Christian Fettkenhauera,
Jens Weberb,
Markus Antoniettia and
Dariya Dontsova*a
aMax-Planck Institute of Colloids and Interfaces, Department of Colloid Chemistry, Research Campus Golm, 14424 Potsdam, Germany. E-mail: dariya.dontsova@mpikg.mpg.de
bUniversity of Applied Science Zittau/Görlitz, Department of Chemistry, 02763 Zittau, Germany
First published on 21st August 2014
Poly(triazine imide)-based carbon nitride materials with BET surface areas up to 200 m2 g−1 were synthesized in ZnCl2 containing salt melts without the use of hard templates. We found that the composition, structural order, optical properties and morphology of the products can be adjusted by careful selection of synthesis parameters. The nature of the salt eutectic and precursor concentration in the melt have an especially large influence, with ZnCl2 being a reactive solvent. This novel synthesis route provides access to easily processable materials with improved optical absorption in the visible range that can be used as composite photocatalysts, CO2 adsorbents or nanocomposite fillers.
In general, molten salt(s) can serve as a solvent for high-temperature materials synthesis, as catalyst (for example, ZnCl2 in trimerization reactions) but also as “soft template” for tailoring micro- and mesoporosity of the resulting products. Previously, ZnCl2 and ZnCl2-containing melts were successfully used for the synthesis of porous covalent triazine-based frameworks11 and porous functional carbons.12
In this contribution, we further explore the utility of salt melt synthesis for the preparation of carbon nitrides, with the main focus on increasing the surface areas of products, as well as on controlling their structure and morphology by varying the synthesis parameters. We report on the preparation of new carbon nitride hybrid materials using zinc chloride containing eutectic mixtures, their characterization by means of X-ray diffraction (XRD), elemental analysis (EA), X-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM), FTIR-spectroscopy and gas adsorption, and discuss potential applications.
Salt mixture A/B(/C) | A content, wt% | B content, wt% | C content, wt% | Molar content of ZnCl2 | Melting temperature, °C (ref. 17) |
---|---|---|---|---|---|
ZnCl2 | 100 | — | — | 1.00 | 318 |
LiCl/ZnCl2 | 8.5 | 91.5 | — | 0.770 | 294 |
NaCl/ZnCl2 | 23.7 | 76.3 | — | 0.580 | 270 |
KCl/ZnCl2 | 36.3 | 63.7 | — | 0.490 | 230 |
CsCl/ZnCl2 | 48.3 | 51.7 | — | 0.575 | 263 |
NaCl/KCl/ZnCl2 | 10.7 | 13.8 | 75.5 | 0.600 | 203 |
Reference g-C3N4 was prepared by heating DCDA with the ramp of 2.3 °C min−1 up to 550 °C and subsequent holding at this temperature for 4 h under constant nitrogen flow (15 mL min−1) in a covered crucible. The final product was thoroughly ground.
k = ln![]() |
![]() | ||
Scheme 1 (a) Idealized condensation scheme of dicyandiamide (DCDA) to graphitic carbon nitride; (b) idealized structure of poly(triazine imide)/Li+Cl−.16 |
The synthesis and work-up procedure yielded colored products, which were analyzed by EA, EDX and XPS with regard to their composition (Table 2). The C/N weight ratios in the products range from 0.59 to 0.61 and are slightly higher than the value for reference g-C3N4 (0.58) but slightly lower than the stoichiometric value for C3N4 (0.64). This might be explained by incomplete amide condensation and the corresponding presence of terminal NH/NH2-groups. The sum of C, N, and H as obtained from EA is typically lower than 100%, usually around 82–87%. The remaining 13–18 wt% are Zn (3–5%), O (10–15%) and Cl (1–3%) ions as analyzed by EDX spectroscopy and XPS for selected samples. Zinc could be removed completely by washing the composites with 10 M HCl. This step is however accompanied by additional protonation of the product surface18 and was hence not employed in the default recipe. The presence of oxygen in products is partially the result of the aqueous work-up of the reaction mixtures but also due to the strong adsorption of water and CO2 under ambient conditions.
