Article Cite This: Chem. Mater. XXXX, XXX, XXX−XXX
pubs.acs.org/cm
Mild Solvothermal Growth of Robust Carbon Phosphonitride Films Brian L. Chaloux,*,† J. Michael Shockley,‡ Kathryn J. Wahl,† Stanislav Tsoi,† Andrew J. Birnbaum,§ and Albert Epshteyn*,† †
Chemistry Division, ‡NRC Postdoctoral Associate, and §Computational Multiphysics Systems Lab, U.S. Naval Research Laboratory, Washington, D.C. 20375, United States
Chem. Mater. Downloaded from pubs.acs.org by UNIV OF SOUTH DAKOTA on 08/24/18. For personal use only.
S Supporting Information *
ABSTRACT: Solvothermal growth of thin (1−2 μm thick), uniform films of carbon phosphonitride on fused quartz substrates has been achieved by self-reaction of 1 m P(CN)3 in diphenyl ether at 185 °C for 7 days in a sealed Teflon-lined digestion bomb. The films are chemically similar to C3N3P powders thermoset from neat (solid) P(CN)3 based on XPS and IR spectra. A difference in bulk elemental composition relative to C3N3Preduced nitrogen and increased oxygen contentis attributable to through-film (as opposed to surface-confined) hydrolysis of unreacted nitriles (P−CN) to hydroxyl species (P−OH). Nanoindentation analyses were performed to assess the mechanical properties of these CPN films. Minimal dependence of mechanical properties on thermal treatment was observed with annealing temperatures in the 100−400 °C range, resulting in an average reduced modulus of Er = 26.5 ± 1.5 GPa and hardness of H = 1.5 ± 0.1 GPa, °C which is on par with high performance polymers. Heating to 550 °C results in higher modulus and hardness (E550 = 51 ± 2 r GPa, H550 °C = 5.0 ± 0.4 GPa), likely due to an increased prevalence of oligomeric phosphine oxide cross-links formed by dehydration of P−OH functional groups.
■
INTRODUCTION Amorphous carbon-based films are utilized extensively in coating applications for their hardness and wear resistance.2,3 These materials are often processed to introduce dopants, such as nitrogen, to improve mechanical properties for hard, functional coatings and wear-resistant surfaces. Nitrided carbon coatings are typically grown by plamsa/vapor deposition methods, producing carbon nitride films with variable CNx compositions.4 While nitrogen doping by these methods can result in elastic films with cross-linked, fullerenelike microstructure,5 it has not been confirmed to produce sought-after crystalline C3N4 phases.6 However, the phases often called “graphitic” carbon nitride (a.k.a. g-C3N4) referring to their sheet-like, mostly sp2 hybridized structuresare amenable to bulk synthesis directly from molecular precursors, albeit as amorphous materials. These related materials, ranging in composition from “melon”-like (linear, C6N9H3)7 to truly graphitic (C3N4) depending on extent of polymerization, are experiencing a renaissance of research interest not just for coating applications, but for their potential as a metal-free, heterogeneous catalysts8−13 and for filtration/ gas sorption.10,12,14 Their syntheses, however, require high temperatures (≥400 °C) and can create large quantities of gaseous byproducts (NH3, HCl, etc.), making film processing difficult.15 For applications in catalysis, filtration, and gas sorption, morphology and physical integrity are often © XXXX American Chemical Society
secondary to chemistry, allowing utilization of g-C3N4 in powder form. That carbon nitrides have been predominantly studied in powder form is likely due to difficulties manipulating these materials: they are insoluble in most solvents,16 and they decompose prior to a measurable melting point.17 Another relevant system are the carbon phosphides, CPx, which have been less extensively studied than their carbon nitride counterparts; however, plasma and sputter-deposited CPx films with varying C:P ratios have been demonstrated.18−20 Computational studies have shown that many of the predicted phases for the CPx system parallel the phases (predicted and experimentally observed) of CNx.20−23 Although molecular precursors to synthesize pure carbon phosphides are less readily available than carbon nitride precursors, mixed CNP materials are accessible24 and have also been investigated by ab initio computational methods.25 Recently, we discovered a phosphorus-substituted analog of g-C3N4 utilizing P(CN)3 as a molecular precursor to the compound C3N3P, illustrated in Figure 1.1,26 The thermal selfreaction of P(CN)3 proceeds at comparatively low temperatures (≥185 °C) and yields C3N3P quantitatively, improving process compatibility with low-temperature materials.1 The Received: June 13, 2018 Revised: August 9, 2018
