Mahsa
Parvizian
and
Jonathan
De Roo
*
Department of Chemistry, University of Basel, Basel, Switzerland. E-mail: Jonathan.DeRoo@unibas.ch; Tel: +41 61 207 10 46
First published on 15th November 2021
Metal nitride nanocrystals are a versatile class of nanomaterials. Depending on their chemical composition, the optical properties vary from those of traditional semiconductor nanocrystals (called quantum dots) to more metallic character (featuring a plasmon resonance). However, the synthesis of colloidal metal nitride nanocrystals is challenging since the underlying precursor chemistry is much less developed compared to the chemistry of metal, metal chalcogenide or metal phosphide nanocrystals. Here, we review chemical approaches that lead (or could lead) to the formation of colloidally stable metal nitride nanocrystals. By systematically comparing different synthetic approaches, we uncover trends and gain insight into the chemistry of these challenging materials. We also discuss and critically evaluate the plausibility of certain suggested mechanisms. This review is meant as a guide for the further development of colloidal nitride nanocrystals.
Ligand-assisted, wet chemical synthesis has provided the necessary control over nanocrystal size and shape.5 However, it is a complex process, combining metal–organic chemistry, inorganic crystallization and surface chemistry to yield an organic/inorganic hybrid object that is colloidally stable. The precursor conversion mechanism of many metals, metal chalcogenide, metal phosphide and metal halide NCs has been elucidated.6–9 and also their surface chemistry is increasingly better understood.10–16 The crystallization mechanism (i.e., nucleation and growth) causes more discussion in the literature.17,18 Approaches based on classical nucleation theory are at odds with the Finke–Watzky based mechanisms.19,20 Recent evidence shows that many NCs do not nucleate via a burst nucleation event, but rather that nucleation is a continuous process.21–23 Continuous nucleation does not necessarily lead to polydisperse ensembles because small NCs grow faster than larger NCs, allowing the newly formed NCs to catch up with the growing NC population.22–25 Even without a comprehensive framework for NC crystallization, chemists have been able to adjust the final NCs size by controlling the precursor conversion rate.26 The three aspects of colloidal synthesis (precursor conversion, crystallization and surface chemistry) are interdependent. An additional, often neglected parameter is the solvent. During high temperature nanocrystal synthesis, certain popular solvents decompose or polymerize, forming products that compromise the purity of the final NCs.27,28 Moreover, the purity of the ligand can considerably influence the solubility of the starting precursors and provide a better capping of the final NCs.29
While metal, metal chalcogenide and metal halide NCs are well studied, the metal pnictides have received less attention, with the exception of metal phosphides.9,30–33 Metal nitrides in particular are a versatile class of nanomaterials but their synthesis is not well developed. One difficulty is the higher thermodynamic stability of the oxides, compared to the nitrides (see Table 1). One can rationalize the lower standard formation free energy for nitrides because of the high bond enthalpy of dinitrogen (941 kJ mol−1) versus dioxygen (500 kJ mol−1).34,35 Another factor is the lower ionic character of metal nitrides since the enthalpy of formation of N3− is much higher than for O2− (2300 kJ mol−1versus 700 kJ mol−1).36 Only AlN and alkali(ne earth) metal nitrides are regarded as saline nitrides and they react quickly with water, releasing ammonia.37 On the other hand, boron and silicon nitride are almost purely covalent compounds with high stability.37 Transition metals like titanium form interstitial nitrides with extremely high chemical, thermal and mechanical stability.37 TiN has the fcc structure of metallic titanium with interstitial nitrogen atoms in the octahedral holes (Fig. 1). Given the above considerations, it is no surprise that metal nitride synthesis is generally carried out at a high temperature and with nitrogen precursors that are more reactive than molecular nitrogen.
Fig. 1 (a) Crystal structure of group 4 metal nitride. The metal is represented in purple and the nitrogen atom in grey. Structures plotted using VESTA.48 Cif file obtained from ref. 49–51 (b) Absorption spectra of aqueous dispersions of TiN, ZrN, and HfN NCs. The inset shows a photograph of the NCs dispersed in water. Reprinted from ref. 40 with permission from John Wiley and Sons, Copyright 2019. |
Here, we review the existing synthetic strategies that lead to metal nitride NCs, with a focus on the chemical conversion reactions. A broader review on nitride nanomaterials, including porous and 2D materials can be found elsewhere.39 We target predominantly wet chemical syntheses and colloidally stable nanocrystal products. Other methods such as gas phase or solid state synthesis are only included when informative. We do not discuss nitride NCs of the alkali metals or alkaline earth metals since there are no synthetic procedures for nanocrystals available and they are extremely sensitive towards humidity, rendering them less technologically interesting. We will go systematically through the periodic system and discuss the nitrides per group, starting with group 4 and ending with group 13.
In 1988, TiN films were prepared using atomic layer epitaxy at 500 °C; TiCl4 precursor was pulsed into a reactor with gaseous ammonia.52 The colour depends on the stoichiometry of the film. For 40% N a golden yellow color is observed while for 50% a dark violet is obtained. Alternatively, Ti(NMe2)4 is reacted with ammonia. Similar strategies apply to ZrN and HfN. An extensive account of customized precursors for Chemical Vapour Deposition (CVD) of the group 4 nitrides is reviewed elsewhere.53
6TiCl4 + 8NH3 → 6TiN + N2 + 24HCl | (1) |
Fig. 2 Diagram of the nonthermal plasma apparatus for TiN NCs synthesis. Reprinted with permission from ref. 55. Copyright 2018 American Chemical Society. |
The particle size is tuned from 3 to 8 nm by decreasing the flow rate of ammonia. The smaller particles are more oxidized than the bigger particles, but all show a plasmon resonance between 800 and 1000 nm. One can obtain ZrN nanocrystals in a similar fashion using ZrCl4 and NH3 as the precursors.56 After synthesis, the powders need to be heated at 200 °C in an Ar atmosphere to remove the ammonium salt by-products. To avoid this last step, TiCl4 was later replaced with tetrakis(dimethylamido) titanium (Ti(NMe2)4).55
6Ti(NMe2)4 + 8NH3 → 6TiN + N2 + 24HNMe2 | (2) |
Also here, the particles are 6–8 nm in size (Fig. 3) and have a well-defined plasmon resonance (around 800 nm). However, the particles are not monocrystalline (consisting of 2–3 nm crystallites) and contain a considerable amount of carbon.
