N-and F-Co-Doped Carbon Quantum Dots Coated on a Ni Foam Substrate as Current Collector for Highly Stable Li-Air Batteries

Because of their high theoretical density, Li-air batteries (LABs) are expected to be critical components of future energy storage devices. However, the commercial use of LABs is limited by their low charge-discharge e ﬃ ciency, high overpotential, and low cyclic stability. Therefore, to improve the cyclic stability of LABs, we manufactured N-and F-co-doped carbon quantum dots (NF-CQDs), thinly coated them on the surface of a Ni foam current collector via ultrasonic spray coating, and used the as-fabricated NF-CQD/Ni as the current collector of an LAB cathode. The NF-CQD/Ni-based cathode maintained remarkable cycling stability over 42 cycles (with a limited capacity was 1000mAh/g) and exhibited a superior speci ﬁ c discharge capacity of approximately 5121.91mAh/g (at 100 mA/g). NF-CQD/Ni-based cathode exhibits superior electrochemical properties, including its high cycling stability and high discharge capacity, primarily contributed by the NF-CQD coating layer, which prevented the oxidation of the Ni foam substrate and increased the electrical conductivity and electrocatalytic activity of the cathode. Therefore, NF-CQD/Ni can serve as an excellent current collector for LAB cathodes.


Introduction
The greenhouse emitted from burning fossil fuel is one of the primary reasons for global warming becoming increasingly serious. Therefore, considerable efforts have been dedicated to replacing fossil fuels with renewable energy sources [1][2][3]. Moreover, to meet the need for electric devices, the demand for energy storage devices with excellent safety, stable performance, and high energy density has increased [4,5]. Li-air batteries (LABs) present an excellent theoretical energy density (11400 Wh/kg) comparable to gasoline (13200 Wh/kg) and are environmentally friendly; therefore, they are expected to be remarkable energy storage systems [6][7][8][9]. However, LABs exhibit critical drawbacks, for instance, a high overpotential and low cycling stability, which restrict their practical application [10,11].
Typical LABs were comprised an electrolyte, a cathode, an anode, and a separator. Discharge-charge process of LABs can be described as follows [12][13][14]: Discharge process : oxygen reduction reaction Á ORR ; 2e -+ 2Li + + O 2 ⟶ Li 2 O 2 ð Þ Charge process : oxygen evolution reaction Á OER ; In particular, the low cycling stability of LABs is ascribed to the incomplete decomposition and accumulation of Li 2 O 2 at the cathode hindering the transfer of O 2 gas to the cathode. Therefore, to improve the cycling stability of LABs, the overpotential during charging should be decreased by improving the electrical conductivity and electrocatalytic activity of the electrode. Various studies have focused on increasing the electrocatalytic activity of LAB cathodes by using cathode materials with high electrical conductivity, such as carbonaceous active materials [15][16][17], metallic active materials [18,19], and electrolytes [20,21]. However, although the current collector plays a critical role by conferring stability and improving the conductivity and adhesion of the cathode, studies on current collectors have been scarce [22,23]. Therefore, surface engineering of current collectors to prevent oxidation and increase conductivity is critical for fabricating highly stable LABs with excellent cycling stability.
In this context, we coated a Ni foam substrate with carbon quantum dots codoped with N and F (NF-CQDs) and used the fabricated sample as the current collector of an LAB cathode. In addition, we estimated the effect of surface engineering on the cycling performances of LABs. Especially, the fabricated NF-CQDs presented large specific surface area, small molecular weight and particle size, high electrical conductivity, and remarkable chemical stability and were synthesized using a simple method. Furthermore, heteroatom doping of CQDs can increase the electrical conductivity and electrocatalytic activity of LAB cathodes. Moreover, the current collector fabricated by depositing a NF-CQD coating layer on the surface of a Ni foam current collector can enhance the cycling stability of LABs by preventing the oxidation of the Ni foam by the water vapor or O 2 in air. In addition, because the ultrasonic spray coating method generates fine droplets via the ultrasonic vibration of the spray nozzle, uniform, dense, and thin coating layers can be deposited on the surfaces of current collectors, unlike the coating layers deposited using other coating methods, such as conventional spray coating or dip coating [24][25][26].

