Femtosecond laser irradiation as a novel method for nanosheet growth and defect generation in g-C3N4

Among the many recently developed photo-catalytic materials, graphitic carbon nitride (g-C3N4) shows great promise as a catalytic material for water splitting, hydrogen generation, and related catalytic applications. Herein, synthesized bulk g-C3N4 is simply irradiated under a 35 fs pulse at mixed photon energies (800 nm and its second harmonic). g-C3N4 was synthesized from melamine following a facile thermal polymerization procedure. The prepared material was introduced, in an aqueous environment, to the femtosecond laser for various lengths of time. The treated material demonstrates a significant increase in surface area, relative to the untreated samples, indicating that irradiation is a successful method for exfoliation. The subsequent characterization reveals that the mixed irradiation process drives significant defect generation and sheet growth, which is not seen under 800 nm irradiation. Extended mixed irradiation results in 4 nm thick nanosheets with lateral dimensions 4× that of the bulk material. The treated material shows improved dye absorption/removal. This novel method of defect generation and nanosheet growth shows great potential as a g-C3N4 pre-treatment method for co-catalytic applications. Herein it is shown that femtosecond laser irradiation drives exfoliation beyond 100 nm particle sizes, and sheet-like morphologies under extended irradiation, which must be taken into account when using this method to improve material performance.


Keywords: gC 3 N 4 , femtosecond laser irradiation, catalysts, nanosheets
(Some figures may appear in colour only in the online journal)

Introduction
Over the past decade, promising semiconductor materials have been developed to solve the myriad challenges associated with improving catalytic performance, especially with respect to environmentally significant applications, such as water-splitting and sustainable hydrogen generation for solar fuel applications [1]. Notably, graphitic carbon nitride (g-C 3 N 4 ) has gathered significant interest as a cost-effective, accessible, non-toxic, and stable semiconductor with desirable electronic properties. g-C 3 N 4 is synthesized via the thermal condensation of melamine, which is a commodity chemical and low-cost precursor. Once condensed, g-C 3 N 4 is considered non-toxic and non-hazardous [2]. As a semiconductor, the band edge positions (conduction band minimum of −0.5 to −1.5 eV and a valance band maximum of 1.3-2.0 eV versus normal hydrogen electrode (NHE)) are appropriately set for hydrogen evolution (H/H 2 O @ 0.00 eV versus NHE), oxygen reduction (O 2 /O 2 − @ −0.35 eV versus NHE) and water splitting (O 2 /H 2 O @ 1.23 eV versus NHE) [3][4][5][6][7][8]. The resulting band gap of 2.7 eV lies in an acceptable range for visible light excitation, hence g-C 3 N 4 has been a promising candidate for next-generation photo-catalytic materials. g-C 3 N 4 nanosheets have shown promising applications in a variety of fields including bio-processing, hydrogen production, water splitting, and wastewater remediation [9][10][11][12][13][14]. Despite these advantages, the active sites in g-C 3 N 4 possess poor intrinsic catalytic activity; an issue traditionally solved with assorted metallic co-catalysts, hetero-junctions for charge separation, or exfoliation methods designed to improve the number and accessibility of the active sites [15][16][17][18][19].
The first step to producing highly functional g-C 3 N 4 systems is to appropriately prepare and exfoliate the material itself. Exfoliation is necessary to separate the bulk g-C 3 N 4 material into nanosheets, flakes, or engineered structures to fully take advantage of the high surface area and unique engineering considerations native to 2D materials. A seemingly endless array of options for preparation and exfoliation exist, including poly-condensation, poly-addition, solgels, and template-based preparations [5,20]. These preparation methods begin with a carbon nitride precursor (urea, thiourea, melamine, etc), and produce process-specific morphologies and properties. Exfoliation methods include thermal oxidation [21], ultrasonic treatments [22], and chemical methods [23]; all of which drive significant alterations to the surface area, porosity, and/or active sites available on the base material, depending on the method [15,24,25]. Meanwhile, laser irradiation offers an exciting pathway to customize 2D materials for next-generation engineering solutions due to the potential structural adaptations, functional group additions, and utility in co-catalytic/hetero-atomic systems [26][27][28][29][30]. These advantages dovetail nicely with many of the challenges associated with g-C 3 N 4 applications as high-intensity, pulsed laser irradiation generated at precise photon wavelengths has proven to be effective for exfoliation, domain growth, and defect generation [14,31] in diverse materials, under both nanosecond [11,12,32] and femtosecond pulse durations [3,13,[33][34][35]. Wu et al performed the first femtosecond pulsed laser irradiation (λ = 800 nm) of g-C 3 N 4 , which resulted primarily in exfoliation and some secondary defect generation [3].
Inspired by these results, this work details the irradiation from both fundamental and second harmonic femtosecond laser pulses (fundamental pulses, λ = 800 nm, and its second harmonic, λ = 400 nm) for domain growth and defect generation in g-C 3 N 4 nanosheets while simultaneously exfoliating the bulk particles. In this environment, g-C 3 N 4 undergoes extreme periods of both kinetic and electronic excitation, which drives a unique growth mechanism in g-C 3 N 4 nanosheets.