Fig. 3 High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images of the obtained TiN NCs via non-thermal plasma synthesis. Scale bars are (a) 20 nm and (b) 5 nm. Reprinted with permission from ref. 55. Copyright 2018 American Chemical Society. |
Fig. 4 Reactions of titanium chloride with tertbutylamine and thiphenylphosphine oxide or tertbutylpyridine. |
When Ti(NMe2)4 is treated with primary amines such as n-propylamine, n-octylamine or tert-butylamine, transamination occurs.59–61 Subsequently, the primary amido group self-condenses and thus forms alkylimido bridges, see Fig. 5. The rate of transamination and condensation is clearly dependent on the steric hindrance between the alkyl groups of the amine since it proceeds swiftly at room temperature for n-propylamine while tert-butylamine only forms the condensed product at 100 °C.61 When using an excess of primary amine, a polymeric, insoluble product is obtained.60 However, sols can be formed in tetrahydrofurane (THF) by controlling the amine to titanium stoichiometry. These sols were then used to coat silica via dip-coating and Ti(C,N) and TiN were obtained upon pyrolysis at high temperature (minimum 800 °C) under nitrogen or ammonia atmosphere respectively.62
Upon addition of one equivalent of primary amine to Ti(NMe2)4, the dimer 4 is formed, which can rearrange to a monomeric imido species by coordination of a Lewis base. Addition of two equivalents of trimethylsilylchloride to the latter complex yields the dichloroimido complex.61,63 The same product can be made by the addition of a primary amine to Ti(NMe2)2Cl2.64 Although most of the above titanium complexes have not been used so far to synthesize TiN NCs, they serve as valuable inspiration for future synthetic development and grant insight in the precursors chemistry of titanium and the rest of the group 4.
Recently, titanium chloride was reacted with potassium amide in pyridine at 300 °C.65 While the TEM image showed an isolated TiN nanocrystal of 5 nm, the XRD pattern was too broad to be consistent with a large ensemble of 5 nm crystals. Indeed, another TEM image clearly shows aggregated particles in an amorphous matrix.
Fig. 6 Temperature-dependent powder XRD analysis of the TiO2 precursor/mpg-C3N4 composites. (+) represents mpg-C3N4; (o), TiO2; and (*), TiN. Reprinted from ref. 66 with permission from John Wiley and Sons, Copyright 2007. |
In 2007, Buha et al. investigated the thermal transformation of pre-synthesized anatase TiO2 NCs into nanocrystalline TiN by using cyanamide or urea as a nitrogen source at 800 °C under nitrogen atmosphere.67 Complete transformation is achieved when starting from 5 nm titania while with 10 and 20 nm titania, anatase impurities are still present. This reflects the solid state nature of the approach. When using a high excess of cyanamide or urea, a large amount of amorphous carbon is also produced (20–25 wt%). This can be minimized to 2 wt% by reducing the cyanamide or urea amount. The resulting TiN NCs have an average crystallite size of 3 nm and are aggregated to 50 nm agglomerates, embedded in amorphous carbon. Similarly, ZrN nanocrystals were synthesized by reacting 10–15 nm ZrO2 nanocrystals with urea at 1100 °C under an ammonia flow.68 In 2009, Antonietti and coworkers presented the “urea glass route” which is a combination of sol–gel chemistry and the above strategy of Buha et al.69 First, TiCl4 is dissolved in ethanol, forming titanium ethoxide and releasing HCl as side product. Then a varying amount of solid urea is added to the solution and stirred until complete dissolution and the solvent is allowed to evaporate. The metal center is mostly coordinated to urea via the carbonyl oxygen as shown by FTIR data. The gel was then heated under N2 flow at 800 °C for 3 hours at a very slow rate of 3 °C min−1. A mechanism has been proposed based on thermal gravimetric analysis (TGA) and powder X-ray diffraction (XRD analysis. First, a glassy intermediate with nanoscale anatase and rutile titania is formed at 400 °C. Second, at 600 °C, loss of oxygen and metal reduction/nitridation is observed. Higher temperatures (800 °C) are required to obtain a complete recrystallization, see Fig. 7. A key parameter R has been defined as the ratio of urea and titanium. For R = 3, a mixture of TiN and TiO2 is obtained. Pure TiN is retrieved for R = 4, while Ti(C,N) was formed for R = 10. The higher R, the higher the carbon contamination so the optimal ratio for TiN is four. The authors expanded the procedure to many other metal nitrides (VN, NbN, GaN, CrN, Mo2N), but each metal has another optimal value for R.69 Later, the method was also applied to produce Sn3N4,70 HfN,47 and ZrN,46 as well as other nanocomposites such as graphene g-C3N4 doped with TiN.71,72 The nitride NCs obtained via this process are typically aggregated powders. Finally, other nitrogen sources were explored such as guanidine carbonate and barbituric acid. However, both features a particularly low solubility and it was not possible to achieve R > 1.69
Fig. 7 XRD pattern of titanium–urea gels (R = 10) treated at different temperatures. The * symbol refers to crystalline TiO2. Reprinted with permission from ref. 69. Copyright 2009 American Chemical Society. |
In 2019, Karaballi and coworkers also synthesized TiN, ZrN and HfN NCs from their parent oxide NPs.40 Their strategy consists of a high temperature (1000 °C) solid-state metathesis reaction of magnesium nitride with commercial TiO2 (17 nm), ZrO2 (19 nm), and HfO2 (43 nm).