Experimental
NF-CQDs were fabricated using a hydrothermal method. Specifically, 4 wt% of urea (Aldrich), 0.6 wt% of sodium fluoride (Aldrich), and 9 wt% of citric acid (Aldrich) were dissolved in deionized water, then moved to a Teflon-lined hydrothermal synthesis autoclave, and heated at 180°C for 6 h. The prepared solution was centrifuged three times at 10000 rpm during 25 min and then dialyzed using a 6-8 kD molecular weight cut-off membrane for 24 h to obtain uniform small-sized particles. The dialysate was dried in a 50°C oven overnight to obtain NF-CQD particles. Furthermore, to confirm the effect of F doping, N-doped CQDs (N-CQDs) were fabricated using the same procedure in the absence of sodium fluoride. Ni foam was coated with NF-CQDs using an ultrasonic spray coating method and a 1 wt% NF-CQD dispersion in N-methyl-2-pyrrolidinone (NMP). The feeding rate of the dispersion, spraying time, and frequency were set at 1 mL/h, 2 min, and 130 kHz, respectively. Pristine Ni foam, N-CQD-coated Ni foam, and NF-CQD-coated Ni foam (with a radius of 0.65 cm) are hereafter denoted as bare Ni, N-CQD/ Ni, and NF-CQD/Ni, respectively.
The element distributions, morphology, and structure of the NF-CQDs were determined by energy-dispersive spectroscopy (EDS; NEO ARM, JEOL) apparatus and a transmission electron microscopy (TEM) instrument. Moreover, the absorbance, crystal structure, chemical bonding structure, and surface functional groups of the N-CQDs and NF-CQDs were evaluated using an X-ray diffraction (XRD) instrument, an ultraviolet-visible (UV-vis) spectrophotometer, a Fouriertransform infrared (FT-IR) spectrometer, and an X-ray photoelectron spectroscopy (XPS; VG Multilab2000m) device, respectively. The thickness of the NF-CQD layer coated on the Ni foam substrate was measured using a scanning transmission electron microscopy (STEM; NEO ARM, JEOL) instrument; the morphologies and structure of the cathode material were analyzed using field emission scanning electron microscope (FESEM, EVO10, Carl Zeiss).
To estimate the electrochemical performance of the LAB cathode, we use coin cells (Hohsen Corporation, CR2032) comprising O 2 -electrode as a cathode and porous polypropylene membrane (Welcose, Celgard 2400) as the separator; tetraethylene glycol dimethyl ether (TEGDME) dissolves lithium bis(trifluoromethanesulfonyl)imide (LiTESI) as the electrolyte [27] and Li foil (Honjo Chemical) as the anode. The cathodes were fabricated by polyvinylidene difluoride (20 wt%, PVDF; Arkema) as the binder and Ketjen black (80 wt%, KB; ECP-600JD, Infochems Inc.) as the cathode active material. A mixture of KB and PVDF was dispersed in NMP (Aldrich) and then spray-coated onto bare Ni, N-CQD/Ni, and NF-CQD/Ni, which were used as current collectors, followed by drying in a 50°C oven overnight. The loading of KB was 0:08 ± 0:01 mg. The assembling of cells was carried out in an Ar-atmosphere glove box (99.999%), and the O 2 and H 2 O contents were maintained below 10 ppm. The assembled battery was aged for 24 h in an O 2filled battery jig; discharge-charge tests were conducted using a cycler system (WBCS3000M1) in the voltage range of 2.0-4.5 V (vs. Li/Li + ) at 25°C. Furthermore, cycling stability tests were performed with a specific capacity of 1000 mAh/g at a current density of 100 mA/g. Moreover, we used FESEM of bare Ni-, N-CQD/Ni-, and NF-CQD/Ni-based cathodes after the cycling testing to confirm the stability.  