Synthesis of bulk g-C 3 N 4
Melamine (99% pure, Sigma-Aldrich) was used without further purification. Thermal polymerization was performed in 3 g batches using a ceramic crucible, fully covered with a secondary ceramic cover, in a box furnace (Lindberg Blue M). The crucible was heated from ambient conditions, in air, at a rate of 15°C min −1 to 550°C and held at that temperature for 2 h, then allowed to cool. This produced a yellowish crust, which was subsequently ground to a powder with a mortar and pestle and dissolved in DI to form a 1.66%(w/v) solution. This solution was ultrasonicated in an ice bath for 2 h at 150 W (pulsed; 2 on, 5 off) to produce the g-C 3 N 4 solution.
2.2. Laser irradiation of bulk g-C 3 N 4 1.5 ml of the g-C 3 N 4 solution was irradiated under 35 femtosecond, 1 mJ pulses from an amplified femtosecond laser (Spitfire Ace by SpectraPhysics at 1 kHz), passed through a type BBO (Beta Barium Borate) nonlinear crystal. The combined fundamental and second harmonic radiation which resulted in 19:1, 800, and 400 nm, available in the supplementary information figure S.1, was focused using a 5 cm focusing lens to a spot size of 20 μm, as shown in figure 1. Irradiation was held for n minutes, where n ranges from 0, through 15, 30, 60, 90, and 120 min, the resulting samples are henceforth denoted as 'CN0', 'CN15', 'CN30', 'CN60', 'CN90', and 'CN120', respectively. Separately, irradiation of the solution was performed without any second harmonic wavelengths (λ = 800, 400 nm is absent) for 120 min, that sample is henceforth denoted as '800CN'. Finally, sample CN120 was re-characterized after aging for 30 d to investigate the long-term stability of the changes in the irradiated g-C 3 N 4 , resulting re-characterizations are hereafter denoted as 'Aged'.