(3) |
The final product is treated with 1 M HCl to remove MgO and unreacted Mg3N2, and could be dispersed in water.40 All compounds have the expected cubic crystal structure and the nanocrystal size is 12, 15 and 11 nm for TiN, ZrN and HfN respectively. The NCs seem aggregated on transmission electron microscopy (TEM) but it is unclear whether this occurred during the sample preparation or already in solution. The reaction could be easily scaled to 1 g of metal nitride. Efforts to significantly increase the particle size by using larger oxide NCs were unsuccessful, indicating new nucleation events of the nitride, rather than a simply anion exchange reaction. The localized surface plasmon resonance of the NCs is blue-shifted and more narrow in the series: TiN, ZrN and HfN (Fig. 1). Finally, all particles were slightly oxidized at the surface and this effect was worse for ZrN and HfN, compared to TiN. This scalable strategy of Karaballi et al. appears to produce high quality group 4 nitrides. While they are highly crystalline, their size dispersion could be improved. The method has already been adopted by other groups to produce, e.g., plasmonic HfN nanocrystals.73
In 2017, Kan et al. explored molten salts (here: a eutectic mixture of MgCl2 and NaCl) as a reaction medium for the synthesis of TiN from TiO2 and Mg powder under N2 atmosphere. The use of a higher Mg equivalent versus TiO2 (up to 4 equivalent), higher heating temperature (up to 1100 °C), or longer reaction time (up to 7 h) leads to phase pure TiN with a better crystallinity as judged by the well-defined diffraction peaks obtained by XRD. The authors obtained 5–30 nm aggregated TiN NPs by using 2.5 equivalent of Mg heated at 1100 °C for 3 h. The authors described the mechanism as being a dissolution–precipitation type based on the increase of the surface area from the starting TiO2 (9.83 m2 g−1) to the obtained TiN (93.09 m2 g−1).74
VN and NbN nanopowders were synthesized via the urea glass route (see above).69 Agglomerated VN NCs with a crystallite size of 6 nm have been synthesized by Choi et al. via a two-step method. First, ammonolysis of VCl4 forms amorphous V(NH2)3Cl. Second, the amorphous precursor is crystallized at 400 °C in the presence of NH4Cl under an ammonia atmosphere.80 This method is one of the main chemical methods to form VN NCs,81 as well as NbN NCs.82
In 1972, Ta3N5 powder and thin films were obtained at 1100 K starting from ammonia and Ta2O5 powder or Ta thin films respectively.83 This strategy was recently pursued to prepare Ta3N5 nanotubes.84 In 1995, TaN films were made using the CVD method with tert-butylimidotris(diethylamido) tantalum as precursor at temperatures ranging from 450 °C to 650 °C.85 Nanocrystalline TaN with a cubic phase has been synthesized via a solid combustion synthesis using K2TaF7, NaN3, and NH4F precursors under high temperatures up to 850 to 950 °C.86
In 2011, Ho et al. have reported three methods for the synthesis of colloidal Ta3N5 NCs.87 In their first method TaCl5 is dispersed in octadecene (ODE). At 310 °C, tris(trimethylsilyl)amine ((TMS)3N, dissolved in trioctylphosphine) is injected and the particles are left to crystallize for 20 min at 300 °C.
(4) |
The resulting NCs (23 nm, see Fig. 8A) could not be redispersed in any solvent after precipitation due to a lack of strongly binding ligands. Smaller (10 nm) NCs were formed at a reaction temperature of 200 °C. The authors report that the synthesis is sensitive to the purity of TaCl5; at least 99.99% purity is required, and even then the yield was very low (which was attributed to the low solubility of TaCl5 in ODE). In their second method, tantalum chloride was dissolved in a mixture of oleylamine (OLAm) and trioctylphosphine oxide (TOPO). Li3N was added at 300 °C.
(5) |
Fig. 8 TEM images of Ta3N5 obtained by Ho et al.87 (a) NPs obtained with TaCl5 and (TMS)3N (scale 20 nm), (b) with Ta(NMe2)5 and (TMS)3N (scale 20 nm) (c) with TaCl5 and Li3N (scale 10 nm). (d) Absorption spectrum of the obtained 10 nm Ta3N5. The inset showing the two limiting regimes. Adapted with permission from ref. 87. Copyright 2010 American Chemical Society. |
The resulting NCs (2–8 nm after 20 min reaction, see Fig. 8B) could be redispersed in hexane after precipitation with methanol. When the reaction is left to proceed for 1 hour, 10 nm NCs are obtained. Their absorption spectrum has two absorption onsets (467 nm and 515 nm) which is explained by the absorption of Ta3N5 (515 nm) and that of a higher bandgap material, likely an oxidized shell. XPS data indeed confirm the presence of oxidic tantalum. To counter the low yields in the previous methods, the authors devised a third method. Pentakis(dimethylamido)tantalum and (TMS)3N were dissolved in ODE and injected in pure ODE at 300 °C (Fig. 8C). However, the formed particles were amorphous and highly sensitive to air.87 For all three methods, the crystallinity of the resulting particles was solely assessed by TEM and an ensemble measurement (e.g., XRD) is missing.