Figure 1(b) shows the ultrasonic spray coating process to preparation of the NF-CQDs coated on Ni foam. The CQDs were synthesized by hydrothermal method, including citric acid as a carbon source and urea as a nitrogen source, as shown in Figure S1. When heated, citric acid reacted with urea to form citric acid amide; then, further heating leads to deamination, dehydration, and dehydrocondensation of citric acid [28]. Sodium fluoride as a fluorine source and F atoms were incorporated into the honeycomb structure of carbon atoms [29]. Ultrasonic spray coating is an efficient method that can be used to produce fine droplets via ultrasonic vibration and deposit uniform and thin coating layers on various substrates [30]. In this study, an NF-CQD dispersion was finely coated on a Ni foam substrate via ultrasonic spray coating. Because the NF-CQDs contained numerous functional groups, the thin and uniform NF-CQD coating layer enhanced the interfacial adhesion between current collector (Ni foam) and active material (KB) [23,31]. Therefore, after the generating and decomposition of Li 2 O 2 during repeated dischargingcharging, KB, which served as the active electrode material, maintained the combined structure of the current collector. Moreover, during cycling, Ni foam was oxidized by the water vapor and O 2 in air and corroded by by-products (LiOH or Li 2 (CO 3 )), especially at high potentials (>4 V vs.

International Journal of Energy Research
Li/Li + ), which decreased the electrical conductivity of the current collector [32,33]. However, the NF-CQD coating layer efficiently prevented the oxidation and corrosion of the Ni foam substrate, thereby enhancing the stability of the KB-Ni foam interface [34,35].
The morphology and element distribution of the NF-CQDs were examined using TEM and EDS mapping. The low-and high-resolution TEM results of NF-CQDs (Figures 2(a) and 2(b)) revealed that the NF-CQDs werẽ 4.5 nm in diameter and their lattice fringe space was 0.32 nm, which corresponded to the (002) plane of graphite. To confirm the content distributions of C, O, N, and F, the TEM-EDS mappings of the NF-CQDs were obtained ( Figure 2(c)). The EDS data suggested that N and F were present in the NF-CQDs. The UV-vis spectrum and photographs of the NF-CQD dispersion are shown in Figure 2(d) and its inset, respectively. The NF-CQD suspension was grayish blue under natural light (left) and presented royal blue fluoresce under 365 nm UV light irradiation (right). UV-vis absorbance spectra of NF-CQDs included two feature peaks at 333 and 233 nm, which corresponded to n -π * transitions of the C=N and C=C bonds and the π -π * transitions of the aromatic sp 2 C atoms of the core domain, respectively [36,37]. By comparing the UV-vis spectrum of the N-CQDs and  International Journal of Energy Research NF-CQDs, it was determined that F doping did not change the characteristic absorption peaks of the CQDs ( Figure S2). These results confirmed that we have successfully synthesized NF-CQDs by hydrothermal method. The XRD was used to examine the structures and phases of N-CQDs and NF-CQDs. The broad peak at 26°in the XRD patterns of N-CQDs and NF-CQDs (Figure 3(a)) was ascribed to the (002) lattice plane of amorphous graphite (JCPDS 65-6212), and these results were consistent with the TEM results [38]. In the XRD results, it was confirmed that the crystal structure of the CQDs was maintained after F doping. Furthermore, the chemical bonding structures of N-CQDs and NF-CQDs were analyzed by XPS, and the results were calibrated with 284.5 eV C 1s binding energy (Figures 3(b)-3(f)). Figure 3(b) displays the XPS survey spectrum of N-CQDs and NF-CQDs having peaks at 282-292, 396-404, 528-535, and 682-690 eV which correspond to the binding energies of C 1s, N 1s, O 1s, and F 1s, indicating that N and F atoms have been successfully doped to CQDs. The C 1s XPS profile of the NF-CQDs included the characteristic peaks of C-F groups (~291.6 eV), O=C groups (~288.6 eV), C-N/C-O groups (~286.7 eV), and C-C groups (~284.8 eV) (Figure 3(c)). In contrast, the characteristic peak of C-F was absent from the C 1s XPS profile of the N-CQDs, indicating that the F atoms were only present in the NF-CQDs. The N 1s XPS profiles of N-CQDs and NF-CQDs (Figure 3(d)) included the characteristic peaks of pyridinic-N (~398.7 eV), pyrrolic-N (~399.9 eV), and graphitic-N (~401.4 eV). The presence of these groups was ascribed to the use of urea, a