Characterization
The pre-exfoliated g-C 3 N 4 (CN0) was used in all laser treatments and adapted for various characterizations. The solutions were diluted with dimethylformamide (DMF, Sigma-Aldritch) at a rate of 1:10 and then deposited onto commercially available silicone wafers via the Langmuir-Blodgett method for atomic force microscopy (AFM, Veeco-Bruker Multimode 8 AFM) imaging to capture any significant morphological and sheet thickness changes. Particle size measurements were taken over a minimum of 10 sheets, using the maximum for height and length, utilizing Gwyddion. Scanning electron microscopy (SEM, ZEISS Ultra Plus) and EDS (EDAX Apollo XL-SDD) was used on those same depositions to determine the morphological and chemical composition of the irradiated samples. The surface area was determined using the Brunauer-Emmett-Teller (BET) method on vacuum-dried powders to gauge the exfoliation effectiveness (Micromeritics Gemini 2390a). The Fouriertransformed infrared spectra (FT-IR) was taken on the same set of powdered samples to determine the presence of significant functional groups and to ensure the g-C-3N-4 fingerprint is present (NEXUS 670 FT-IR ESP). Chemical bonding was determined using x-ray photoelectron spectroscopy (XPS) to assess the influence that irradiation has on the g-C 3 N 4 bonding (Thermo ESCALAB 25, Al-Kα x-ray source). The deconvolution parameters can be found in figure S.2. A Shimadzu UV-2501PC spectrometer was used to collect the UV-vis spectra via diffuse reflectance spectroscopy. Samples were prepared by drop-casting and vacuum drying the solutions onto transparent substrates (UV transparent acrylic). Absorbance was calculated from the reflectance data according to equation (1). The data was transformed using the Kubelka-Munk approximation, equation (2), to find the band gap

Photo-catalytic characterization
Methylene blue (MB) dye degradation was performed to measure the photocatalytic activity of the irradiated samples and to compare the results to the bulk material. The procedure was standardized between the samples, with 5 mg of dried, powdered g-C-3N-4 per 15 ml of MB standard (with a known concentration of 5 mg l −1 ). The solution was agitated with a magnetic stir bar (300 rpm) in a dark environment for 2 h. Afterward, the solution was exposed to UV illumination under a 6 W, 365 nm light source (Analytik Jena). At predetermined time steps, samples were removed and centrifuged at 3000 rpm for 20 min, then the absorbance spectra were taken (Shimadzu UV-VIS-NIR, UV-2600i) to establish an initial concentration drop and the MB removal rates. The change in concentration was determined by taking the spectral intensity at 664 nm and applying the Beer-Lambert law, , and l is the optical path length (cm) The resulting concentrations were then converted to a relative percent reduction in MB for each catalytic species, while first-order fits were used to determine the dye removal rates. The initial absorbance drop is shown in figure S.3.

Laser-induced morphological changes
AFM images are shown in figures 2(a)-(f). It is clear that the initial bulk preparation has large particles of material, while sheets emerge with increased femtosecond irradiation. The CN0, CN30, CN60, CN90, CN120, and 800CN particle heights (the approximate volume can be found in S.4) range from 2.3 μm ± 230%, 54 nm ± 150%, 104 nm ± 70%, 88 nm ± 90%, 10 nm ± 120%, and 42 nm ± 100% respectively. The sheet sizes demonstrate a large initial variation, which is preserved during the irradiation, however, the average significantly increases with irradiation. Variations in sheet size may even increase with irradiation. Comparing particle sizes in (a)-(c) with (d) reveals the initial particles are broken and reordered into 5 nm thick sheets. The aspect ratio figure 2(g) matches these observations, where a sharp increase is observed after a critical duration of laser irradiation. This suggests that the first hour of irradiation breaks the bulk g-C 3 N 4 into smaller sheets or particles, which later reorder to larger nanosheet or nano-flake-like structures under additional irradiation. The BET surface area of the materials in dry powder form as a function of irradiation time is shown in figure 2(g). A rapid initial surface area increase is observed which plateaus after 30 min. The increase in BET surface area provides further evidence that exfoliation occurs but the extent may be difficult to assess due to potential re-stacking or aggregation of sheets upon drying. It is clear that mixed kinetic and electronic excitation is more effective for generating high surface area sheets relative to pure kinetic excitation; CN120 shows a two-fold increase relative to the 800CN material. However, other studies covering the production of monolayer g-C 3 N 4 report specific surface areas well above 100 m 2 g −1 [15,23,36].
These results suggest that the laser irradiation separates bulk g-C 3 N 4 via kinetic excitation of atoms and/or absorbed water in the target sheets, driving sufficient physical layer separation for exfoliation, similar to laser exfoliation processes for other 2D materials commonly seen in nanosecond [32] and femtosecond [13,34,35] studies. However, there is a lateral growth mechanism that dominates the system once the bulk has been sufficiently exfoliated, best illustrates in the extended sheets seen in figure 2(e). This is likely related to the number of stacked sheets interacting with any particular irradiation event; fewer sheets may be more difficult to exfoliate, however, these limitations do not seem to interfere with the sheet growth. Notably, this growth mechanism is absent in irradiation lacking electron excitation (800CN, figure 2(f)), indicating that resonant single and, possibly, multi-photon effects (at 400 nm) must be playing a part in this mechanism.