(6) |
In 2013, a sol–gel pathway was demonstrated with Ta(NMe2)5 and n-propylamine.88 Transamination occurs readily but the condensation appears to be slower than for titanium (see section on titanium nitride for details on transamination and condensation). Upon solvent evaporation a gel was formed. Solvent removal by vacuum resulted in a xerogel that converted in an ammonia atmosphere at 800 °C into Ta3N5 powder with a crystal size of 20 nm. Thin films were obtained by coating the a silica slides with the sol. Firing at 600 °C leads to TaN while heating to 800 °C yields Ta3N5.88
It is thus clear from the above that nanocrystals of the group 5 nitrides have not been convincingly obtained in colloidal form. While this is the first challenge for the future, the second challenge is to control the oxidation state and the polymorphism of these materials.
Fig. 9 Crystal structure of Ni3N. The nickel atoms are represented in dark red and the nitrogen atoms in cyan and white. Structures plotted using VESTA.48 Cif file obtained from ref. 89. |
Recently a one-step synthesis of colloidally stable Ni3N NCs has been reported by Shanker et al.97 Nickel acetate is used as precursor and OLAm is chosen as ligand. Different nanocrystal sizes are obtained (5 to 80 nm) by varying the time and temperature in a narrow window, see Fig. 10. The NCs are reasonably monodisperse and colloidally stable. The authors detect ammonia during the reaction and thus infer that ammonia is the nitrogen source. However, it is unclear how ammonia is being generated since the most logical reaction would be the formation of oleylacetamide and NiO. Furthermore, the synthesis requires a specific reagent purity and supplier.
(7) |
Fig. 10 TEM images of the different Ni3N NCs obtained. (a) 5.6 ± 0.9 nm, (b) 10.6 ± 1.0 nm, (c) 36.6 ± 0.8 nm, (d) 56.4 ± 6.5 nm, and (e) 76.4 ± 3.8 nm. (f) High resolution transmission electron microscopy (HRTEM) imageof a Ni3N nanocrystal with a size of 56.4 ± 6.5 nm. Reprinted with permission from ref. 97. Copyright 2021 American Chemical Society. |
Fig. 11 Crystal structure of Cu3N and the doped Cu3PdN. The copper is represented in dark red, the palladium in grey and the nitrogen atom in cyan. Structures plotted using VESTA.48 Cif file obtained from ref. 111 and 112. |
In 1939, Juza et al. synthesized Cu3N powders by reacting CuF2 with ammonia gas (aminolyis reaction).113 Phase pure copper nitride was obtained at reaction temperatures between 250 °C and 325 °C (6 hours reaction time).114 At higher temperatures, Cu(0) is formed. Alternatively, Zachweija and Jacobs synthesized Cu3N by thermal decomposition of [Cu(NH3)x](NO3).111 In 2005, Choi et al. combined copper(II) chloride and sodium azide in superheated toluene or THF.109 The intermediate copper azide decomposes to Cu3N powder at 185 °C.
(8) |
(9) |
The reaction temperature is remarkably lower than previously reported for the similar synthesis of other metal nitrides such as TiN and GaN. Unfortunately, the reaction time is very long (about three days) and metal azides are thermally unstable and shock sensitive, which limits the reaction to small scales. Finally, the obtained Cu3N is not phase-pure and some NaCl impurities remain present in the sample.
(10) |
The authors hypothesize that NO2 and O2 are formed as side products but did not provide experimental evidence. This hypothesis was based on the solid state decomposition reaction of Cu(NO3)2·3H2O into CuO, NO2 and O2 at around 180 °C.116 The size of the NCs is tuned by using different reaction temperatures (220–280 °C) and reaction times (5–20 min). The conditions and their outcomes are also summarized in Table 2. Pure Cu3N NCs are only obtained with 32 equivalents of ODA. Cu2O is obtained with 8 equivalents, while 80 equivalents lead to Cu(0) NCs. Forty and thirteen equivalents lead to a mixture of Cu/Cu3N or Cu3N/Cu2O mixture respectively. Please note that the copper concentration also changes in these experiments since ODA is the solvent. Wang et al. argue that nitrate (NO3−) is the nitrogen source for the nitride (N3−), even though this would involve the transfer of eight electrons to N(V) in nitrate. They based their argument on the observation that copper nitrate is the only precursor that delivers copper nitride. Copper chloride does not yield any products while copper acetate yields a Cu(0)/Cu2O mixture and copper sulfate delivers copper sulfide NCs. However, no other hypotheses were tested or disproved, thereby leaving the option open that the nitride is provided by ODA, as is the case for certain syntheses of InN NCs (see below).