precursor of the N-CQDs, as the N source. The pyridinic-and pyrrolic-N presented two p-electrons, which originated from their aromatic π-systems and increased the electrical conductivity of the N-CQDs [39,40]. The O 1s XPS profiles of N-CQDs and NF-CQDs (Figure 3 3 eV). Furthermore, the F 1s XPS profiles presented two peaks at~687.2 and~685.9 eV, which corresponded to the covalent and semi-ionic bonds of C-F ( Figure 3(f)). These groups were not observed in the F 1s XPS profile of the N-CQDs, indicating that F atoms were only present in the NF-CQDs. The electronegativity of F (4.0) is higher than that of C (2.5); accordingly, the electron/ion conductivity of the NF-CQDs increases because the polar C-F bonds presented a dipole moment [23]. Therefore, owing to F doping, the C atoms of the NF-CQDs became positively charged, and the activity of the π-electrons of the NF-CQDs was higher than that of the π-electrons of the N-CQDs, causing the higher electrocatalytic activity of NF-CQDs than N-CQDs [39]. Consequently, the NF-CQD coating layer increased the electrical conductivity and electrocatalytic activity of the LAB cathode, thereby promoting the decomposition and formation of Li 2 O 2 during charging and discharging [41].
Chemical properties of functional groups on the surface of N-CQDs and NF-CQDs were determined by FT-IR spectrum (Figure 4 [42][43][44]. The characteristic peak of the C-N bonds in the FT-IR spectroscopy of the NF-CQDs confirmed that the CQDs were doped with N atoms. Moreover, the presence of the characteristic peaks of the C=N and C=C bonds indicated that the hydrothermally synthesized NF-CQDs presented a hetero-polyaromatic structure. The characteristic peaks of C-F 2 and C-F in the FT-IR spectrum of the NF-CQDs suggested that the CQDs were doped with F atoms. These results were consistent with the aforementioned XPS spectra. Furthermore, STEM results demonstrated that the NF-CQD coating layer was uniform and thin (26.5-28.5 nm) (Figure 4(b)). Because the NF-CQDs surface exist a great quantity of oxygen containing functional groups, the NF-CQD coating layer strongly interacted with the Ni foam Furthermore, the morphology of the Ni foam substrate did not change after N-CQD and NF-CQD coating via ultrasonic spraying ( Figure S3). Moreover, the uniform and thin CQD coating layers effectively prevented the oxidation of Ni foam during cycling. Therefore, these findings confirmed that the NF-CQD coating layer enhanced the cycling stability of LABs.
To assess the electrochemical catalytic activities of the bare Ni, N-CQD/Ni, and NF-CQD/Ni electrodes, LSV measurements were carried out ( Figure 5). LAB cathodes involve two key reactions: ORR and OER. They all involve multistep electron transfer processes with inherently slow kinetics [45]. In the ORR polarization curve region, the onset potential of NF-CQD/Ni (0.887 V) was higher than that of other electrodes (N-CQD/Ni: 0.81 V and bare Ni: 0.806 V), imply-ing that the NF-CQD/Ni has high ORR catalytic activity. The limiting current density of NF-CQD/Ni-based cathode (-4.755 mA cm -2 ) was superior then that of bare Ni and N-CQD/Ni-based electrodes (-2.256 and -3.985 mA cm -2 , respectively) as shown in Figure 5(b). This result was ascribed to that F atoms can improve the ionic/conductivity of the interfacial layer. In the OER polarization curve region, the NF-CQD/Ni-and N-CQD/Ni-based electrodes showed the superior negative E onset , exhibiting excellent OER catalytic performance. During the discharge-charge process of LABs, the ORR and OER electrochemical kinetics are crucial; they are generally evaluated by the potential gap (E OER−ORR ) between OER (at 10 mA cm -2 ) and ORR (at -2 mA cm -2 ) [46]. The NF-CQD/Ni-based and N-CQD/Nibased electrodes showed a potential difference of 1.07 V,  International Journal of Energy Research which may be due to the efficient electron transport due to the oxygen reaction promoted by N doping.