Melt | C, wt% | N, wt% | H, wt% | C/N, wt% | 100 − (C + N + H) | Color | SBET, m2 g−1 |
---|---|---|---|---|---|---|---|
Ref. bulk C3N4 | 35.0 | 60.1 | 2.06 | 0.58 | 2.8 | Yellow | 10 |
LiCl/ZnCl2 | 30.6 | 50.8 | 2.70 | 0.60 | 15.9 | Pale yellow | 20 |
NaCl/ZnCl2 | 29.6 | 49.1 | 3.04 | 0.60 | 18.3 | Brown | 193 |
KCl/ZnCl2 | 30.1 | 49.2 | 3.06 | 0.61 | 16.0 | Yellow | 42 |
CsCl/ZnCl2 | 30.2 | 49.8 | 3.13 | 0.59 | 16.5 | Yellow | 68 |
NaCl/KCl/ZnCl2 | 29.7 | 50.3 | 3.13 | 0.59 | 16.9 | Brown | 71 |
Pure ZnCl2 | 31.5 | 52.5 | 2.86 | 0.60 | 13.1 | Pale yellow | 58 |
XRD and FTIR spectroscopy investigations showed that the synthesized materials have features typically observed for previously described carbon nitride polymers, but differ from each other with respect to the degree of structural order. Fig. 1a illustrates that products prepared in NaCl/ZnCl2, CsCl/ZnCl2 and NaCl/KCl/ZnCl2 eutectics are mostly X-ray amorphous as they show only broad halos at 2θ ∼ 15° and 27°, LiCl/ZnCl2-derived material shows quite defined reflections corresponding to higher crystallinity, while the product obtained from pure ZnCl2 is characterized by outstanding structural order. The main 2θ reflection around 27° is present in the diffractograms of all materials and corresponds to interplanar stacking of carbon nitride layers. Another reflection observed for most of the materials (except for NaCl/KCl/ZnCl2 and NaCl/ZnCl2 eutectics) at 2θ ∼ 12–15° is related to in-plane structural packing motifs. The appearance of XRD patterns of products prepared in ZnCl2 and LiCl/ZnCl2 suggests that these materials are poly(triazine imide)-like as corroborated by recent investigations of acid washed poly(triazine imides) obtained from LiCl/KCl.19 KCl/ZnCl2-derived material seems to be composed of two phases characterized by different crystallinity.
![]() | ||
Fig. 1 (a) WAXS patterns and (b) FTIR spectra of bulk-C3N4 and products synthesized in salt melts containing ZnCl2. Vertical line indicates 1350 cm−1 position. |
In principle, crystallinity of carbon nitrides prepared in different ZnCl2-containing melts increases in the row NaCl/KCl/ZnCl2 ∼ NaCl/ZnCl2 < CsCl/ZnCl2 < KCl/ZnCl2 < LiCl/ZnCl2 < bulk g-C3N4 < ZnCl2, and therefore can be tuned by varying the nature of the salt eutectic used for the synthesis.
Fig. 1b shows that the main IR absorption bands displayed by the synthesized materials are similar to those of reference g-C3N4. Namely, the large absorption band at 1200–1650 cm−1 corresponds to stretching vibrations of CN heterocycles, a strong vibration at ∼800 cm−1 is attributed to deformation vibrations of triazine or tri-s-triazine ring, and the broad band between 2400 and 3400 cm−1 indicates stretching vibrations of surface hydroxyl groups (OH–) and terminal and residual amino-groups (NH2–, NH–). The presence of the distinct absorption band at ∼1350 cm−1 in ZnCl2-, LiCl/ZnCl2- and KCl/ZnCl2-derived materials suggests that they are built of triazine rather than of tri-s-triazine rings20 that is in good agreement with the XRD data. Besides, the amorphous nature of NaCl/ZnCl2, CsCl/ZnCl2 and NaCl/KCl/ZnCl2-derived products causes broadening and subsequent overlap of the absorption bands, while improved structural order (products from pure ZnCl2, LiCl/ZnCl2 and KCl/ZnCl2) results in more defined vibration peaks.