Precursora | [Cu] (mM) | Solventb/ligandc | Ligand/Cu ratio | Temperature (°C) | Time (min) | Morphology | Phase | Size (nm) | Ref. |
---|---|---|---|---|---|---|---|---|---|
a HMT = hexamethylenetetraamine, PPC = pyrrole-2-carbaldpropyliminato Cu(II). b ODE = 1-ODE. c ODA = octadecylamine, OLAm = oleylamine, HDA = hexadecylamine. | |||||||||
Cu(NO3)2·3H2O | 101 | ODA | 32 | 240 | 10 | Cubic | Cu3N | 15 | 115 |
Cu(NO3)2·3H2O | 101 | ODA | 32 | 280 | 5 | Spherical | Cu3N | 10 | 115 |
Cu(NO3)2·3H2O | 101 | ODA | 32 | 260 + 240 | 5 + 5 | Spherical | Cu3N | 15 | 115 |
Cu(NO3)2·3H2O | 101 | ODA | 32 | 220 + 240 | 5 + 5 | Spherical | Cu3N | 15 | 115 |
Cu(NO3)2·3H2O | 101 | ODA | 32 | 220 | 20 | Irregular | Cu3N | 50 | 115 |
Cu(NO3)2·3H2O | ODA | 40 | 240 | 10 | Cu + Cu3N | 115 | |||
Cu(NO3)2·3H2O | ODA | 13 | 240 | 10 | CuO + Cu3N | 115 | |||
Cu(NO3)2·3H2O | 24 | ODE/ODA | 67 | 250 | 30 | Cubic | Cu3N | 26 | 117 |
Cu(NO3)2·3H2O | 24 | ODE/HDA | 69 | 250 | 30 | Cubic | Cu3N | 19 | 117 |
Cu(NO3)2·3H2O | 24 | ODE/OLAm | 63 | 250 | 30 | Cubic | Cu3N | 11 | 117 |
Cu(NO3)2·3H2O | 24 | OLAm | 250 | 30 | Irregular | Cu3N | 117 | ||
Cu(NO3)2·3H2O | 25 | ODE/OLAm | 31 | 240 | 15 | Cubic | Cu3N | 118 | |
Cu(NO3)2·3H2O + Pd(acac)2 | 25 | ODE/OLAm | 31 | 240 | 15 | Cubic | Cu3PdN | 16 | 118 |
Cu(NO3)2·3H2O + Pd(acac)2 | 25 | ODE/OLAm | 31 | 190 | 30 | Cubic | Cu3PdN | 119 | |
Cu(NO3)2·3H2O | 25 | ODE/OLAm | 31 | 210 | 30 | Cubic | Cu3N | 119 | |
Cu(NO3)2·3H2O | 75 | ODA/OLAm | 45 | 240 | 20 | Cubic | Cu3N | 25 | 105 |
Cu(NO3)2·3H2O | 88 | ODA/OLAm | 41 | 240 | 30 | Cubic | Cu3N | 25 | 103 |
Cu(NO3)2·3H2O | 88 | ODA/OLAm | 41 | 250 | 30 | Cubic | Cu3N | 20 | 103 |
Cu(NO3)2·3H2O | 88 | ODA/OLAm | 41 | 260 | 30 | Cubic | Cu3N | 10 | 103 |
Cu(NO3)2·3H2O | 88 | ODA/OLAm | 41 | 230 | 30 | Spherical | Cu3N | 20 | 103 |
Cu(NO3)2·5H2O + HMT | 50 | Hexanol | — | 200 | 60 | Spherical | Cu3N | 80 | 122 |
Cu(acetate)2·H2O + NH3 | 10 | Nonanol | — | 190 | 60 | Spherical | Cu3N | 200 | 123 |
PPC | 50 | ODA | 74 | 260 | 5 | Spherical | Cu3N | 3 | 124 |
Cu(NO3)2·3H2O | 50 | ODA | 74 | 260 | 5 | Cubic | Cu3N | 19 | 124 |
Cu(OMe)2 | 92 | BnNH2 | 117 | 140 | 15 | Spherical | Cu3N | 2 | 125 |
CuI + KNH2 + NH3 | 21 | Pyridine | — | −35 + 130 | 10 + 30 | Spherical | Cu3N | 4 | 126 |
Also in 2011, Wu and co-workers described the formation of Cu3N from Cu(NO3)2·3H2O in the presence of both a primary amine (surfactant) and octadecene (ODE) (solvent), see eqn (11).117 By changing the primary amine (ODA, hexadecylamine or OLAm), the size of the final Cu3N varies from 26 ± 5.6 nm to 18 ± 2.3 nm and 10 ± 2.7 nm respectively (Fig. 12). All NCs have a cubic morphology and are phase-pure copper nitride.
(11) |
Fig. 12 TEM images and the corresponding size histograms of Cu3N NCs synthesized from copper nitrate and various primary amines in ODE solvent, according to ref. 117. (a, b) Octadecylamine (c, d) hexadecylamine, and (e, f) oleylamine. Reprinted with permission from ref. 117. Copyright 2011 American Chemical Society. Note that the authors labelled the axis diameter. Given the cubic morphology of the nanocrystals, this is confusing and the cube edge length is a more appropriate measure for the size. |
In a first stage, the reaction is heated to 150 °C with a concomitant color change from dark blue to green and then yellow, indicating the reduction of Cu(II) to Cu(I).115 Subsequently, the solution is heated to 250 °C for 30 min and the solution turns brown, consistent with the formation of copper nitride. According to Wu et al. ODE is necessary as solvent to obtain well-defined nanocubes since irregular and polydisperse Cu3N NCs have been obtained using only OLAm. Most likely, other solvents would be more suitable since ODE polymerizes during nanocrystal synthesis.27 Nonetheless, this method was adopted by other groups and Vaughn II et al. used it to produce Cu3N and Cu3PdN, albeit with half the number of amine equivalents, see also Table 2.118 The reaction mixture is degassed for 10 min at 120 °C and then heated to 240 °C under argon for 15 min. To produce Cu3PdxN, Pd(acac)2 is also added at the start. The average diameter of the quasicubic Cu3PdN NCs is 16 ± 2 nm. Cu3N and Cu3PdN can also be grown on Pt and Au nanoparticle seeds to form nanoscale heterostructures at 190 °C see Fig. 13.119
Fig. 13 Time-dependent growth studies of Cu3PdN on Pt nanocubes at 170 °C for (a–f) 5 min, (g–l) 10 min, and (m–r) 15 min. The data for each time point include (a, g, and m) TEM images, (b, h, and n) HAADF-STEM images, (c–e, i–k, and o–q) energy dispersive X-ray spectroscopy (STEM-EDS) element maps for Pt (red), Cu (blue), and Pd (green), and (f, l, and r) HRTEM images of individual particles. Reprinted with permission from ref. 119, Copyright 2019 American Chemical Society. |
In 2019, Braman et al. synthesized Au decorated Cu3N NCs. First, Cu3N NCs (10 ± 5 nm) were obtained via a similar approach to the one of Vaughn et al. but with a higher amine equivalent and a lower heating temperature (210 °C). Subsequently, Au NPs were grown on top of the Cu3N cubes. The resulting heterostructure demonstrated a better catalytic activity than pure Cu3N.120
In 2014, Xi et al. performed the reaction in a mixture of ODA and OLAm.105 First the mixture is heated to 110 °C and than quickly heated to 240 °C for 10 min. They obtained uniform nanocubes of Cu3N of around 20 nm. The growth mechanism was investigated by analyzing reaction aliquots at 2, 5 and 10 min. From TEM measurements (Fig. 14), Xi et al. conclude that first small NPs are formed, which subsequently ripen into bigger nanocubes. This growth mechanism was later confirmed under slightly different conditions.121 In addition, long reaction times (at 260 °C) lead to the decomposition of the formed Cu3N to Cu(0) particles. Xi et al. also studied the magnetic properties of the Cu3N NCs and observed paramagnetic and ferromagnetic contributions. Such properties are not expected for d0 materials and are assumed to originate from surface defects.105
Fig. 14 (A) TEM of the Cu3N NCs synthesized according to ref. 105, after 2, 5 and 10 min. (B) The proposed growth procedure of Cu3N nanocubes. Reproduced from ref. 105 with permission from the Royal Society of Chemistry. |
In 2019, Yin et al. used the same strategy of Xi et al. By tuning the heating temperature, the size and morphology of the final Cu3N is changed, see Table 2.103 Above 240 °C cubic NCs are formed while below 240 °C spherical NCs are formed. The size of the final Cu3N decreases from 25 nm to 10 nm by increasing the reaction temperature from 240 °C to 260 °C. The Cu3N NCs are stable even after 2 weeks of air exposure and no degradation is observed in the crystal structure and morphology.