To assess the cathode electrochemical performance, we fabricated cathodes by coating KB as the active material on bare Ni, N-CQD/Ni, and NF-CQD/Ni current collectors. The initial constant current discharge-charge curves of the fabricated cathodes are displayed on Figure 6 over a current density of 100 mA/g in the voltage range of 2.0-4.5 V. The initial specific capacities of the bare Ni-based cathode (2917.97 mAh/g) were lower than that of the N-CQD/Ni-and NF-CQD/Ni-based cathodes (3963.52 and 5121.91 mAh/g, respectively). Furthermore, the discharge-charge voltage gap of the NF-CQD/Nibased cathode (1.603 V at 2000 mAh/g) was narrower than that of the bare Ni-based cathode (1.771 V at 2000 mAh/g). This result was ascribed to the higher ORR and OER electrocatalytic activities of the NF-CQD/Ni-based cathode than the bare Ni-based cathode during charging, which promoted the efficient decomposition of Li 2 O 2 . The cycling stability of the fabricated cathodes was assessed at a limited capacity of 1000 mAh/g, at a current density of 100 mA/g, and in the voltage range of 2.0-4.5 V (Figure 6(b)). The cycling stability of the NF-CQD/ Ni-based cathode (42 cycles) was superior to those of the bare Ni-and N-CQD/Ni-based cathodes (13 and 31 cycles, respectively). During the discharging-charging test, Li 2 O 2 formed continuously as the product of the discharging reaction and decomposed at the cathode surface, thereby lowering the cycling stability of the LAB. In particular, the cycling stability of the LABs featuring the NF-CQD-based cathode was higher than those of the other LABs because F doping caused the electrochemical catalytic activity and electrical conductivity of the electrode to increase, thereby inducing the efficient decomposition of Li 2   International Journal of Energy Research cathodes, the CQD coating layers suppressed the oxidation of the Ni foam substrates, which improved the cycling stabilities of the corresponding LABs. Moreover, compared with other reported electrodes without carbon coating, the NF-CQD/Nibased cathode exhibits excellent performance in terms of specific capacity and cycle stability (Table S1) [47][48][49][50].
The discharge-charge voltage curve of bare Ni-, N-CQD/ Ni-, and NF-CQD/Ni-based cathodes are shown in Figure 7. The bare Ni-based cathode exhibited a stable dischargingcharging voltage gap during the first five cycles (Figure 7(a)). However, after 10 cycles, the charging overpotential of the bare Ni-based cathode increased slowly, causing a decrease in discharge capacity. In contrast, the N-CQD/Ni-based cathode presented a stable voltage gap for 25 cycles. This was attributed by the increase in the conductivity of the N-CQD coating layer induced by N doping, which promoted the ORR and OER at the cathode. Furthermore, the NF-CQD/ Ni-based cathode presented excellent cycling stability and a stable voltage gap for 30 cycles. These results were ascribed to the F doping of the NF-CQD coating layer; the electrocatalytic activity of the cathode materials is increased by increasing the activity of the π-electrons of N-CQDs. Therefore, for the bare Ni-based cathode with a low electrocatalytic activity, Li 2 O 2 was only partially decomposed, and it accumulated at the surface of the cathode, causing a decrease in its electrical conductivity. Thereby, the ORR and OER at the bare Nibased cathode were disturbed, and the cycling stability of the cathode was poor. However, the NF-CQD coating layer prevented the oxidation of the Ni foam substrate and promoted the efficient decomposition of Li 2 O 2 by increasing catalytic activity and conductivity efficiency of electrodes, thereby increasing the cycling capability of the cathode [51,52].