Laser-induced chemical changes
Chemical species changes are expected under powerful irradiation, especially under conditions where the irradiation photon energy is sufficient to electronically excite the material, creating the potential for photochemical versus purely ablative processes. Individual photons at 400 nm and 800 nm have energies of 3.1 and 1.5 eV, respectively (S.5), which is sufficient to electronically excite g-C 3 N 4 , generating free electron/hole pairs that can participate in reordering and redox-based events [37,38]. To fully examine the chemical changes occurring in the material it is helpful to identify the current species and potential defect locations/species. Figure 3(a) illustrates the predominant sites of interest. Two carbon species (blue) are initially present in g-C 3 N 4 ; an intratri-s-triazine carbon (N-C=N) and a terminal carbon which may have irregular functional remainder groups on edges and defect locations. Nitrogen (green) is similar as only three species are expected in pristine g-C 3 N 4 ; a central (N-C 3 ), edge (C-N=C), and defective variant (N-H 2 ), which may also have irregularities. Defective species (red) can exist in many forms, however, under irradiation, only solvent interactions or reordering of existing molecules is expected. Finally, intra-sheet defects can exist, which may offer additional sites for adsorbed species or functionalization (purple) [3,26,28,[39][40][41].
The UV-vis absorbance spectra were used to determine the band gaps derived from equation (2). CN0 to CN90 share a band gap between 2.8 and 2.84 eV, while C120 seems to break down into two absorptive regions, with band energies of 2.65 and 3.04 eV respectively. This behavior suggests that, at some critical point, the electronic structure is so damaged by defects that it reorders into states no longer indicative of g-C 3 N 4 . These results are available in figure S.6.
FT-IR provides insight into the functional groups being introduced during the irradiation process. The spectra obtained for the various materials can be seen in figure 3. The g-C 3 N 4 fingerprint consists of a broad peak at 3000-3500 cm −1 associated with the N-H stretching mode, C-N and C=N stretching modes in the tri-s-triazine aromatic ring from 1200 to 1775 cm −1 and a sharp peak at 808 cm −1 related to the triazine breathing modes [3,42]. A peak at 2180 cm −1 arises with increasing irradiation time, evolving into a There is a clear increase in exfoliated sub-100 nm particles with irradiation time, which then form larger 'sheets' between 5 and 10 nm thick after 120 min. The increase in (g) aspect ratio (width/length) illustrates the conversion of particles into sheets, while (h) the fitted BET surface area shows the trend in exfoliation. Samples CN0-CN60, and are comparable to 800CN morphologically, although the size of the particles decreases. This is in contrast to the development of sheets observed in (d), (e) CN90 and CN120.
broader, sharper peak in CN120. This is matched by the evolution of a second aromatic breathing mode at 760 cm −1 , which indicates the presence of broken tri-s-triazine units, resulting in semi-isolated aromatic structures [43][44][45]. Additional reordering is apparent in the C-N and C=N fingerprint regions, highlighted in the CN120 spectra. These shifted peaks match with the NO and NO 2 fingerprints; NO 2 has asymmetric, symmetric, and scissor breathing modes at 1527 cm −1 , 1350 cm −1 , and 880 cm −1 respectively, while NO has asymmetric and symmetric modes at 1540 cm −1 and 1375 cm −1 [46]. This strongly indicates that after some critical degree of defect generation, nitrogen will reorder into various oxidative states with the surrounding aqueous environment. Similar events will occur with carbon, generating C-O/C=O, and some weak fingerprints were detected for these groups, notably C-O stretching at 1200 cm −1 and a potential weak shoulder at 1777 cm −1 associated with C=O stretching [47]. This suggests that a secondary oxidative process arises under extreme irradiation environments. A final peak between 2170 and 2180 cm −1 arises in the triple bond region, associated with the asymmetric stretching mode of cyano-groups (C≡N) [48]. This peak clearly develops with irradiation time, indicating that it may be the primary initial defect until oxidative effects become significant. Retesting the FT-IR spectra of the Aged sample ( figure 3(c)) indicates that the oxidative species are no longer apparent, and the functional groups present more closely resemble the initial CN0 spectra, with the exception of the C≡N groups. The C≡N signal intensity has clearly decreased but still appears much more significant than in CN0. This suggests that there is a limit to the number of C≡N groups that may be integrated into g-C 3 N 4 before the material becomes unstable. As a degree of C≡N groups are still present after aging, it is likely that some defect species are stable on intra-sheet pores or sheet edges. None of these functional group alterations are observed in the 800CN sample; the only difference between CN0 and 800CN seems to be the signal intensity, which may be an artifact of the FT-IR process or due to the increased sheet exfoliation.
The XPS spectra, figure 4, provide some of the greatest insights into the changes occurring under the laser exfoliation process. HR-XPS fits of the C 1s, N 1s, and O 1s spectral regions in the CN0, CN120, and the Aged samples are compared, while an overview of the remaining samples and fitting methodology can be found in figure S.2.