Amorphous CsCl/ZnCl2–C3N4 sample was investigated by XPS in order to unravel the structure of the carbon nitride polymer and intercalated Zn2+ species. The C 1s spectrum (Fig. 2a) shows a main carbon species with a binding energy of 288.2 eV corresponding to CN3 bonds.21 In addition to the C 1s peak of adventitious carbon at 285.0 eV, there is a weak peak at 283.3 eV which results from carbide impurities in the sample. The deconvolution of the N 1s signal reveals three contributions (Fig. 2b): at 401.2 eV (NHx groups), 399.7 eV (C–NC) and 397.6 eV (deprotonated N-sites). The number of contributions as well as their chemical shifts matches the calculated values for N 1s electrons in PTI prepared in LiCl/KCl.22 The C/N ratio is then calculated as a ratio of the area of the main carbon peak to the total area of nitrogen peaks. The value of 0.59 fits the one for poly(triazine-imide) reported by Wirnhier et al.10a and differs from fully condensed heptazine-based C3N4 (0.64). Furthermore, the ratio of amino-nitrogen atoms (NHx groups) to ring nitrogen atoms (C–N
C) calculated as a ratio of the corresponding peak areas, is equal to 0.49 and is close to the theoretical value of 0.5 for the idealized structure of PTI ((C3N3)2(NH)3). For heptazine-based structure this value would be much lower because only uncondensed terminal amino-groups would contribute to the XPS spectrum and only weak peaks of –NHx are reported for this case.21 All these findings prove that the carbon nitride component of the composite is based on poly(triazine imide).
XPS investigations further confirmed the presence of O, Zn and Cl in the sample. O 1s signal (Fig. S1a†) consists of three contributions: at 529.3 eV (assigned to O–Zn bonds23), 531.5 eV (–OH groups)24 and 533.0 eV (adsorbed water).25 The ratio between these three contributions is 1:
16
:
8. The presence of Zn–O bonds is further confirmed by Zn 3p signal (Fig. S1b†) which shows two main and almost equal contributions: at 88.4 eV for Zn–O bonds26 and at 89.6 eV for Zn–Cl and Zn–OH bonds. The peak of Cl 2p (Fig. S1c†) at 198.1 eV, by-turn, indicates the presence of Zn–Cl bonds. Basing on these findings, we conclude that the intercalated Zn species are mainly zinc oxide and zinc oxychloride. The latter one is the product of hydrolysis of ZnCl2 during the work-up procedure. The qualitative differences between surface and bulk of the product were analyzed by comparing the XPS spectra before and after Ar-ion bombardment of the sample which allows removing the surface layer of the material. The Cl 2p signal after ion bombardment reveals two peaks that correspond to different Zn–Cl bonds (Fig. S2a†). The contribution at 198.2 eV is due to zinc oxychloride while the one at 199.5 eV indicates the presence of zinc chloride27 which didn't hydrolyze being entrapped in the structure. The presence of ZnCl2 in the bulk of the product is also supported by Zn 3p signal (Fig. S2b†) which shows now an additional contribution at 91.4 eV due to Cl–Zn–Cl bonds. The amount of zinc oxychloride in the bulk of the product is much higher than the amount of zinc oxide, though at the surface their quantities were almost even. This finding together with the fact that the amount of Zn in the bulk of the material is at least 3 times higher than at the surface illustrates the difficulty to remove the intercalated Zn species ascribed to the diffusion limitations. In contrast, the oxygen content in the bulk of the product is at least two times lower than at the surface that is explained by the removal of surface hydroxyl groups and surface-adsorbed water. The weight content of the elements calculated from XPS data is in good agreement with the values obtained from EA, ICP and EDX measurements (Table S1†).
The morphology of the prepared materials was investigated by SEM. It was found to be fairly homogeneous within each of the samples and variant depending on the nature of alkali metal chloride constituting the salt melt (Fig. 3). Carbon nitrides are obtained as crystalline densely-packed nanosheets of 100–200 nm in diameter from pure ZnCl2 and LiCl/ZnCl2 eutectic melts. NaCl/ZnCl2, CsCl/ZnCl2 and NaCl/KCl/ZnCl2 melts give amorphous nanoparticles with the diameters of 50 ± 10 nm, 60 ± 10 nm and 40 ± 10 nm, respectively. KCl/ZnCl2 eutectic gives rise to the mixed product morphology containing larger sheets (1–2 μm in diameter), and spherical nanoparticles aggregates (∼300 nm in diameter).
