Ag₂S is a semiconductor with a narrow band gap of 1.08 eV [34]. When AgTB precursor was stirred with octylamine (OA) in toluene at room temperature, monodispersed Ag₂S nanocrystals can be isolated after 3 hours (Figure 2a). Typical XRD pattern of the nanoparticles prepared from AgTB is shown in Figure 2d. The diffraction pattern revealed good monoclinic crystallinity and fitted well with the α-phase of bulk Ag₂S (JCPDS 14-72). This is known to be the stable silver sulfide phase which commonly exists at room temperature.

The same α-phase silver sulfide nanocrystals were obtained when the reaction was carried out using other amines such as dodecylamine (DDA) and oleylamine (OLA). TEM analysis showed that spherical nanoparticles with reasonable size distribution were produced at room temperature in all these cases. The Ag₂S nanocrystals obtained from different amines, however, are slightly different in sizes as shown by the size histograms in Figure 2a–c. Thus, average particle diameters of 9.2 ± 1.9, 8.3 ± 1.5 and 7.5 ± 0.9 nm are obtained for reaction with OA, DDA and OLA respectively. In conclusion, the Ag₂S particle sizes can be controlled by varying the chain length of the alkylamine used, and longer-chain amines tend to produce smaller-sized particles.

Typical XPS spectra of the Ag₂S nanocrystals are shown in Figure 3, with binding energies corrected with reference to the C 1s peak at 284.7 eV. The doublet arising from Ag 3d_5/2 and 3d_3/2 was detected at 368.0 and 374.1 eV respectively, while the S 2p photoelectron peak appears at 161.7 eV. These values are close to those of bulk Ag₂S [35]. There is no O 1s peak (531.0 eV) detected on the spectrum, indicating that there is no by-product such as Ag₂SO₄ (368.3 eV) or Ag₂O (368.4 eV) produced. Peak area analysis of the Ag 3d_5/2 and S 2p peaks, after accounting for elemental sensitivity factors, gives an elemental ratio of 1.97:1 for Ag to S.

Similarly, we found that CuTB precursors readily decompose by reacting with alkylamines at room temperature to give uniform Cu_2−xS nanoparticles. TEM analysis (Figure 4a–c) indicated average diameters of 8.1 ± 1.1, 6.1 ± 0.5 and 5.8 ± 0.4 nm for the nanoparticles produced from OA, dioctylamine (DOA) and OLA, respectively.

It is well known that copper sulfides exist in many different phases and compositions (Cu_xS, x: 1 → 2). Non-stoichiometric copper sulfide is readily formed and has been utilized as superionic conductor or p-type semiconductor. XPS peak area analysis gave an elemental Cu to S ratio of 1.72:1 for our samples, thus corresponding to x ~0.28 in the Cu_2−xS general formula. The XRD pattern in Figure 4d clearly revealed the rhombohedral structure, which fits well with the standard bulk Cu_2−xS phase (JCPDS 85-1693).

While DOA also reacts with CuTB to produce copper sulfide nanoparticles at room temperature, we notice that the growth rate is slightly slower and the particles produced are smaller than those prepared with primary amine (i.e., OA). We believe that, since OA is less bulky compared to DOA, it attacks the precursor with less hindrance and thus causes the reaction to occur at a faster rate. In addition, DOA is expected to play a better role to efficiently prevent the nanoparticles from aggregation and thus will tend to produce smaller particles. On the other hand, our experiments confirmed that tri-substituted amine does not result in the formation of Cu_2−xS, probably due to its bulkiness.

β-In₂S₃ is an n-type semiconductor with a band gap of 2.0–2.2 eV [36]. It has promising applications in the preparation of green and red phosphors for photoconductors and photovoltaics [37,38,39]. In addition, it can serve as a host for a number of metal ions to form semiconducting and/or magnetic materials. In₂S₃ nanocrystals have not received as much attention so far, due to the lack of a straightforward preparation methodology. In this case, we have successfully prepared these nanoparticles by decomposing InTB precursor in alkylamines.

Reacting InTB with OA could produce spherical In₂S₃ with an average diameter of 3.7 ± 0.6 nm (Figure 5a). However, reaction of InTB with OLA or DOA could not proceed at room temperature until the addition of a trace amount of a short chain amine. Thus, upon adding propylamine into OLA, In₂S₃ nanoparticles with average diameter of 2.6 ± 0.4 nm could be isolated (Figure 5b). The In₂S₃ nanoparticles prepared from OLA and OA show a UV-Vis absorption onset at 355 nm (3.50 eV) and 437 nm (2.84 eV), respectively (Figure 5c). Compared to the band-gap of bulk In₂S₃ (2.2 eV or 564 nm) [36], it is clear that the excitonic transition is blue-shifted due to strong quantum confinement in these In₂S₃ nanocrystals. Figure 5d shows the XRD pattern of the as-prepared In₂S₃ nanoparticles, confirming the tetragonal β-phase of In₂S₃ (JCPDS 32-0456).