C 1s
The C 1s spectra, seen in figures 4(a), (b), have been deconvoluted to identify four significant peaks, additional deconvolutions are available in figure S.2. These are related to the adventitious carbon peak overlapped with any C-O bonds in the g-C 3 N 4 sheets themselves (C-C, C-O), the tri-s-triazine terminal carbon shown in figure 3, the edge carbon constituting the primary species in g-C 3 N 4 , and a carbonate related peak (associated with both the adventitious carbon and g-C 3 N 4 ) [49][50][51][52]. Higher binding energy shake-up satellite peaks, due to insulated domains charging at different rates, manifest above 290 eV [53,54]. The g-C 3 N 4 C 1s peaks have been identified at 284.8, 286.26 ± 0.17, 287.86 ± 0.17, and 289.22 ± 0.24 eV for samples from CN0 to CN120 and Aged. These are associated with adventitious carbon, the terminal carbon bonded to an amine (C-NH 2 ), the edge carbon (N-C=N), and the carbonate-related peak (N-C-O) respectively [27,50,55]. Carbon peak shifts are relative to the adventitious carbon, illustrated in figure S.2. C≡N could not be deconvoluted from the C-N peak [31,49,50,56]. The FWHM, peak positions, and peak assignment details are Figure 3. The structure of melamine and g-C 3 N 4 in (a), blue and green denote carbon and nitrogen species intrinsic to the material while red highlights possible defect species, and purple illustrates preferred metal doping sites. The FT-IR spectra are compared between (b) the respective irradiation times, regions of interest indicated in CN120, and (c) the CN0, CN120, Aged, and 800CN samples, illustrating the instability intrinsic to the highly defective material. available in figure S.2. It is clear that, with laser irradiation time, the trends in the NCO and C-NH 2 bonds increase, then drop off after CN90 to CN120, matching the FT-IR trends. This comes at the expense of the N-C=N bonds, indicating that these bonding states reorder to generate defects in the carbon bonding in the material. The decrease from CN0 to CN60 and increase from CN60 to CN120 indicate that this reordering is preferable over gaseous decomposition or another chemical loss mechanism. A 30% drop is observed between CN0 and CN60, which is nearly fully recovered by CN120; this indicates that carbon bonding in g-C 3 N 4 has an induced defect limit, at which unstable domains will reorder to the initial state under irradiation, rather than continuing to develop defects. This reordering and apparent structural recovery are likely associated with the domain growth mechanism seen in the AFM analysis. Notably, the NCO peak has the largest shift between CN0 and CN120, suggesting that the carbonate defect/oxidative carbon states are more stable and less inclined to revert than the C-N reordered states. There are at least two competitive processes occurring; increased irradiation time leads to exfoliation which reveals new active surface area. This leads to the surface-sensitive XPS detecting a relative increase in fully integrated tri-striazine units as the material bulk becomes exposed. Conversely, the irradiation introduced new defective states, and a rise in the correlated defective bonds will be apparent. This may explain why the C-NH 2 (terminal defects) tends to remain constant between CN0 and CN30; terminal defects may be more unstable, and may more easily bond with one another, resolving into nanosheets.
The stability of the samples can be compared by looking at the Aged samples. The adventitious carbon, C-NH 2 , N-C=N, and NCO peaks have area ratios of 45.11, 11.92, 30.12, and 12.85%, respectively. This is a significant increase relative to the CN120 sample and is likely related to absorbed surface species present in the aging environment. Oxidation drops in the Aged sample to the lowest level among all the samples, as does the N-C=N, while the defect-related C-NH 3 species is higher than any purely laser-irradiated sample. This suggests that oxidized carbon atoms do not reorder to N-C=N with time, but will resolve as more thermodynamically stable sp 2 /sp 3 species.