A
DOI: 10.1021/acs.chemmater.8b02508 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials
use. Ethereal solvents (diphenyl ether, diethyl ether, and THF) were distilled from Na−benzophenone ketyl. P(CN)3 was freshly sublimed prior to use. Slide Preparation. All films were prepared on glossy, (amorphous, see Figure S6) fused quartz slides cut to dimensions of 1.5 cm × 2.5 cm × 1.5 mm. Four batches, A−D, were prepared where all samples from the same batch were simultaneously deposited in the same reactor. All slides were cleaned with soap and water, treated overnight with hot acid piranha solution (7:1 H2SO4 + 30% H2O2), rinsed with deionized (DI) water, and then dried at 185 °C prior to use. Warning: piranha solutions are extremely corrosive. Use appropriate personal protective equipment and splatter guards and cover solutions with a watch glass at all times. Items should not be placed into piranha before preliminary cleaning, as a substantial quantity of organic contaminant may result in boilover. Slides were prescored with a glass cutter prior to cleaning to enable bisection after film deposition and analysis of fresh, identical samples by orthogonal analytical techniques. Film Deposition and Postprocessing. Six slides at a time were placed in a 23 mL Teflon (PTFE) cup and secured in an upright position with a notched silicone rubber bar (Figure 2a). 1.0 g of
Figure 1. Structural representation of C3N3P, produced as a black powder by heating crystalline P(CN)3 at ≥185 °C under inert atmosphere (e.g., Ar, N2, vacuum).1
resultant amorphous carbon nitride analogue, which we have dubbed carbon phosphonitride (CPN), exhibits good thermal and chemical stability and an optical band gap (∼1.4 eV) lower than g-C3N4 (1.7−2.8 eV).17,27−29 These attractive materials properties can potentially be harnessed for use in devices; however, because P(CN)3 polymerizes below its melting point and exhibits modest vapor pressure at elevated temperature (∼55 mmHg at 185 °C),30 it suffers some of the same process difficulties as g-C3N4. To circumvent challenges inherent to traditional carbon nitride syntheses, researchers have explored the use of chemical vapor deposition (CVD) from volatile monomers, electrolytic deposition, and solvothermal film growth processes. Only in several instances, by clever design of volatile monomers, have g-C3N4 films been successfully prepared by CVD/pyrolysis.17,31,32 Only a single report exists of C3N3P CVD from a similar monomer.24 Although electrochemical methods have been widely successful at depositing films, the electrochemistry at play and high operating voltages result in products with composition closer to CNx (x ≤ 0.66) than C3N4.33−38 Recently, several groups have developed promising, mild solution-based methods for the preparation of carbon nitride films. Zhang et al. demonstrated film deposition by solvolysis and sol processing of melon-like carbon nitride in nitric acid,16 while Xie et al. demonstrated direct solvothermal film growth of triazine-based carbon nitride from molecular precursors in acetonitrile.39 Herein, we report a solvothermal, direct growth method for the preparation of smooth CPN films from solutions of P(CN)3, which is similar in concept to the growth of carbon nitride from acetonitrile.39 CPN films were found to grow more slowly from diphenyl ether and form thinner, more uniform films than their CN analogue deposited from acetonitrile.39 This solvothermal method for growing CPN overcomes some of the limitations inherent to working with P(CN)3, enabling thin film growth of carbon phosphonitride on a variety of substrates under mild conditions. Having such films is an important first step toward enabling the use of carbon phosphonitride for coating, semiconductor, and electrocatalytic applications; it also enables detailed measurements of their mechanical properties.
