing linearity and sensitivity [13,14]. It is thus a challenge to realize a uniform dopant loading in COF materials.

Oxygen-Sensing Operation
The testing atmosphere was controlled by pure N 2 and pure O 2 flows which were mixed with desired ratio and then imported into a quartz chamber. Both N 2 and O 2 flows were controlled by flowmeters. Each sample was immobilized in the quartz chamber and kept for at least 5 s to achieve atmosphere balance. Steady emission spectra were recorded by the F7000 (Hitachi) spectrometer under luminescence mode (5 nm × 5 nm). Each measurement was repeated three times to get a mean value.

Results and Discussion
3.1. Characterization on CuPX and CuPX-COF, X = Br or Cl 3.1.1. Single Crystal Structure of CuPBr As depicted in Scheme 1 and Section 2.3, the dopant CuPX (X = Br or Cl) was in-situ synthesized in the micropores of Br-COF and Cl-COF. For comparison convenience, CuPX (X = Br and Cl) was synthesized as a reference compound under a similar condition, without Br-COF and Cl-COF. The molecular identity of CuPX (X = Br or Cl) has been confirmed above by NMR, MS and elemental analysis. Additionally, CuPBr single crystal was obtained and presented in Figure 1. Detailed geometric parameters are listed in Supplementary Materials. Two Cu(I) ions are coordinated by two Brions and a POP ligand, with Cu . . . Cu distance of 2.62 Å which is rather close to the radius sum of two Cu atoms (1.28 + 1.28 Å). The two bridging Brions help to stabilize these two Cu(I) ions. The crystal cell length values are measured as a = 12.36 Å, b = 13.80 Å and c= 13.98 Å. This small size ensures the successful loading of CuPBr in Br-COF micropores (with pore size~2 nm), which will be further discussed below. Owing to the free rotation of phenyl rings in POP ligand, each CuBr molecule is far away from the each other, with weak intermolecular interaction, as shown by its Hirshfeld surface plotting shown in Figure 1. Even in packing mode, there is no obvious aggregation or π-π interaction between CuPBr molecules. This is good news for an oxygen-sensing probe since the interaction between probe molecules always compromises sensing performance by barricading O2 impact, resulting in bi-exponential excited state lifetime and thus non-linear quenching behavior [9,10,16].

Electronic Structure of CuPBr
It has been reported that an oxygen-sensing procedure based on luminescence quenching is generally a dynamic one, where 3 O2 (ground state) attacks probe excited electrons, resulting in excited state 1 O2 and probe emission quenching [9,10]. As a consequence, the electronic structure of probe plays an important role in controlling sensing sensitivity, response time and even the linearity of calibration curve. Most reported metal-based probes are charge-transfer-based (CT-based) ones [9][10][11][12]17]. The virtues of a CT-based probe include large Stokes shift to avoid excitation light interference, broad distribution of excited state electrons to increase collision probability with O2, and long lifetime to allow more collision chances with O2 [19]. The electronic structure of CuPBr is revealed by TD-DFT method [17,19]. It is observed from Figure 2 that the highest occupied molecular orbital (HOMO) of CuPBr is composed of Cu and Br atoms, with rather slim contribution from POP ligand, while its lowest unoccupied molecular orbital (LUMO) is basically the π* of the POP ligand, admixed with contribution from Cu d orbital. The onset electronic transition corresponds to a transition from HOMO to LUMO, with excitation energy of 3.40 eV. It is thus assigned as a mixed character of (M + X)LCT. Here M means metal, X means halogen atom, L denotes phosphorous ligand, and CT means, as above mentioned, charge transfer. The observation of such CT transition shall favor the oxygen-sensing behavior of CuPBr, which will be confirmed below. In addition, this excitation energy is found much higher than those of [Cu(N-N)(POP)] + (<3.0 eV) [19]. We attribute this high transition energy to the strong coordination effect from Brions.  Owing to the free rotation of phenyl rings in POP ligand, each CuBr molecule is far away from the each other, with weak intermolecular interaction, as shown by its Hirshfeld surface plotting shown in Figure 1. Even in packing mode, there is no obvious aggregation or π-π interaction between CuPBr molecules. This is good news for an oxygensensing probe since the interaction between probe molecules always compromises sensing performance by barricading O 2 impact, resulting in bi-exponential excited state lifetime and thus non-linear quenching behavior [9,10,16].