To compare the cycling stabilities of the electrodes after the discharge-charge tests, we analyzed the FESEM of the bare Ni-, N-CQD/Ni-, and NF-CQD/Ni-based cathodes before first discharging, after first discharging and charging, and after the cycling stability tests (Figure 8). Before first discharging, bare Ni-, N-CQD/Ni-, and NF-CQD/Ni-based . The products which were formed at first discharging process on the bare Ni cathode were peony flower-like (Figure 8(b)), but they were not completely decomposed after charging (Figure 8(c)). Compared with bare/Ni, the particles on the surface of N-CQD/Ni-and NF-CQD/Nibased cathodes did not exhibit peony flower-like morphology after first discharging, and they showed the decomposed discharge products and decreased particle size after first charging (Figures 8(f), 8(g), 8(j), and 8(k)). Also, the particles on the surface of N-CQD/Ni-and NF-CQD/Ni-based electrodes were sticky compared to bare Ni electrodes after charge-discharge tests because the N-CQDs and NF-CQDs coated on Ni foam enhance cycling stability by the strong connection between the current collector and electrode. The discharge product of LABs is Li 2 O 2 ; however, the X-ray energy of Li (52 eV) is too small to be distinguished by EDS, so we analyzed the percentages of the O element to confirm the discharge product content on the surface of electrodes (  [53,54]. Therefore, owing to the poor ORR and OER activities of the bare Ni-based cathode, the peony flower-like Li 2 O 2 particles grown on its surface caused its cycling stability to decrease by blocking its pores and decreasing its active sites. The Li 2 O 2 particles that accumulated at the N-CQD/Ni-based cathode surfaces also presented a peony flower-like structure; however, the 2-D Li 2 O 2 sheets were shorter and smaller than those grown on the surface of the bare Ni-based cathode (Figure 8(h)). Moreover, the Li 2 O 2 particles deposited on the NF-CQD/Ni-based cathode were spherical with bump surface (Figure 8(l)). The contents of discharge products on surface of electrodes were determined by analyzing the atomic percentage of O element. As shown in Table 1 particles on the gas transfer at the cathode was lower than that of the 2-D Li 2 O 2 sheets; therefore, the NF-CQD coating layer enhanced the cycling stability of the NF-CQD-based cathode. Therefore, the NF-CQD/Ni-based cathode exhibits highly stable electrochemical performance, which is ascribed to the NF-CQD layer coated on the Ni foam substrate. The NF-CQD/Ni-based cathode exhibits remarkable cycling stability due to the following: (a) the NF-CQD coating layer prevented the oxidation of the Ni foam substrate during discharging-charging, and (b) the NF-CQDs increased the electrical conductivity and electrocatalytic activity of the cathode, facilitating the efficient decomposition of Li 2 O 2 .

Conclusion
In this study, we fabricated a NF-CQD/Ni-based cathode for highly stable LABs using a hydrothermal method followed by ultrasonic spray coating. The NF-CQD/Ni-based cathode displayed an excellent initial discharge capacity of 5121.91 mAh/g at a current density of 100 mA/g, a low discharge-charge voltage gap of 1.603 V at 2000 mAh/g, and outstanding cycling stability over 42 cycles (with a limited capacity of 1000 mAh/g). These results are attributed to the effect of the NF-CQD coating on the Ni foam substrate as follows: (i) the NF-CQD coating layer prevented the oxidation of the Ni foam matrix and (ii) co-doping with N and F improved the electrocatalytic activity and electrical conductivity of the cathode. We believe that coating NF-CQD layers on current collectors can be used to fabricate highly stable LABs.

Data Availability
The data used to support the findings of this study are included within the manuscript and the supplementary information files.  Figure S1: the kinetic mechanism of the formation of CQDs. Figure S2: UV-vis absorbance spectrum of N-CQDs. Figure  S3: FESEM images of (a, d) bare Ni, (b, e) N-CQD/Ni, and (c, f) NF-CQD/Ni. Table S1: specific capacity and cycle stability of without carbon coating electrodes reported previously. (Supplementary Materials)