■
Figure 2. Apparatus for deposition of CPN films from P(CN)3 solution: (a) 23 mL PTFE cup with quartz slides and silicone divider; (b) acid digestion bomb assembly (cup, lid, and pressure vessel). P(CN)3 and 10 mL of preheated diphenyl ether (Ph2O) were added to the cup. The Teflon lid was secured tightly, and the closed assembly was then heated to 150 °C on a hot plate for 1 h to dissolve the P(CN)3. The cup was subsequently placed in a Parr digestion bomb (Figure 2b), closed tightly, and heated in a vented (to air) convection oven at 185 °C for 7 days. After deposition, the apparatus was transferred back into a drybox and disassembled. Slides were removed individually from the cup, washed with diethyl ether and THF, and gently buffed with a THF-wet KimWipe to remove loosely adhered particulates. Samples were heat treated under vacuum for 4 h at 100, 250, 400, or 550 °C to remove residual solvent and anneal the CPN films. The low end of heat treatment temperatures was chosen to be high enough °C = 0.6 mmHg)30 and to sublime any unreacted P(CN)3 (P100 vapor °C residual Ph2O (P100 = 3 mmHg). The high end of treatment vapor temperatures was chosen to be sufficiently below the decomposition temperature of C3N3P measured in prior work (∼800 °C at 10 °C min−1 heating).1 Intermediate temperatures of 250 and 400 °C were chosen as convenient, evenly spaced intervals. Samples were stored wrapped in aluminum foil in individual plastic bags (in the drybox). Cleavage of coupons into smaller units was performed immediately prior to analyses. Sample names are abbreviated by batch letter and heat treatment temperature, e.g., D400 = Batch D, 400 °C heating. Analytical Evaluation. Scanning electron microscope (SEM) images were collected on a Zeiss Leo SEM and FIB performed on a FEI Nova 600 NanoLab SEM, both operating at 5 keV. AFM images were acquired with an Asylum Cypher AFM operated in contact mode. Nanoindentation measurements were made with a Hysitron Ubi instrumented indenter operated in load control configuration. A diamond indenter of Berkovich pyramidal geometry was impressed to peak loads of 100, 500, and 1000 μN at 5 s loading and unloading
EXPERIMENTAL SECTION
Manipulations were performed in an argon-filled drybox, except where otherwise noted. P(CN)3 was prepared by refluxing PCl3 with a stoichiometric amount of AgCN, as previously described.1,40 Warning: P(CN)3 hydrolyzes to toxic HCN gas and should therefore be handled in a dry, well-ventilated environment. Explosive decomposition has been observed to result f rom rapid heating above 200 °C in a sealed system; protective shielding and a slow temperature ramp are recommended when heating above 100 °C. All solvents were dried and deoxygenated before B
DOI: 10.1021/acs.chemmater.8b02508 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials rates with a hold time at peak load of 5 s. The unloading curves (examples in Figure S20) were analyzed using the Oliver and Pharr method41 to determine reduced modulus (Er) and hardness (H). X-ray photoelectron spectra (XPS) were collected on a Thermo Scientific K-Alpha XPS; binding energies were calibrated to adventitious carbon (C 1s = 284.8 eV) and spectra fit with Shirley backgrounds and symmetrical Gaussian−Lorentzian (GL) peaks. Infrared (IR) spectra were collected on a ThermoFisher 670 with a Pike MIRacle attenuated total internal reflectance (ATR) attachment using a diamond window and mercury cadmium telluride MCT-A detector at a resolution of 4 cm−1. X-ray diffraction (XRD) was performed in grazing incidence mode (0.2° incident angle) on a Rigaku SmartLab diffractometer using Cu Kα radiation. Samples were kept under argon prior to analysis. For SEM, AFM, nanoindentation, IR, and XRD, samples were transferred over Drierite (CaSO4) desiccant in a nitrogen-purged vessel to prevent hydrolysis. For XPS, samples were loaded directly into an air-free vacuum transfer module (sold by Thermo Fisher Scientific) in a drybox and pumped into the K-Alpha antechamber without air exposure. IR and XRD were performed rapidly under air. SEM and XPS were acquired under dynamic vacuum. Nanoindentation and AFM were performed at room temperature in a chamber purged with dry nitrogen.
bulk compositions. The expected elemental makeup of C3N3P is C = 43 at. %, N = 43 at. %, and P = 14 at. %, which are approximately the values found (when converted from wt %) for P(CN)3 polymerized neat.1 CPN films grown from Ph2O, however, were found to contain a substantial amount of oxygen (22−28 at. %) and a significantly reduced amount of nitrogen (12−18 at. %). The silicon content, albeit low (2300 cm−1, except for the melamine-derived samples, which assay for unreacted −NH2/−NH− groups.17 Unlike C3N4 derived from melamine, amines are not a natural substituent for carbon phosphonitride. However, the peak observed at 2400 cm−1 is particularly unusual and therefore helps assessment tremendously. The first instinct is to assign this vibrational mode to nitriles in a chemically distinct environment from the species at 2200 cm−1. However, an extensive survey of literature shows that in no compoundswhether organic, inorganic, or organometallicare the CN stretching modes shifted to such a high frequency. The only fundamental vibrational modes that occur near 2400 cm−1 are the N−D/O−D stretch, P−H stretch, and S−H stretch.45 Of these candidates, only P−H is a plausible functional group for our films. Ultimately, the 2400 cm−1 mode (P−H), 3600−2500 cm−1 mode (O−H), elevated oxygen content, and diminished nitrogen content can all be explained by partial hydrolysis of the films (Figure 10). At least some nitriles (CN) remain
Figure 9. Calculated reduced elastic moduli and hardness (in compression) of heat-treated samples from batches A and D. Means and standard deviations from multiple indentations are represented by bar heights and error bars, respectively.