Electronic Structure of CuPBr
It has been reported that an oxygen-sensing procedure based on luminescence quenching is generally a dynamic one, where 3 O 2 (ground state) attacks probe excited electrons, resulting in excited state 1 O 2 and probe emission quenching [9,10]. As a consequence, the electronic structure of probe plays an important role in controlling sensing sensitivity, response time and even the linearity of calibration curve. Most reported metal-based probes are charge-transfer-based (CT-based) ones [9][10][11][12]17]. The virtues of a CT-based probe include large Stokes shift to avoid excitation light interference, broad distribution of excited state electrons to increase collision probability with O 2 , and long lifetime to allow more collision chances with O 2 [19]. The electronic structure of CuPBr is revealed by TD-DFT method [17,19]. It is observed from Figure 2 that the highest occupied molecular orbital (HOMO) of CuPBr is composed of Cu and Br atoms, with rather slim contribution from POP ligand, while its lowest unoccupied molecular orbital (LUMO) is basically the π* of the POP ligand, admixed with contribution from Cu d orbital. The onset electronic transition corresponds to a transition from HOMO to LUMO, with excitation energy of 3.40 eV. It is thus assigned as a mixed character of (M + X)LCT. Here M means metal, X means halogen atom, L denotes phosphorous ligand, and CT means, as above mentioned, charge transfer. The observation of such CT transition shall favor the oxygen-sensing behavior of CuPBr, which will be confirmed below. In addition, this excitation energy is found much higher than those of [Cu(N-N)(POP)] + (<3.0 eV) [19]. We attribute this high transition energy to the strong coordination effect from Brions. Owing to the free rotation of phenyl rings in POP ligand, each CuBr molecule is far away from the each other, with weak intermolecular interaction, as shown by its Hirshfeld surface plotting shown in Figure 1. Even in packing mode, there is no obvious aggregation or π-π interaction between CuPBr molecules. This is good news for an oxygen-sensing probe since the interaction between probe molecules always compromises sensing performance by barricading O2 impact, resulting in bi-exponential excited state lifetime and thus non-linear quenching behavior [9,10,16].

Electronic Structure of CuPBr
It has been reported that an oxygen-sensing procedure based on luminescence quenching is generally a dynamic one, where 3 O2 (ground state) attacks probe excited electrons, resulting in excited state 1 O2 and probe emission quenching [9,10]. As a consequence, the electronic structure of probe plays an important role in controlling sensing sensitivity, response time and even the linearity of calibration curve. Most reported metal-based probes are charge-transfer-based (CT-based) ones [9][10][11][12]17]. The virtues of a CT-based probe include large Stokes shift to avoid excitation light interference, broad distribution of excited state electrons to increase collision probability with O2, and long lifetime to allow more collision chances with O2 [19]. The electronic structure of CuPBr is revealed by TD-DFT method [17,19]. It is observed from Figure 2 that the highest occupied molecular orbital (HOMO) of CuPBr is composed of Cu and Br atoms, with rather slim contribution from POP ligand, while its lowest unoccupied molecular orbital (LUMO) is basically the π* of the POP ligand, admixed with contribution from Cu d orbital. The onset electronic transition corresponds to a transition from HOMO to LUMO, with excitation energy of 3.40 eV. It is thus assigned as a mixed character of (M + X)LCT. Here M means metal, X means halogen atom, L denotes phosphorous ligand, and CT means, as above mentioned, charge transfer. The observation of such CT transition shall favor the oxygen-sensing behavior of CuPBr, which will be confirmed below. In addition, this excitation energy is found much higher than those of [Cu(N-N)(POP)] + (<3.0 eV) [19]. We attribute this high transition energy to the strong coordination effect from Brions.  The above analysis on CuPBr single crystal has suggested that its molecular size is no larger than 1.5 nm. Aiming at a tentative evaluation on the possible CuPX doping in X-COF micropores, the monolayer structure and stacking structure of X-COF should be simulated. The monolayer structure of Br-COF and its energy-minimized stacking mode were optimized by universal force-field model and shown as Figure 3 [15]. The diameter of Br-COF micropore is measured as~2.2 nm which is large enough to load CuPBr (a = 12.36 Å, b = 13.80 Å and c = 13.98 Å). There are three Brions in each Br-COF micropore, two of them are able to react with [Cu(CH 3 CN) 2 (POP)]BF 4 , to form one CuPBr molecule. In other words, theoretically, there should be one and only one CuPBr molecule in each Br-COF micropore, due to the restriction of geometric space and charge balance. It is till observed from Figure 3 that Br-COF layers tend to take an offset ABA staking which is an energy-favored structure (285 kcal/mol), compared to the geometrical energy values of 532 kcal/mol for ideal AA stacking mode and 451 kcal/mol for ideal AB staking mode. Considering that Cl-COF was obtained with Br-COF as a starting compound by an ionic exchange reaction, they should have nearly identical backbone microstructure, except for their different counterions (Brfor Br-COF and Clfor Cl-COF).