![]() | ||
Fig. 3 Representative SEM images of carbon nitrides synthesized in (a) ZnCl2, (b) LiCl/ZnCl2, (c) KCl/ZnCl2, (d) NaCl/ZnCl2, (e) CsCl/ZnCl2 and (f) NaCl/KCl/ZnCl2. |
ZnCl2 salt melt derived materials have higher specific surface areas as determined by Brunauer–Emmett–Teller (BET) analysis of the gas adsorption data compared to the reference g-C3N4 (see Table 1). However, the values are much lower than those reported for other types of materials prepared in ZnCl2 melts (CTFs11 and carbons12). The appearance of nitrogen adsorption–desorption isotherms suggests the absence of any kind of inner porosity in final MCl/ZnCl2–C3N4 products, which is typical for carbon nitrides prepared in salt melts.10a Hence the estimated surface area values are reflecting only the external surface area of material particles, in good agreement with the results of SEM investigations.
The structural order and morphology of the products seem to correlate with the amount (χ) of ZnCl2 in the salt melt at a fixed melamine to salt weight ratio (1:
5) (see Table 1). Thus, at χ(ZnCl2)∼0.6, products are obtained as amorphous spherical nanoparticles. Being a strong Lewis acid, ZnCl2 interacts with cyano-groups of precursors and intermediates. This interaction facilitates their solubilization in the salt melt, but is also the source of the discussed structural complexity. The second component of the salt mixture (MCl, M = Li+, Na+, K+, Cs+) allows tuning the solubility of the reaction intermediates in the salt melt and thus influences the onset of “polymer–salt melt” phase separation. Li+ and K+ cations favor the solubilization of the reaction intermediates that results in slower phase separation and allow the growth of crystalline poly(triazine imides). Na+ and Cs+ seem to cause rather quick supersaturation of the melt with oligomeric intermediates that leads to the formation of numerous amorphous nanoparticles, which cannot crystallize into more extended networks.
We found that the microstructure of products is mainly defined already at 400 °C. This means that the final morphology is determined by precipitation and crystallization of intermediary condensates at the point of vanishing solubility in the salt melt. No significant changes in FTIR spectra and WAXS diffractograms can be observed when the synthesis temperature is further increased from 400 to 600 °C (Fig. S3†). However, the degree of condensation of the polymeric networks increases with raising the reaction temperature, resulting in slightly improved C/N ratios (0.59 to 0.61) and narrowing the band gap of the resulting semiconductors, as already visually observed as an intensification of products colors. In the case of applying a one-step heating procedure (Scheme S1b†), materials obtained at 550 °C are characterized by slightly higher values of the BET surface areas if compared to those synthesized at 600 °C (144 and 115 m2 g−1, respectively) while providing the same C/N ratio (0.61, Table S3†), thus 550 °C was selected as the temperature of choice for running the condensation. This is in good agreement with previous observations however on bulk condensation of C3N4 precursors.
Variations of the heating rate and holding time (Fig. S4 and S5†) have only minor influence on products structures, though the best products compositions characterized by a C/N ratio of 0.61 were obtained at 10° min−1 ramp and 6 hours holding time. Other studied parameters gave lower C/N ratios in the products (Tables S4 and S5†).
As the morphology is given by a product precipitation at rather early stages of the condensation, the concentration of the precursor in the melt must have crucial impact on the morphology and crystallinity of ZnCl2-derived carbon nitrides. Here, three different regimes were established: both low (1:
1 precursor to salts weight ratio) and high (1
:
10, 1
:
20 wt ratios) melamine concentrations give highly crystalline products, while concentrations in between (1
:
2, 1
:
5 wt ratios) lead rather to amorphous polymeric structures (Fig. S6†). The crystalline species have sheet-like morphology characterized by different diameters: 0.3–1 μm for 1
:
1 ratio, 1.5–2 μm for 1
:
10, and 50–80 nm for 1
:
20. The intermediary concentrations yield nanoparticles 50–100 nm in size (Fig. 4). The BET surface areas of products after acidic work-up vary between 60 and 190 m2 g−1 (Table S6†).