(12) |
The standard synthesis is performed at 140 °C under inert atmosphere, in a pressure tube. Over the course of the reaction, the color changes from dark blue to red. The average crystal size (determined from XRD) is quite insensitive to the reaction conditions (80 °C–180 °C, 5–60 min) and varies from 1.8 to 2.8 nm. The authors further suggested a precursor conversion mechanism by analyzing the reaction by-products with gas chromatography–mass spectrometry (GC-MS), see Fig. 15. In the first step, copper(II) methoxide oxidizes BnNH2 to benzenemethanimine and is reduced to Cu(I). This proposal is based on the detection of methanol in GC-MS. This reactivity is special for Cu(OMe)2 since other Cu(II) salts such as Cu(II) chloride make stable complexes with amines and fail to produce Cu3N. The benzenemethanimine intermediate reacts with benzylamine, yielding N-benzylidenebenzylamine (detected by GC-MS) and ammonia. The in situ formed ammonia reacts with Cu(OMe) to form Cu3N. Traces of benzonitrile are also observed at the GC, which is justified by further oxidation of benzenemethanimine in presence of Cu(II). This synthetic strategy was adopted by Liu and co-workers, but also requires a specific reagent purity and supplier.128
Fig. 15 Proposed reaction mechanism of Cu(OMe)2 with BnNH2 that leads to the formation of Cu3N.125 Molecules in red are the one detected by GC-MS. |
(13) |
By condensation of the copper amide under reflux conditions, small copper nitride NCs are obtained (4.2 ± 0.7 nm), see Fig. 16. Only pyridine is observed at the surface of the NCs according to FT-IR and elemental analysis, which could explain why these NCs are air and moisture sensitive.
Fig. 16 TEM images of Cu3N NCs synthesized by sol–gel chemistry according to ref. 126. Reproduced from ref. 126 with permission from the Royal Society of Chemistry. |
Alternatively, Nakamura et al. have formed Cu3N from copper(II) acetate in nonanol by bubbling ammonia through the mixture at 190 °C for 1 h.123 The final NCs are not colloidally stable and form aggregates of about 200 nm with crystallite sizes of 20–30 nm. The same colour changes are observed; from blue to yellow to dark brown. Single phase CuO is formed by using shorter chain alcohols such as pentanol, hexanol and heptanol at 130 °C. With increasing temperature, an increasing amount of copper nitride is obtained and from 170 °C pure copper nitride is formed. The authors propose the following mechanism. First, ammonia forms a complex with Cu(II), evidenced by a color change from green to blue. Second, the authors hypothesize Cu(II) is reduced by nonanol. The color change to yellow is consistent with a Cu(I) species but the authors do not detect the direct product, nonyl aldehyde. They do detect 9-ODEamide, of which it is unclear how it is exactly formed. Finally, the Cu(I) species reacts with ammonia to form copper nitride. In the background, there is also the esterification of acetate with nonanol to nonyl acetate. According to the mechanism, a lot of protons are released but the authors did not try to uncover the fate of these protons. Nor were alternative hypotheses considered such as the reduction of Cu(II) by ammonia itself.
Fig. 17 Proposed reaction mechanism for Cu3N formation from hexamethylenetetraamine and Cu(NO3)2·5H2O.122 |
Also in 2018, Sithole et al. synthesized a novel copper precursor: pyrrole-2-carbaldpropyliminato Cu(II) (PPC), see Fig. 18.124 PPC is synthesized from pyrrole-2-carboxaldehyde, copper acetate and propylamine in deionized water, filtered and recrystallized. In a second step, PPC is heated with ODA to form Cu3N (Fig. 18). The resulting spherical NCs are about 1.5–4 nm in diameter, in contrast to the 18 nm cubes that are obtained with copper nitrate under identical conditions. The photoluminescence of the smaller Cu3N NCs is slightly blue-shifted to the emission of the larger NCs, which could possibly be ascribed to quantum confinement. Indeed, we estimated from the calculated dielectric constant and the effective hole and electron masses, an exciton Bohr radius of about 3 nm. However, a direct relation between size and optical properties is currently missing.