N 1s
The N 1s peak can be deconvoluted into four species as seen in figures 4(b), (d). These are identified as the edge bonds between carbon and nitrogen in the tri-s-triazine units (C-N=C), the central bonds between carbon and nitrogen in the tri-s-triazine units (N-C 3 ), some edge or terminal defects present throughout the g-C 3 N 4 sheets (N-H 2 ), and various convoluted nitride bonds (NO x ), which are beyond the resolution of the system and are therefore best fitted with a broad peak. These are situated at 398.55 ± 0.38, 399.93 ± 0.41, 401.16 ± 0.47, and 402.57 eV ± 0.51 eV respectively. The C≡N bond cannot be meaningfully distinguished from the C-NH 2 peak, attempts to introduce a new peak at 399.5 ± 0.6 eV resulted in poor or overly convoluted fits [31,49,56]. As the only source of nitrogen in the system is from the melamine precursor, any species changes must be due to the irradiation-driven reordering of g-C 3 N 4 . It is apparent that the irradiation drives an initial increase in the N-H 2 peak, which tails off at longer irradiation times, matching the behavior seen in the C 1s C-NH 2 peak. An inversely reciprocal relationship is clear between the N-H 2 and C-N=C species; this suggests that the irradiation damages the C-N=C bonds, which reform as defective edges, again agreeing with the C 1s results. This behavior appears to be largely independent of the central N-C 3 peaks, which show a small but steady decrease, indicating that the defective generation mechanism is focused on the edges of the tri-striazine units, as illustrated in figure 3 and collaborated in the C 1s deconvolution. Although the slight decrease in the N-C 3 species suggests that irradiation will reorder N-C 3 bonds, the steady decrease suggests that if a defect is generated on the central N-C 3 bond, it is permanent and will not reorder into a stable g-C 3 N 4 system. The NO x bonds follow an unusual trend; from CN0 to CN90 these species match the defect patterns seen in the N-H 2 and C 1s species, however, N-H 2 in the CN120 sample increases sharply. This could be explained by an oxygen transfer from carbon to nitrogen with severe irradiation. However, it seems more likely that domain growth events stabilize NO x bonds, or are facilitated by these structures. This would suggest that oxidized nitrogen states are greatly preferred to oxidized carbon species. The aging comparison in the N 1s species shows that C-N=C, N=C 3 , N-H 2 , and NO x reorder with time to 65.54, 24.03, 9.78, and 0.65%, respectively (figure S.2). The C-N=C peak nearly doubles from the C120 sample, and increases by 25% from CN0; this indicates that edge nitrogen species in g-C 3 N 4 will reorder with time from defects to the original orientation, potentially even removing preexisting terminal defects (from the polymerization process) in favor of maintaining tri-s-triazine units. The N-C 3 bond remains nearly the same as in CN120, slightly less than seen in CN0, supporting the prior indicators for high stability at the central nitrogen site. NO x is nearly absent after the aging, indicating that oxide bonding on edge nitrogen is not a stable defect over longer time spans, as it appears to either break off or reorder.