Simulated Structure of X-COF
The above analysis on CuPBr single crystal has suggested that its mol no larger than 1.5 nm. Aiming at a tentative evaluation on the possible CuP X-COF micropores, the monolayer structure and stacking structure of X-CO simulated. The monolayer structure of Br-COF and its energy-minimized st were optimized by universal force-field model and shown as Figure 3 [15]. of Br-COF micropore is measured as ~2.2 nm which is large enough to loa 12.36 Å, b = 13.80 Å and c = 13.98 Å). There are three Brions in each Br-CO two of them are able to react with [Cu(CH3CN)2(POP)]BF4, to form one CuP In other words, theoretically, there should be one and only one CuPBr mol Br-COF micropore, due to the restriction of geometric space and charge bal observed from Figure 3 that Br-COF layers tend to take an offset ABA stakin energy-favored structure (285 kcal/mol), compared to the geometrical ener 532 kcal/mol for ideal AA stacking mode and 451 kcal/mol for ideal AB s Considering that Cl-COF was obtained with Br-COF as a starting compoun exchange reaction, they should have nearly identical backbone microstructu their different counterions (Brfor Br-COF and Clfor Cl-COF).

XRD Analysis, SEM Morphology, IR Spectra and Microporous Structu
The recorded XRD curves of X-COF and CuPX-COF (X = Br and Cl) Figure 4. There is a sharp XRD peak around 3.3° and a broad one around 2 XRD curve. The first peak matches the simulated XRD peak of Br-COF exchange and bridging-reaction with [Cu(CH3CN)20(POP)]BF4, these two p preserved in Cl-COF and CuPX-COF, with no obvious spectral shift or rela variation. This observation suggests that the hexagonal microstructu constructed and well preserved after loading Cu-based probes. On the ot detectable XRD peaks from dopant CuPBr are observed, which means molecules have been uniformly distributed into COF micropores, with no ag phase separation (More explanation words can be found from Supplementa The recorded XRD curves of X-COF and CuPX-COF (X = Br and Cl) are shown in Figure 4. There is a sharp XRD peak around 3.3 • and a broad one around 27 • in Br-COF XRD curve. The first peak matches the simulated XRD peak of Br-COF. After ionic exchange and bridging-reaction with [Cu(CH 3 CN) 20 (POP)]BF 4 , these two peaks are well preserved in Cl-COF and CuPX-COF, with no obvious spectral shift or relative intensity variation. This observation suggests that the hexagonal microstructure has been constructed and well preserved after loading Cu-based probes. On the other hand, no detectable XRD peaks from dopant CuPBr are observed, which means that dopant molecules have been uniformly distributed into COF micropores, with no aggregation or phase separation (More explanation words can be found from Supplementary Materials).  To confirm the above statement, SEM images of Br-COF, CuPBr-COF and CuPCl-COF are shown in Figure 5. Spherical-liked nanoparticles with diameter of ~1 μm are observed for Br-COF. After ionic exchange and bridging-reaction with [Cu(CH3CN)20(POP)]BF4, the spherical morphology has been well preserved, admixed with some structural fragments. It seems that these gentle operations (ionic exchange and bridging-reaction at ambient condition) have slim impact on Br-COF structure. The elemental mapping of CuPBr-COF is shown in Figure 5 as well. Uniform distribution is observed for Cu element, with no obvious aggregation, suggesting that CuPBr molecules have been uniformly distributed in COF micropores (see Figure S3 of Supplementary Materials for more elemental mapping photos and TEM images). The successful dopant loading in CuPX-COF is further analyzed with IR spectral comparison between CuPX, CuPX-COF and X-COF, X = Br, Cl. The IR spectra of CuPX are similar to each other owing to their rather similar molecular composition. As shown in Figure 6, there are two characteristic bands, peaking at 2925 cm −1 and 1075 cm −1 . The To confirm the above statement, SEM images of Br-COF, CuPBr-COF and CuPCl-COF are shown in Figure 5. Spherical-liked nanoparticles with diameter of~1 µm are observed for Br-COF. After ionic exchange and bridging-reaction with [Cu(CH 3 CN) 20 (POP)]BF 4 , the spherical morphology has been well preserved, admixed with some structural fragments. It seems that these gentle operations (ionic exchange and bridging-reaction at ambient condition) have slim impact on Br-COF structure. The elemental mapping of CuPBr-COF is shown in Figure 5 as well. Uniform distribution is observed for Cu element, with no obvious aggregation, suggesting that CuPBr molecules have been uniformly distributed in COF micropores (see Figure S3 of Supplementary Materials for more elemental mapping photos and TEM images).  