Fig. 18 UV-Vis absorption and photoluminescence spectra of Cu3N obtained via (a) PPC and (b) Cu(NO3)2·3H2O method. Reprinted from ref. 124 with permission from the Royal Society of Chemistry. (c) Reaction mechanism proposed for Cu3N formation from the reaction of pyrrole-2-carbaldpropyliminato Cu(II) (PPC) and ODA shown below.124 |
Fig. 19 (a) Crystal structure of Zn3N2 metal nitride. The metal is represented in purple and the nitrogen atom in grey. Structures plotted using VESTA.48 Cif file obtained from ref. 130. (b) Toluene solutions of OLAm capped Zn3N2 NCs sampled at different reaction times and excited with a 365 nm lamp. (c) Absorption (thin lines) and emission (thick lines) spectra of Zn3N2 NCs with mean diameters of 2.1 ± 0.4 nm (solid lines), 4.6 ± 2.3 nm (dashed lines) and 7.1 ± 2.0 nm (dotted lines). Reproduced from ref. 136 with permission from the Royal Society of Chemistry. |
Various physical methods or nitridation approaches have been employed to synthesize Zn3N2 as thin films,137 powders,138 hollow balls,139 or nanowires.134 One chemical approach was reported by Taylor et al. where diethylzinc was reacted with an ammonia flow in ODE, with OLAm as the capping ligand.
(14) |
The size is tuned by the stepwise addition of the diethylzinc precursor. Thus, NCs emitting in the range 500–1100 nm were obtained,140 with a photoluminescence quantum yield greater than 50%. The authors further investigated the effect of the ratio between the diethylzinc to ammonia on the growth. A lower ammonia flow rate or higher diethylzinc injection promotes growth and large particles are obtained. Higher ammonia flow rates lead to smaller NCs.136
GaN and InN nanocrystals are considered the possible next generation of QDs. While the exciton Bohr radius of gallium nitride is 3.2 nm, the exciton Bohr radius of indium nitride is 8 nm. Together with the formation of solid solutions (InxGa1−xN), this provides ample opportunity to tune the optical properties using either quantum confinement or compositional control. GaN is being investigated for high frequency and high power transistor design.147,148 InN is a promising compound for optoelectronic applications due to its low bandgap, small effective mass and efficient electron transport. InN is also considered for low-cost, high-efficiency solar cells, photomasks, laser diodes, sensors, and biological imaging.149–153 InN has a complex plasmonic behavior due to a nonparabolic dispersion of its conduction band.154,155
In2O3 + 2NH3 → 2InN + 3H2O |
GaN NCs in a silica matrix were synthesized by first forming an oxide gel of gallium and silicon. The gel is then annealed at 900 °C for 5 h under NH3 atmosphere.157 Variations on this strategy exist where pure GaN nanocrystalline powders are formed by annealing Ga2O3, GaO(OH), or NH4[Ga(OH)2CO3] under an NH3 atmosphere at 900–950 °C.158–161 Similar to the case of TiN, the solid state reaction between Ga2O3 (or γ-Al2O3) and cyanamide CN2H2 yields GaN (or AlN) NCs.67,162 Finally, the urea glass route was also successful in synthesizing GaN nanopowders.
Janik et al. formed GaN NCs via pyrolysis at 450–500 °C of a gallium imide [Ga(NH)3/2]n precursor (obtained by reaction of [Ga(NMe2)3]2 and NH3).166 In 1999, this method was improved by crystallizing the [Ga(NH)3/2]n precursor in hot trioctylamine (360 °C for 24 h) under a flow of ammonia.167 During cooldown of the reaction mixture, a solution of hexadecylamine in trioctylamine was added, which improved the colloidal stability of the final GaN NCs. These NCs have a diameter of 3 nm and feature a first exitonic emission at 3.65 eV, higher than the bulk bandgap of GaN.
β-GaN NPs of 3–8 nm in size were obtained via microwave heating of Ga(NH2)3 in ionic liquid at 300 °C for 1 h. The colorless suspension of Ga(NH2)3 was obtained by dissolving GaCl3 in liquid NH3 at −35 °C and adding KNH2. The obtained NCs show a bandgap around 3.9 eV which is 0.6 eV above the β-GaN bulk bandgap. The authors attributed this blue shift to quantum confinement. The fluorescence properties of the as-synthesized NCs are shown in Fig. 20.168
Fig. 20 (a) Emission spectrum at 274 nm and excitation spectrum at 523 nm of the synthesized GaN. (b) Photographs of the as synthesized GaN NCs in daylight and under excitation at 254 nm. Reprinted from ref. 168 with permission from the Royal Society of Chemistry. |
In 2006 Dingman et al. reported the synthesis of InN nanofibres at low temperatures (203 °C), catalyzed by nanometer sized droplets of In metal (also called a solution–liquid–solid (SLS) mechanism).169 The authors prepared two different indium precursors (iPr2InN3 and tBu2InN3) by reacting trialkylindium first with methanol and subsequently with trimethylsilylazide. Crystalline InN was then formed by reacting the as mentioned precursors with 1,1-dimethylhydrazine (H2NNMe2) at 203 °C for 20 h. The role of 1,1-dimethylhydrazine is to reduce part of the indium precursors to In droplets, which then catalyze the InN crystallization.
(15) |
(16) |
(17) |
In 2006 Choi et al. synthesized InN NCs from indium bromide and sodium azide under solvothermal conditions.170 The indium azide intermediate decomposes into InN and N2.