![]() | ||
Fig. 4 SEM images of materials synthesized in NaCl/ZnCl2 salt melts using different precursor to salt ratios: (a) 1![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
We can only speculate about the origin of this morphology changes. We suggest that at high precursor concentrations, a molecular complex between melamine and ZnCl2 is initially formed, which consumes all ZnCl2. Further transformations do not occur in a solution anymore but in solid state, and rather the structure of complexes between ZnCl2 and intermediates determines the high order in the final products. Upon decreasing melamine concentration, the initially formed molecular complex is redissolved at higher temperatures, but while polymerizing quickly reaches supersaturation and precipitates as amorphous nanoparticles. In some cases (NaCl/ZnCl2, CsCl/ZnCl2, NaCl/KCl/ZnCl2) even spinodal decomposition of the polymer–salt melt phase is assumed to take place, yielding highly microporous materials (see below). At low precursor concentration, demixing of phases occurs slowly, and the polymer (PTI) has enough time to crystallize. It ejects the solvent because the cohesion energy of carbon nitride is higher than the secondary valence to the metal center.
![]() | ||
Fig. 5 Selected SEM images of materials synthesized in ZnCl2 using different precursor to salt ratios: (a) 1![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Due to the very strong donor–acceptor interactions of ZnCl2 with the precursor, zinc cyanamide (also called zinc carbodiimide, Zn(CN2)) is formed as a by-product when performing the reaction in pure ZnCl2. At 1:
10 precursor to salt ratio, the yield of ZnCN2 is ∼50%. When the ratio is further lowered down to 1
:
20, solely ZnCN2 is formed (Fig. S7a and b†). These results prove the precoordination between the salt and the monomer via N–Zn donor–acceptor bonds formation and point out the alternative reaction pathway in the case of ZnCl2 if compared to LiX/KX (X = Cl, Br) melts. LiX/KX melts only weakly solubilize poly(triazine imides), therefore phase demixing occurs early, and big crystalline structures can grow from that melts. On the contrary, strong binding of Zn to condensation intermediates ensures their solubilization, and at later condensation stages the reaction progresses towards the thermodynamically more stable Zn(CN2) rather than to C3N4. Thus, ZnCl2 should be always considered as a reactive solvent in the case of carbon nitride synthesis.
Zn(CN2) can be easily removed by washing the product mixture with diluted acid, or converted to zinc oxide by re-heating the composite in air (2 hours at 400 °C; note that contrary to carbon nitride Zn(CN2) is not oxidation stable). In this way, PTI/ZnO composites can be obtained. Fig. 6 shows a TEM image and SAED pattern of such a composite containing ∼40 wt% C3N4 and ∼60 wt% ZnO (crystallite size: 13 ± 2 nm) and having a BET surface area of 270 m2 g−1 (Fig. S8†). The photocatalytic activity of this material is discussed below.
Optical absorption and emission properties of ZnCl2-derived carbon nitrides are compared with those of g-C3N4 in Fig. 7a and b, respectively. As it can be seen, NaCl/ZnCl2- and CsCl/ZnCl2-derived materials absorb visible light over a wide wavelength range. If compared to g-C3N4, the absorption band edge of these composites is obviously smeared to longer wavelengths. The appearance of the absorption spectra of NaCl/ZnCl2–C3N4, CsCl/ZnCl2–C3N4 and PTI/ZnO composite (Fig. S9†) is typical for dyade materials, such as carbon@TiO2.28 In a dyade, two components form a joint electronic system, and the altered absorption spectrum likely points out at charge transfer between carbon nitride and Zn containing species (such as ZnO). At the same time, the steady state photoluminescence of the products from NaCl/ZnCl2 and CsCl/ZnCl2 excited at 300 nm is negligible, that implies that the relaxation of excitons occurs via non-radiative pathways: charge transfer at heterojunction and/or non-radiative recombination at the impurity states in the crystal lattice.