O 1s
Due to the nature of the oxidative status in the material and signal-to-noise ratio, deconvolution beyond a set of peaks is not interpretively significant in the O 1s spectra. In that regard, a fit of two peaks loosely correlated to NCO and NO x must be performed; these peaks are centered at 531.9 ± 0.4 and 533.9 ± 0.75 eV. They follow the same trends as the other high-resolution scans; nitrogen oxide species increase with exfoliation while the carbonate species increase with defect generation. The nitrogen oxide species may increase as well with defect generation, but if so, it is outweighed by the oxidation of carbon species. Aging heavily favors the NCO species, where NCO and NO x reorder to 66.5 and 33.5% respectively (figure S.2). This seems to indicate that the aging process permits oxidation of defect centers on carbon species in g-C 3 N 4 , while the nitrogen species either reintegrate into the sheet, resolve as N-C 3 during domain growth, or are broken off as independent species, driving the decreased nitrogen ratio seen in the elemental composition results in the figure S.7.

Survey
The overall survey results indicate that for CN0 through CN120, the ratios (relative to the sum fitted area) of C 1s, N 1s, and O 1s bonds remain constant at 45.6 ± 3.7%, 51.1 ± 4.5%, and 3.4% ± 23.5% respectively. This indicates that there is relatively little volatile species loss or oxygen integration. However the Aged sample dramatically alters these trends; the C 1s, N 1s, and O 1s jump to 64.05, 16.77, and 19.18 respectively (figure S.7). This indicates that there is a relative loss of 66% in the N 1s bonds, taken over by O 1s species. This suggests that there are substitution and/or removal reactions between the defective nitrogen species and ambient oxygen. This might occur at the C≡N defect site, which would correlate to aging effects in the FT-IR spectra. It cannot be ruled out that this change is due to environmental contamination or adsorption.
It must be noted that the Aged sample very nearly reverted chemically to match CN90, which is more similar to CN0 than CN120. This indicates that the chemical processes observed are short-term and, should they be desired in an application, must be additionally stabilized. This does not hold for the morphology of the material; aging the samples shows some agglomeration which was trivially circumvented via bath sonication (30-60 min) in all samples with the exception of 800CN, which shows agglomeration (see figure 2) following a similar 60 min bath sonication. Generally, the solutions were observed to settle within an hour, while suspensions in DMF would remain stable for greater than 8 h.