(18) |
(19) |
The solvent used during the solvothermal treatment determines the final properties of the NCs. While toluene delivered 10 nm aggregated wurtzite InN, hexadecane yielded poorly crystalline InN. The reaction typically takes 3–5 days during which the reaction temperature was increased step by step to avoid explosive azide decomposition. Also InxGa1−xN alloyed particles were synthesized via this strategy.170
The reaction of potassium amide (KNH2) and indium iodide in liquid ammonia yields an insoluble indium amide product. The latter is probably highly condensed, featuring imido bridges. Decomposition of this insoluble product at 400 °C under nitrogen atmosphere delivers InN, In, KNH2, and NH3.171 Alternatively, InI3 and NaNH2 were reacted in anhydrous benzene in a Teflon-lined pressure bomb.172 The crystallite size of the resulting InN NCs varies from 9.9 ± 5 nm to 35.1 ± 10 nm by increasing the reaction time or temperature. At ambient pressure, InN NCs (6.2 ± 2 nm diameter) have been synthesized using InBr3 and NaNH2 in hexadecane under an ammonia atmosphere.
(20) |
A post-synthetic acid treatment (to remove In(0)) and functionalization with OLAm yielded colloidally stable particles, see Fig. 21.173 This method has been further studied by Palomaki et al. where they use In2S3 instead of InBr3 and ODE as their solvent.174
(21) |
Fig. 21 TEM images of the colloidally stable InN obtained by Hsieh et al. after acid treatment. (a) Low magnification image, (b) high resolution image and (c) diffraction pattern obtained via fast Fourier transform (FFT). Reprinted with permission from ref. 173 with permission from the Royal Society of Chemistry. |
To improve the size dispersion, size-selective precipitation was performed by the dropwise addition of acetonitrile to OLAm-capped InN NCs, and centrifugation. The final nanocrystal size was tuned from 4.5 to 9.9 nm by varying the reaction time from 15 min to 12 h at 220 °C. The NCs size also increases from 4.1 to 8.0 nm by increasing the reaction temperature from 175 °C to 200 °C (at a fixed reaction time of three hours). Below 175 °C no InN was formed.
When InCl3·4H2O, NaNH2, and sulfur were reacted in a pressure bomb at 160 °C for 30 h, wurtzite InN crystals (50–200 nm) were obtained.175 According to XRD, the reaction proceeds through a In2S3 intermediate. Even larger particles (500 nm) were obtained using In(NO3)3·5H2O as precursor.
Fig. 22 TEM images of the NPs. (a) Overview of the as prepared sample, showing the InN branches stemming out of larger In(0) NPs. (b) HRTEM image of one of the InN branches in (a) at the surface of the In(0) nanoparticle. XRD pattern for the (c) as-prepared sample and (d) acid treated sample. The vertical lines show the XRD patterns of bulk zinc blende-InN (red) and tetragonal In(0) (blue). Reprinted with permission from ref. 176. Copyright 2016 American Chemical Society. |
Intrigued by the high dissociation energy of C–N bonds (330 kJ mol−1), Beaulac et al. investigated the mechanism of this reaction by analyzing the by-products via1H NMR, 13C NMR, 2D COSY NMR, and mass spectrometry.177 The authors propose that oleylamide is oxidized to oleylimine, thereby reducing In(III) to In(0), see Fig. 23. A second equivalent of oleylamide performs a nucleophilic attack on the imine, yielding a secondary aldimine and releasing NH2−. The latter goes on to react with In(III) to form InN. The proposed mechanism is based on the observation of secondary aldimine, ammonia, hydrogen, InN and In as reaction products. Oleylimine is too unstable to be isolated. In(0) is always formed in larger yields than InN, meaning that not all reactive NH2− are converted to InN.
Fig. 23 Overall reaction. Reaction 1a, alkylamide precursor oxidation; reaction 1b, formation of the secondary imine leading to NH2− production; reaction 2, nitride formation; R1, R2, R3, three hypothetical reduction processes concomitant with reaction 1a; S1, detrimental hypothetical side reaction. Orange, reactants; blue, intermediates (not isolated); red, identified products. Uncoordinated forms depicted here for simplicity. Reprinted with permission from ref. 177. Copyright 2018 American Chemical Society. |
In 2019, Chang Choi et al. designed a hot-injection strategy to obtain GaN QDs.180 Gallium chloride and stearic acid were dissolved in ODE and heated to 280 °C under argon. Subsequently, a solution of lithium bis(trimethylsilyl)amide in hexane was rapidly injected.
(22) |
The nanocrystal size was hard to determine from TEM but surely below 5 nm, which is also confirmed by the very broad nature of the XRD reflections. The PL emission of the QDs could be tuned from 315–355 nm, while the bulk PL is at 365 nm. Successful doping with zinc red-shifted the PL to 330–460 nm.180
(23) |
Since group 4 and 5 nitrides typically require high temperatures for crystallization, the logical conclusion is that their crystallization rate is low. Therefore, one should design nitride precursors that decompose even slower, or to apply strategies that enhance the rate of crystallization by lowering the barrier for bond forming and bond breaking. To move the field forward in a rational way, we need to understand the mechanism of the colloidal syntheses that do provide satisfactory results. While the precursor conversion has been studied for InN, there are many systems for which the mechanism is left unexplored. Even more problematic, the crystallization part of the mechanism has not been studied for any of the nitrides discussed here. However, the crystallization is likely the most difficult step in the whole process and thus the most critical to understand to make progress. Finally, it will be equally important that researchers treat colloidal synthesis as a true chemical reaction that has a chemical equation (which needs to be written down) and for which an accurate yield needs to be reported. By putting more rigor into writing the Experimental section, we can enhance the reproducibility of the field. Given the use of early transition metal nitrides for many applications such as catalysis, photothermal therapy, superconductors, and electrochemical capacitors, synthetic efforts should focus on providing these NCs as size-tunable colloidal particles. The field of colloidal metals nitrides is clearly challenging and still young, but adding colloidal nitrides (with synthetic control) to the portfolio of colloidal materials would open up many exciting possibilities for the future of materials science and nanotechnology.
This journal is © The Royal Society of Chemistry 2021 |