μm over the course or several days. Utilizing a sealed apparatus (Figure 2) was crucial to prevent loss of P(CN)3 monomer under deposition conditions. Diphenyl ether (Ph2O) was found to be an ideal solvent, combining sufficiently low vapor pressure (∼110 mmHg at 185 °C), dissolution power for P(CN)3 (≥1 m at 140 °C), and inertness to the reactive monomer (from NMR, Figures S16 and S17). Despite growing with low surface roughness compared to overall film thickness, a substantial number of micrometersized particulate dot the surface of all CPN films we have characterized. Surface coverage of these particles is low, but non-negligible. However, their presence suggests that the film growth process is likely of a sol−gel type: reactions of P(CN)3 in solution generate intermediates which eventually fall out of solution and act as nucleation sites for further growth. Sol−gel behavior was also observed qualitatively in NMR scale test reactions (Figures S16−S18): given several days reaction time, dark gels formed within the 4 mm inner diameter tubes. The most marked difference between the growth of CPN and an analogous solvothermal deposition of g-C3N4 from acetonitrile 39 is film morphology. While Xie et al.’s solvothermally grown g-C3N4 films appear to be compacts of bumpy 1−5 μm diameter particles,39 the building blocks of CPN films grown from P(CN)3 in Ph2O appear by AFM (Figure 7 and Figure S4) to be closer to tens of nanometers in size. Chemistry. The primary quandary arising from the characterization of our solvothermally grown CPN films is the substantial difference in elemental compositionparticularly nitrogen and oxygen contentfrom the expected values (C = 43 at. %, N = 43 at. %, and P = 14 at. % for P(CN)3 and C3N3P). Assaying for only 12−18 at. % nitrogen and 22−28 at. % oxygen, it might be reasonable to assume that metathesis of these elements occurs between P(CN)3 and Ph2O over the course of the reaction. Although Ph2O and Ph3N share similar
Figure 10. Illustration of likely chemical motifs in CPN films: phophorus-bridged triazines, unreacted nitriles, hydrolyzed phosphonous acids (PIII), and phosphine oxides (the PV tautomer).
unreacted, as is evident in both IR (peak at 2200 cm−1) and N 1s XPS (∼399 eV) spectra. Like P(CN)3, it follows that in the presence of atmospheric (or other sources of) moisture these bonds hydrolyze to P−OH + HCN. In the case of C3N3P powder, such hydrolysis is probably limited to the surface,1 while CPN films appear to be sensitive throughout their thickness. P−OH accounts for the broad IR absorbance from 3600−2500 cm−1 and contributes to the O 1s XPS peak at ∼533 eV. F
DOI: 10.1021/acs.chemmater.8b02508 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials
Figure 11. ATR-IR spectra of batch D CPN samples with peaks labeled according to assignments from Table 2. Molecular diphenyl ether disappears at or below 400 °C. At 550 °C, PO stretching shifts to higher wavenumber, indicating dehydration.