To confirm the above statement, SEM images of Br-COF, CuPBr-COF and CuPCl-COF are shown in Figure 5. Spherical-liked nanoparticles with diameter of ~1 μm are observed for Br-COF. After ionic exchange and bridging-reaction with [Cu(CH3CN)20(POP)]BF4, the spherical morphology has been well preserved, admixed with some structural fragments. It seems that these gentle operations (ionic exchange and bridging-reaction at ambient condition) have slim impact on Br-COF structure. The elemental mapping of CuPBr-COF is shown in Figure 5 as well. Uniform distribution is observed for Cu element, with no obvious aggregation, suggesting that CuPBr molecules have been uniformly distributed in COF micropores (see Figure S3 of Supplementary Materials for more elemental mapping photos and TEM images). The successful dopant loading in CuPX-COF is further analyzed with IR spectral comparison between CuPX, CuPX-COF and X-COF, X = Br, Cl. The IR spectra of CuPX are similar to each other owing to their rather similar molecular composition. As shown in Figure 6, there are two characteristic bands, peaking at 2925 cm −1 and 1075 cm −1 . The former peak is assigned as the IR absorption from Cu-X cluster, while the latter one is The successful dopant loading in CuPX-COF is further analyzed with IR spectral comparison between CuPX, CuPX-COF and X-COF, X = Br, Cl. The IR spectra of CuPX are similar to each other owing to their rather similar molecular composition. As shown in Figure 6, there are two characteristic bands, peaking at 2925 cm −1 and 1075 cm −1 . The former peak is assigned as the IR absorption from Cu-X cluster, while the latter one is attributed to the off-plane bending vibration of C-H bond from POP ligand [20]. The IR spectra of Br-COF and Cl-COF are nearly identical to each other due to their identical COF structure, peaking at 1587 cm −1 , 1448 cm −1 and 1273 cm −1 , respectively. The first two IR bands are attributed to vibrations of C = C bonds of phenyl rings, while the latter one is considered as in-plane bending vibration of C-H bond [20]. All above mentioned IR peaks are traced from the IR spectra of CuPX-COF (X = Br and Cl), especially the IR peaks from Cu-X cluster (2925 cm −1 ). It is thus confirmed that dopant CuPX has been successfully in-situ synthesized in the micropores of X-COF, X = Br, Cl.
Materials 2022, 15, x FOR PEER REVIEW attributed to the off-plane bending vibration of C-H bond from POP ligand [20]. spectra of Br-COF and Cl-COF are nearly identical to each other due to their identic structure, peaking at 1587 cm −1 , 1448 cm −1 and 1273 cm −1 , respectively. The first bands are attributed to vibrations of C = C bonds of phenyl rings, while the latte considered as in-plane bending vibration of C-H bond [20]. All above mentioned IR are traced from the IR spectra of CuPX-COF (X = Br and Cl), especially the IR peak Cu-X cluster (2925 cm −1 ). It is thus confirmed that dopant CuPX has been successf situ synthesized in the micropores of X-COF, X = Br, Cl. The above statement is finally confirmed by the N2 adsorption/desorption iso of CuPX-COF and X-COF, X = Br, Cl, as shown in Figure 7. As for X-COF, a sh uptake is observed at low pressure, suggesting the presence of micropores in samples. Their Brunauer-Emmett-Teller (BET) surface area values are determined m 2 /g for Br-COF and 955 m 2 /g for Cl-COF, with pore size values of 16.6 Å and respectively. The smaller porous parameters of Br-COF than those of Cl-C explained by the larger size of Br -(3.92 Å) than Cl -(3.62 Å). After in-situ synthesis/loading, their BET surface area values are greatly decreased (lower t m 2 /g). It is thus confirmed that CuPX dopant has been successfully synthesized/loaded into COF micropores. The doping level of CuPX in CuPX-COF is then discussed by their element and thermal gravimetric analysis (TGA) curves. As mentioned in Section 3.1.3, th The above statement is finally confirmed by the N 2 adsorption/desorption isotherms of CuPX-COF and X-COF, X = Br, Cl, as shown in Figure 7. As for X-COF, a sharp N 2 uptake is observed at low pressure, suggesting the presence of micropores in X-COF samples. Their Brunauer-Emmett-Teller (BET) surface area values are determined as 775 m 2 /g for Br-COF and 955 m 2 /g for Cl-COF, with pore size values of 16.6 Å and 17.3 Å, respectively. The smaller porous parameters of Br-COF than those of Cl-COF are explained by the larger size of Br -(3.92 Å) than Cl -(3.62 Å). After in-situ dopant synthesis/loading, their BET surface area values are greatly decreased (lower than 10 m 2 /g). It is thus confirmed that CuPX dopant has been successfully in-situ synthesized/loaded into COF micropores.