Photo-catalytic results
The photocatalytic/absorptive results are plotted in figure 5, where C/C 0 is calculated as the percent decrease in concentration relative to the initially measured concentration (select spectral data available in S.7). Solution pH was measured at 7 ± 0.1 before and after dye degradation, which matched similar works [33,[57][58][59]. As pH strongly impacts the degradation and absorbance behavior of catalytic degradation systems, comparison between results is only meaningful at matched pH values [59][60][61]. The reported results are not optimized with regard to pH and can be directly compared to the literature, seen in table 1, while the degradative performance relationship between the laser-treated samples is preserved. The rate is determined from the linear fit of the concentration change. The photocatalytic rates during illumination are nearly identical from CN0 through CN120 (insert). This suggests that photocatalytic effects are not significantly improved due to irradiation. Dye degradation in g-C 3 N 4 is a complex process largely governed by intermediate species including holes (h+), super-oxide radicals (·O 2 -), singlet oxygen (·O 2 ), and hydroxyl radicals (·OH) [62,63]. Adsorption is more dependent on surface area and van der Waals forces, such that adsorption and catalytic activity are not intrinsically linked. In contrast to the catalytic rate, the dye absorption rates (calculated from the non- illuminated period) increase dramatically with laser irradiation; CN0 and CN15 remove dye at a rate of 0.12 and 0.13% min −1 respectively, while CN30 through CN90 see a steady increase in removal rate, with a slight decrease in the CN120 samples. This closely follows the NCO/N-H 2 /NO x XPS and partially matches the FT-IR trends, but does not correlate well with the BET surface area behavior. From that correlation, it appears that the non-illuminated dye removal/ absorption rate, either catalytic or degradative, is closely affiliated with the generation of defects in the material rather than the surface area. The initial MB concentration drop, where the first concentration measurement taken for each sample is compared to the known initial MB concentration, increases strongly with irradiation; CN0 and CN120 absorb 45 and 75% respectively. MB has a charged sulfur group which would be more strongly attracted to electron-donating or charged regions in the g-C 3 N 4 material. As laser irradiation drives the generation of electron-donating hydroxyl and amine groups, it is logical to conclude that some of the increase in MB absorption is due to interactions between the newly generated species and the MB dye. These results suggest that functional defect groups are suitable for improved organic absorption, which may be useful in reaction systems with absorbed species' lifetime limitations. While the untreated g-C 3 N 4 has poor performance, comparable to other similar materials (table 1), the CN120 sample performs remarkably, at the higher end of comparable materials. A significant difference between the experimental designs compared is the powers and wavelengths of the irradiation sources; various techniques and standards are employed due to material requirements or due to the specific nature of the target application. Our 6 W, 365 nm, light source represents only a fraction of the available solar spectrum (popular for solar applications) and is in the UV range (as opposed to various visible light targeting applications). UV-A (310-400 nm) irradiation was chosen due to the band-gap overlap with g-C 3 N 4 , enabling electron excitation as the primary thermodynamic driving force for degradation. Such features should be taken into consideration when comparing various literature results. It is clear that further improvements in performance might be obtained via doping or co-catalyst loading.

4. Conclusion
It is clear that the combined fundamental and second harmonic irradiation drives two competing processes; an exfoliation and a defect generation route. Exfoliation is effective for a 10× increase in the BET surface area in the g-C 3 N 4 material. Bonds associated with defects in the XPS highresolution spectra can reach as high as 30% of the relative species bonding. With respect to dye degradation, laser irradiation, and defect generation appear insufficient to produce an improved catalytic material; a co-catalyst is necessary to utilize the introduced defects for oxidative purposes. However, defective states act as enhanced absorptive sites for MB; a behavior that has the potential to be utilized in targeted diffusion-limited catalytic systems to increase the lifetime of the target species on the catalyst surface.
Post-exfoliation irradiation results in expanded g-C 3 N 4 domains, with nanosheets increasing by four-fold on average relative to the as-prepared material. This sheet growth mechanism is only seen under irradiation tailored for both electronic and kinetic excitation, indicating that irradiationrelated chemical changes are necessary precursors for this method of sheet growth. Defect generation appears to be a multi-purpose tool; it allows for functionalization by increasing species absorption rates and the number of active sites, by permitting a higher integration of co-catalyzed sites, while simultaneously operating as a novel growth mechanism in g-C 3 N 4 .

Acknowledgments
This work was supported by the Natural Sciences and Engineering Research Council of Canada-Discovery (GPIN-2020-06053) and MITACS-Accelerate (IT30845) Projects.

Data availability statement
The data cannot be made publicly available upon publication because the cost of preparing, depositing and hosting the data would be prohibitive within the terms of this research project. The data that support the findings of this study are available upon reasonable request from the authors.