The P−H stretching mode observed at 2400 cm−1 is accounted for by tautomerization of P−OH to P(O)H, illustrated schematically in Figure 10. Although unusual at first glance, this behavior is well documented for phosphinous and hypophosphinous (PIII) acids.46−48 Hypophosphorous acid (H3PO2) exhibits strong bands at 2380 cm−1 for P−H stretching and 1180 cm−1 for PO stretching,46 while substituted compounds such as diphenyl phosphite show both O−H stretching at ∼3500 cm−1 and P−H stretching at ∼2300 cm−1 in solution.48 Substantial hydrolysis of P−CN bonds (∼50−60%) can easily account for the increase in oxygen content and reduction in nitrogen content of solvothermally grown CPN films. Additionally, inclusions of diphenyl ether may account for the apparently high carbon content, offsetting the carbon lost to P−CN hydrolysis. The mechanism by which water was introduced to the CPN films is still a mystery, but XPS and IR spectra show that the chemistry of films produced by solvothermal growth (CPN) and thermally polymerized powders (C3N3P)1 are qualitatively similar. Unfortunately, XPS proved only marginally useful for chemical analysis, as peak positions in the literature are too variable to be positively assigned to single species in the CPN films. Infrared spectra of CPN films heat-treated above 100 °C change in small, but noticeable ways (Figure 11). The simplest is disappearance of the sharp vibrational modes associated with diphenyl ether (Figure 11, peaks 4 and 5). The weak intensity of the Ph2O modes and breadth/overlap of peaks attributable to amorphous CPN preclude determination of whether molecular Ph2O has evaporated at elevated temperature or if it otherwise decomposed and incorporated itself into the film. EDS (Table 1, D100−D550) shows negligible change in carbon and oxygen content versus temperature, suggesting Ph2O pyrolysis rather than evaporation. When analyzing changes in the broad peaks associated with amorphous CPN, it is important to note that the IR baseline changes substantially versus heat treatment temperature from 600−1300 cm−1. The shifting baseline is likely due to differences in film thickness, and therefore contributions from Si−O vibrational modes (Figure S10), as changes do not directly correlate with treatment temperature. With this in mind, there are two noteworthy chemical changes evident in IR spectra of the film. First, the weak P−H (2400 cm−1) and CN (2200 cm−1) modes visible in sample D100 are absent in D400 and D550. For CN, this indicates completion of reaction of the dangling nitrile groups. A decrease in P−H intensitywhich
tracks with the decreasing P 2p shoulder observed in XPS at 130.4 eV (Figure 4)might result from a shift of the favored oxyphosphine tautomer from P(O)H (hydride) to P−OH (hydroxyl), except the O−H stretching mode also weakens with increasing treatment temperature. Combined with a concurrent increase in intensity of the PO stretching vibrations, what likely occurs is dehydration of P−OH to P− O−P, shifting the tautomeric equilibrium away from hydride form to hydroxyl form. Substantial bimolecular dehydration of phosphoric acid to pyrophosphate (H4P2O7) is observed at 170 °C, consistent with our observations.49 Second, when heating to the highest temperature, 550 °C, the asymmetric PO stretching mode initially observed at 1190 cm−1 appears to shift and broaden toward 1400 cm−1. This behavior is indicative of multiple dehydration reactions, forming clusters with phosphorus pentoxide-like structure (T2symmetric PO stretch ≈1400 cm−1 in P4O10).50 Unlike bimolecular dehydration, which may proceed between adjacent phosphines at lower temperatures, the polymolecular processes necessary for three- and four-center cross-link formation are only likely at temperatures high enough for the less common bis-hydroxyl-substituted phosphines to encounter each other. Mechanical Properties. CPN films grew slowly from solution, achieving thicknesses of only 1−2 μm after 7 days of deposition. Despite some amount of adhered particulate, the films were uniform enough to perform mechanical analysis of hardness (H) and reduced elastic modulus (Er) by nanoindentation (Figure 9 and Figure S19). Although reduced modulus is related to Young’s modulus by Poisson’s ratio, ν, (EYoung ≈ Er(1 − ν2) when Er ≪ Eindenter), Poisson’s ratio for our CPN films is unknown; thus, only Er is reported here for the samples measured. For the majority of materials, however, 0 (cork) > ν ≥ 0.5 (rubber), meaning that the value of Young’s modulus is expected to be 75%−100% of the reduced modulus as reported.51 Surprisingly little variability in the average values of Er and H was observed for heat treatment temperatures between 100 and 400 °C, indicating minimal structural evolution of films under these conditions (either chemical or morphological). As they are heated to 550 °C, films clearly do evolve, although the variability in both Er (±15 GPa) and H (±2.6 GPa) from batch to batch is substantial. Initially, we suspected that the increase in apparent modulus and hardness on annealing at 550 °C might be attributable to crystallization of the film, a change in thickness (thus increasing substrate effects on indentation), or a change in elemental composition. However, grazing incidence XRD (Figure S6) showed no crystallization, G
DOI: 10.1021/acs.chemmater.8b02508 Chem. Mater. XXXX, XXX, XXX−XXX
Chemistry of Materials
■
and elemental composition was shown to be unchanged (to within error) by both XPS and EDS (Table 1). Thickness is unlikely to be the cause, either, as the only data included in statistical analyses were those where contact depths calculated by the Oliver and Pharr method41 were