Doping Level Determined by Elemental Data and Thermal Analysis
The doping level of CuPX in CuPX-COF is then discussed by their elemental data and thermal gravimetric analysis (TGA) curves. As mentioned in Section 3.1.3, there are three halogen atoms in each X-COF micropore, two of them are able to react with [Cu(CH 3 CN) 2 (POP)]BF 4 , to form one CuPX molecule (X = Br and Cl). In other words, theoretically, there should be one and only one CuPX molecule in each X-COF micropore, due to the restriction of geometric space and charge balance. The recorded C/N/H composition of CuPX-COF is comparable to the theoretical C/N/H composition of CuPX-COF, which confirms the 1:1 loading in each X-COF micropore.
A more precise result is given via the TGA curves of X-COF, CuPX and CuPX-COF, as shown in Figure 8. To assist weight loss assignment, differential thermal gravimetric (DTG) curves are plotted. Br-COF and Cl-COF have three endothermic peaks, centering at 68 • C, 466 • C and 560 • C. The former one is attributed to the thermal release of adsorbent molecules such as water, while the latter two ones are attributed to the thermal decomposition and collapse of COF structure. CuPBr and CuPCl depict mono endothermic peak, centering at 384 • C and 305 • C, respectively. The endothermic peaks of CuPX-COF are composed of those from dopant CuPX and host X-COF, with minor temperature shift, due to the interaction between CuPX and X-COF. composition of CuPX-COF is comparable to the theoretical C/N/H composition of CuPX-COF, which confirms the 1:1 loading in each X-COF micropore. A more precise result is given via the TGA curves of X-COF, CuPX and CuPX-COF, as shown in Figure 8. To assist weight loss assignment, differential thermal gravimetric (DTG) curves are plotted. Br-COF and Cl-COF have three endothermic peaks, centering at 68 °C, 466 °C and 560 °C. The former one is attributed to the thermal release of adsorbent molecules such as water, while the latter two ones are attributed to the thermal decomposition and collapse of COF structure. CuPBr and CuPCl depict mono endothermic peak, centering at 384 °C and 305 °C, respectively. The endothermic peaks of CuPX-COF are composed of those from dopant CuPX and host X-COF, with minor temperature shift, due to the interaction between CuPX and X-COF.

Photophysical Parameters of CuPX under N2 and O2: Quantum Yield and Lifetime
Some crucial photophysical parameters of CuPX are recorded so that their oxygensensing performance can be tentatively evaluated. It is observed from Figure 9 that, under pure N2 atmosphere, CuPX exhibits Gaussian-liked blue emission, peaking at 440 nm for CuPBr and 450 nm for CuPCl. Their emission quantum yields (Φ) are determined as 0.30 and 0.31, with excited state lifetime (τ) as long as 6.7 μs and 6.8 μs, respectively. These long-lived excited states suggest that they have a phosphorescent nature, which allows enough chances to be quenched by O2. Given pure O2 atmosphere, CuPX emission is obviously quenched, with emission quantum yields decreased to 0.11 for CuPBr and 0.24 for CuPCl, respectively. Their lifetimes are quenched to 2.5 μs and 5.2 μs. This observation

Photophysical Parameters of CuPX under N 2 and O 2 : Quantum Yield and Lifetime
Some crucial photophysical parameters of CuPX are recorded so that their oxygensensing performance can be tentatively evaluated. It is observed from Figure 9 that, under pure N 2 atmosphere, CuPX exhibits Gaussian-liked blue emission, peaking at 440 nm for CuPBr and 450 nm for CuPCl. Their emission quantum yields (Φ) are determined as 0.30 and 0.31, with excited state lifetime (τ) as long as 6.7 µs and 6.8 µs, respectively. These long-lived excited states suggest that they have a phosphorescent nature, which allows enough chances to be quenched by O 2 . Given pure O 2 atmosphere, CuPX emission is obviously quenched, with emission quantum yields decreased to 0.11 for CuPBr and 0.24 for CuPCl, respectively. Their lifetimes are quenched to 2.5 µs and 5.2 µs. This observation suggests that CuPX emission is quenchable by O 2 , which endows CuPX with oxygensensing possibility. On the other hand, there is no obvious spectral shift or bandshape change, indicating that the CT-based excited state is well preserved. The absorption spectra of CuPBr upon pure N 2 and pure O 2 atmospheres are recorded and compared in Figure S4 (Supplementary Materials). No obvious difference is observed. This is because the oxygensensing mechanism is a dynamic one, via a dynamic collision between CuPBr triplet excited state and O 2 molecules. CuPBr ground state takes no participation in the sensing procedure. As a consequence, the electronic transition of CuPBr ground state (namely its absorption) is immune from O 2 level variation. Aiming at a better understanding on CuPX excited state, corresponding emissive and non-emissive probabilities (k r and k nr ) are calculated by Equations (1) and (2). Φ = k r /(k r + k nr ) (1) Materials 2022, 15, x FOR PEER REVIEW 10 of 14 triplet excited state and O2 molecules. CuPBr ground state takes no participation in the sensing procedure. As a consequence, the electronic transition of CuPBr ground state (namely its absorption) is immune from O2 level variation. Aiming at a better understanding on CuPX excited state, corresponding emissive and non-emissive probabilities (kr and knr) are calculated by Equations (1) and (2).