![]() | ||
Fig. 7 (a) UV-visible light diffuse reflectance spectra and (b) steady-state photoluminescence spectra of g-C3N4 and MCl/ZnCl2–C3N4 (M = Li, Na, K, Cs) excited at 300 nm. |
The product obtained from LiCl/ZnCl2 melt on the other hand absorbs less visible light and is characterized by slightly stronger emission than g-C3N4. Both absorption and emission spectra indicate that this semiconductor has a wider band-gap compared to the reference carbon nitride. This is in good agreement with the assignment that this structure is based on triazine imide moieties, as previously reported salt melt products.19,22 The optical properties of the material synthesized in KCl/ZnCl2 lie in between those of LiCl/ZnCl2 and NaCl/ZnCl2- or CsCl/ZnCl2-derived composites, that is in agreement with XRD and SEM data discussed above.
The water-washed MCl/ZnCl2-derived composites typically possess higher specific surface areas than the acid-treated ones. The most striking example is NaCl/ZnCl2–C3N4, whose BET surface area dropped from 700 m2 g−1 (value after water washing) to 193 m2 g−1 after the acidic treatment (Fig. S11†). Such behavior may be explained by the collapse of the internal structure during acidic work-up or the removal of high surface area impurities, such as nanoparticles onto the hybrid. In the first case, the composite material after water washing may be envisaged as a poor, amorphous analogue of metal organic frameworks, in which triazine imide (oligomeric) moieties play the role of organic ligands, which are interconnected by Zn2+ ions, ZnCl2, ZnO clusters, or other species. Such microporous composite materials show very promising adsorption properties, which will be discussed later.
The highly microporous NaCl/ZnCl2-derived C3N4 composite (water-washed only) was investigated with respect to its CO2 adsorption properties (Fig. 8 and S12, S13†). It shows quite high CO2 uptake of 3.6 mmol g−1 at 0 °C and 2.5 mmol g−1 at 30 °C. The steep rise of the amount of adsorbed CO2 at low pressures indicates the presence of rather small micropores, which could indeed be interesting for separation applications. The N2 uptake at 0 °C and 30 °C was also studied and initial calculation with respect to the CO2/N2 selectivity α were undertaken. Ideal adsorbed solution theory (IAST)29 gives α = 225 (0 °C) and α = 100 (30 °C) at 1 bar and a gas composition of CO2/N2 of 0.15/0.85 that reflects an approximate composition of a flue gas. Such selectivities are indeed competitive to some reported zeolites30 and MOFs31 given also the simplicity of synthesis and purification procedures as well as the relatively high product yield of 50% (based on total C content in the precursor and the product).
![]() | ||
Fig. 8 CO2 and N2 adsorption–desorption isotherms of microporous NaCl/ZnCl2-derived C3N4 composite (after water washing step). |
The photocatalytic properties of ZnCl2-derived carbon nitrides were evaluated in a model reaction assay, the photodegradation of Rhodamine B (RhB) under blue light (λ = 420 nm) irradiation. Here, two selected examples of remarkable performance are the microporous NaCl/ZnCl2-derived C3N4 hybrid (water-washed), again, and the C3N4/ZnO (40 wt%/60 wt%) nanocomposite. The estimated rate constants of RhB degradation accomplished by these two photocatalysts are k0 = 21 × 10−3 min−1 and k0 = 12 × 10−3 min−1, respectively; these are ∼10 and ∼5 times higher than the one observed for the reference g-C3N4 (2.2 × 10−3 min−1, Fig. S14†). We mainly attribute these enhanced activities to the increased surface areas of the composites and improved absorption of visible light due to the formation of the dyadic structures. Additionally, the C3N4/ZnO hybrid is characterized by an improved crystallinity of ZnO NPs that results in better transport of the photo-generated and transferred electrons to the surface-adsorbed RhB and O2 molecules.
Footnote |
† Electronic supplementary information (ESI) available: O 1s, Zn 3p and Cl 2p XPS spectra of CsCl/ZnCl2–C3N4, schemes of heating procedures; WAXS patterns, FTIR spectra, EA and surface areas of products synthesized at different temperatures, heating rates, holding times and ratios; WAXS patterns and FTIR spectra of materials made in pure ZnCl2; diffuse reflectance spectra of NaCl/ZnCl2–C3N4 and C3N4/ZnO composites, CO2 and N2 adsorption–desorption experimental isotherms and predictions of gas uptake of CO2 and N2 at 303 K, Rhodamine B degradation data. See DOI: 10.1039/c4ra08236b |
This journal is © The Royal Society of Chemistry 2014 |