Emission Spectra under Various O2 levels
The oxygen-sensing performance of CuPX-COF (X = Br and Cl) is tentatively discussed by comparing its steady emission spectra upon addition of various O2 levels. It is observed from Figure 10 that CuPX-COF exhibits characteristic emission bands from CuPX and X-COF, peaking at 440 nm and 630 nm for CuPBr-COF, 450 nm and 630 nm for CuPCl-COF, respectively. The former emission band of each CuPX-COF sample comes from dopant CuPX, while the latter one comes from host X-COF. It is observed that the dopant blue emission is obviously quenched by increasing O2 level, but X-COF red emission is just slightly quenched. In this case, an emission color change from blue (under pure N2) to red (under pure O2) is observed, as shown in Figure 10. To reveal the nature of COF red emission quenching at 630 nm (O2 quenching or photodegradation), emission monitoring of CuPBr-COF at 630 nm upon pure N2 ad pure O2 atmospheres is performed and shown as Figure S6 (Supplementary Materials). Upon pure O2-pure N2-pure O2 cycles, COF red emission is correspondingly quenched-recovered-quenched. As a consequence, we tentatively conclude that the COF emission quenching is mainly caused by O2 quenching effect, instead of photodegradation.

Emission Spectra under Various O 2 levels
The oxygen-sensing performance of CuPX-COF (X = Br and Cl) is tentatively discussed by comparing its steady emission spectra upon addition of various O 2 levels. It is observed from Figure 10 that CuPX-COF exhibits characteristic emission bands from CuPX and X-COF, peaking at 440 nm and 630 nm for CuPBr-COF, 450 nm and 630 nm for CuPCl-COF, respectively. The former emission band of each CuPX-COF sample comes from dopant CuPX, while the latter one comes from host X-COF. It is observed that the dopant blue emission is obviously quenched by increasing O 2 level, but X-COF red emission is just slightly quenched. In this case, an emission color change from blue (under pure N 2 ) to red (under pure O 2 ) is observed, as shown in Figure 10. To reveal the nature of COF red emission quenching at 630 nm (O 2 quenching or photodegradation), emission monitoring of CuPBr-COF at 630 nm upon pure N 2 ad pure O 2 atmospheres is performed and shown as Figure S6 (Supplementary Materials). Upon pure O 2 -pure N 2 -pure O 2 cycles, COF red emission is correspondingly quenched-recovered-quenched. As a consequence, we tentatively conclude that the COF emission quenching is mainly caused by O 2 quenching effect, instead of photodegradation. For a comparison between CuPBr-COF and CuPCl-COF, sensitivity is defined as I0/I100, where I0 means the emission intensity at 0% O2 and I100 denotes that at 100% O2, respectively. The sensitivity values of CuPBr-COF and CuPCl-COF are determined as 12.25 and 1.50, respectively, where CuPBr-COF shows a much higher sensitivity than CuPCl-COF. Considering their nearly identical geometric structure and composition, we attribute this sensitivity difference to the heavy-atom-turbulence effect of Br in CuPBr, which increases the phosphorescent nature of CuPBr emission, favoring 3 O2 attack. CuPCl emission has less phosphorescent composition to be quenched by 3 O2, leading to its limited sensitivity. This observation explains why CuPCl excited state (τ = 6.8 μs in pure N2 vs. τ = 5.2 μs in pure O2) is less quenched in pure O2, compared to the case of CuPBr (τ = 6.7 μs in pure N2 vs. τ = 2.5 μs in pure O2). In addition, the porous structure of X-COF offers a high specific-surface-to-volume ratio, which improves sensitivity by allowing more dopant molecules to meet and be quenched by O2.
The selectivity of CuPBr-COF is tentatively discussed via its emission spectra upon various gases, including CO2, H2, CH4, C2H2 and moisture. It is observed from Figure S7 (Supplementary Materials) that CuPBr-COF emission bands (440 nm and 630 nm) are nearly constant upon the first four gases, indicating a good selectivity. This is because they are closed-shell structure and are not able to accept energy from CuPBr phosphorescence. Moisture, however, has quenching effect on CuPBr emission since H2O may quench the triplet CuPBr excited state. Thus, to ensure precise and reliable result, testing gas should be dried before sensing.

Response and Recovery
Aiming at an evaluation on the correlation between O2 presence and CuPX-COF emission, CuPX-COF emission is monitored when testing atmosphere is switched between pure N2 and pure O2. It is observed from Figure 11 that CuPBr-COF (440 nm) and CuPCl-COF (450 nm) emission remains at a high level in pure N2 atmosphere. Upon pure O2 atmosphere, their emission is instantly quenched to a low level and preserved. Their emission intensity can be recovered back to a high level given a pure N2 atmosphere. To compare their sensing response performance, response time is defined as the time for each sample to lose 95% of its initial emission intensity (from pure N2 to pure O2), while recovery time is defined as the time for each sample to recover 95% of its initial emission intensity (from pure O2 to pure N2). The response time values of CuPBr-COF and CuPCl-COF are determined as 13 s and 13 s, while their recovery time values are determined as 38 s and 40 s. Their rather similar response/recovery performance is attributed to their For a comparison between CuPBr-COF and CuPCl-COF, sensitivity is defined as I 0 /I 100 , where I 0 means the emission intensity at 0% O 2 and I 100 denotes that at 100% O 2 , respectively. The sensitivity values of CuPBr-COF and CuPCl-COF are determined as 12.25 and 1.50, respectively, where CuPBr-COF shows a much higher sensitivity than CuPCl-COF. Considering their nearly identical geometric structure and composition, we attribute this sensitivity difference to the heavy-atom-turbulence effect of Br in CuPBr, which increases the phosphorescent nature of CuPBr emission, favoring 3 O 2 attack. CuPCl emission has less phosphorescent composition to be quenched by 3 O 2 , leading to its limited sensitivity. This observation explains why CuPCl excited state (τ = 6.8 µs in pure N 2 vs. τ = 5.2 µs in pure O 2 ) is less quenched in pure O 2 , compared to the case of CuPBr (τ = 6.7 µs in pure N 2 vs. τ = 2.5 µs in pure O 2 ). In addition, the porous structure of X-COF offers a high specific-surface-to-volume ratio, which improves sensitivity by allowing more dopant molecules to meet and be quenched by O 2 .
The selectivity of CuPBr-COF is tentatively discussed via its emission spectra upon various gases, including CO 2 , H 2 , CH 4 , C 2 H 2 and moisture. It is observed from Figure  S7 (Supplementary Materials) that CuPBr-COF emission bands (440 nm and 630 nm) are nearly constant upon the first four gases, indicating a good selectivity. This is because they are closed-shell structure and are not able to accept energy from CuPBr phosphorescence. Moisture, however, has quenching effect on CuPBr emission since H 2 O may quench the triplet CuPBr excited state. Thus, to ensure precise and reliable result, testing gas should be dried before sensing.

Response and Recovery
Aiming at an evaluation on the correlation between O 2 presence and CuPX-COF emission, CuPX-COF emission is monitored when testing atmosphere is switched between pure N 2 and pure O 2 . It is observed from Figure 11 that CuPBr-COF (440 nm) and CuPCl-COF (450 nm) emission remains at a high level in pure N 2 atmosphere. Upon pure O 2 atmosphere, their emission is instantly quenched to a low level and preserved. Their emission intensity can be recovered back to a high level given a pure N 2 atmosphere. To compare their sensing response performance, response time is defined as the time for each sample to lose 95% of its initial emission intensity (from pure N 2 to pure O 2 ), while recovery time is defined as the time for each sample to recover 95% of its initial emission intensity (from pure O 2 to pure N 2 ). The response time values of CuPBr-COF and CuPCl-COF are determined as 13 s and 13 s, while their recovery time values are determined as 38 s and 40 s. Their rather similar response/recovery performance is attributed to their nearly identical geometric structure and composition. The recovery time is 3-fold longer than the response time. This is because the recovery process is a dynamic diffusion-controlled one, [21]. In addition, it is observed that there is a gradual smooth increase for CuPX-COF emission in pure N 2 atmosphere, indicating the adsbrobed/residual O 2 in sample, which is attributed to the micropores of X-COF having high affinity for O 2 gas. erials 2022, 15, x FOR PEER REVIEW 12 of controlled one, [21]. In addition, it is observed that there is a gradual smooth increase CuPX-COF emission in pure N2 atmosphere, indicating the adsbrobed/residual O2 sample, which is attributed to the micropores of X-COF having high affinity for O2 gas (a) (b) Figure 11. Emission monitoring (a) and Stern-Volmer plots (b) of CuPX-COF, X = Br, Cl.

Calibration Curve
The above discussion has confirmed a dynamic quenching mechanism of CuPX-C for O2. In this case, the steady emission intensity upon various oxygen levels can analyzed by Stern-Volmer equation described by Equation (3) [22,23]. Here, Ksv is Ste Volmer constant, [O2] means oxygen level.

I0/I = C + Ksv[O2]
An ideal Stern-Volmer equation should be a linear one, given a condition that pro molecules are uniformly distributed and their emission is homogeneously quenched O2. The Stern-Volmer plots of CuPBr-COF follow a linear response, as expected, w fitting equation of I0/I = 1.033 + 0.110*[O2], R 2 = 0.999. But those of CuPCl-COF are no linear ones and fail to obey Equation (3). It has been above mentioned that CuPCl emiss has less phosphorescent composition than CuPBr emission. As a consequence, it assumed that there should be multiple sensing sites in CuPCl-COF, some of them oxygen-quenchable, while the others are not. In this case, a two-site Demas model shou be applied to describe CuPCl-COF steady emission spectra, as shown by Equation [22,23]. Here, f1 and f2 are fractional contributions of sensing sites (f1 + f2 = 1), Ksv1 and K are corresponding Stern-Volmer constants of sensing sites.

Calibration Curve
The above discussion has confirmed a dynamic quenching mechanism of CuPX-COF for O 2 . In this case, the steady emission intensity upon various oxygen levels can be analyzed by Stern-Volmer equation described by Equation (3) An ideal Stern-Volmer equation should be a linear one, given a condition that probe molecules are uniformly distributed and their emission is homogeneously quenched by O 2 . The Stern-Volmer plots of CuPBr-COF follow a linear response, as expected, with fitting equation of I 0 /I = 1.033 + 0.110*[O 2 ], R 2 = 0.999. But those of CuPCl-COF are non-linear ones and fail to obey Equation (3). It has been above mentioned that CuPCl emission has less phosphorescent composition than CuPBr emission. As a consequence, it is assumed that there should be multiple sensing sites in CuPCl-COF, some of them are oxygen-quenchable, while the others are not. In this case, a two-site Demas model should be applied to describe CuPCl-COF steady emission spectra, as shown by Equation (4) [22,23]. Here, f 1 and f 2 are fractional contributions of sensing sites (f 1 + f 2 = 1), K sv1 and K sv2 are corresponding Stern-Volmer constants of sensing sites. ])}, R 2 = 0.999. It is observed that K sv1 is comparable to that of CuPBr-COF, but K sv2 is close to 0, which means that its emission is nearly non-quenchable by O 2 . This conclusion is consistent with the obvious emission of CuPCl under pure O 2 atmosphere. Some important sensing parameters of CuPX-COF are compared to literature ones in Table 1. It is observed that CuPBr-COF is a promising one, showing virtues of high sensitivity, linear calibration curve, short response time, along with visual color change during sensing procedure.

Conclusions
As a conclusion, this paper reported two oxygen probes based on halogen-bridged Cu cluster and their oxygen-sensing performance. They were loaded into COF micropores by an in-situ method. The resulting composite samples were characterized in detail to confirm the successful probe loading, including single crystal analysis, DFT calculation, XRD, SEM, IR, N 2 adosprtion/desorption, and TGA. The doping level was determined as 22%. The halogen-bridged Cu clusters showed blue emission peaking at~440 nm, while COF host showed red emission peaking at 630 nm. These halogen-bridged Cu clusters had long emissive lifetime of~6.7 µs and high emission quantum yield of 0.30 in pure N 2 atmosphere. Given pure O 2 atmosphere, lifetime and quantum yield were quenched to 2.5 µs and 0.11, showing oxygen-sensing possibility. A linear oxygen-sensing calibration curve was observed, with sensitivity of 12.25, response time of 13 and recovery time of 38 s. Sample emission color was changed from blue to red when testing atmosphere was changed from pure N 2 to pure O 2 , which was detectable by eyes. It was found that Br-containing probe was superior to Cl-containing one by showing higher sensitivity and linear calibration curve, due to the heavy atom turbulence effect. For further effort, sensitivity can be further improved by incorporating more heavy atoms into probe structure. This work proposed a method of constructing an oxygen-sensing system by in-situ (one-step) synthesizing lightemitting Cu(I) cluster in luminescent porous COF, so that probe molecules can be uniformly distributed in COF micopores. Sensitivity and linearity of the calibration curve can be improved by this method, compared to the sensing systems prepared by two-step methods.