CdS is one of the most studied metal sulfides, due to its various applications. When CdTB precursor was mixed with OA, spherical CdS nanoparticles with an average diameter of 5.3 ± 0.7 nm were produced (Figure 6a). No reaction had happened, however, in the sole presence of OLA at room temperature. Again, CdS nanoparticles with an average diameter of 4.4 ± 0.3 nm were obtained upon adding a small amount of propylamine into the OLA (Figure 6b).

The monodispersity of the prepared CdS nanoparticles is manifested in their absorption spectra, which exhibit a clear distinct band rather than a shoulder or threshold (Figure 6c). The band-edge absorption is blue-shifted, occurring at 432 nm for OA-capped and 424 nm for OLA-capped nanocrystals. These gave an estimated particle diameter of 4.9 and 4.0 nm respectively from the Brus equation, in good agreement with those obtained from TEM analysis. CdS is known to exist in two structures: Cubic zinc blende phase and hexagonal wurtzite phase. In our sample, XRD analysis (Figure 6d) suggested a wurtzite crystal structure (JCPDS 41-1049). This observation is interesting, since CdS nanoparticles with zinc blende structure are often produced in ambient conditions, whereas wurtzite structure is often obtained at high temperatures.

From the above experimental results, we have confirmed that long chain alkylamines such as OA and OLA can react with the thiobenzoate precursors at room temperature to produce monodispersed metal sulfide nanoparticles. Comparatively short chain amines such as propylamine and ethylamine yield larger particles that precipitate quickly from the reaction mixtures. By mixing a small amount of these shorter chain amines into the long chain amines, on the other hand, enables monodispersed nanoparticles to be isolated again. Thus, the long chain amines are needed in this reaction as capping groups to efficiently prevent aggregation of the nanoparticles.

After repeating the reactions using different types of amine, we could also conclude that the types of amine used will affect the rate of formation of the specific nanocrystals. Thus, for OLA and aliphatic amines with chains longer than 14 carbon atoms, In₂S₃ nanoparticles cannot be produced at room temperature. Hence, heating was needed to initiate this particular reaction. On the other hand, the reaction can be induced at room temperature when a small amount of shorter chain amines are added into the long chain amine.

An important aspect of our findings is that MTB precursors are broken down by amines at room temperature. We have found that in other common capping reagents, e.g., TOPO, thiols or carboxylic acids, the decomposition of MTB precursors can only occur at elevated temperatures. We determined the onsets of these reactions by slowly heating up the MTB precursors in these media, and the results are summarized in Table 2. We could hence conclude that the decomposition of thiobenzoates in amines is not a simple pyrolysis process.

In addition to the above, we have also confirmed that tertiary amines such as trioctylamine do not react with the MTB precursors even at reflux conditions. Secondary amines, such as DOA, can produce monodispersed nanoparticles from CuTB and AgTB precursors at room temperature, but not with InTB and CdTB. Thus, it seems that amines with an active hydrogen atom are needed for the reaction to proceed at room temperature.

On the basis of all the above observations, we propose a general reaction mechanism as follows. The reaction probably arises from an initial attack of the amine group onto the electron-deficient (i.e., electrophilic) carbonyl carbon of the MTB precursor [40], and this is followed by an elimination of an amide to produce the metal sulfides (Scheme 1):

For the InTB and CdTB precursor, the reaction with secondary DOA occurs only at elevated temperatures. We believe this is because the bulkier DOA could not access the carbonyl carbon of these di- and tri-substituted MTB precursors readily. As a comparison, we have also performed reactions with shorter chain secondary amines, such as diethylamine and dipropylamine. It was found that reactions of these secondary amines with MTB can occur readily at room temperature. The particles produced, however, are rather large in size, as expected from our proposition that amines are playing dual roles as the nucleophilic reagent and the capping agent. From these results, it can be seen that properties such as coordination ability, steric hindrance, solubility, as well as stability constants are all important factors in the successful preparation of nano-sized crystals from the MTB precursors.

In order to support the proposed mechanism in Scheme 1, we performed DFT calculations by modeling the reaction between MTB precursors with a simple amine-ethylamine. For all the four types of precursors, B3LYP/LanL2DZ optimized transition state structures were located as shown in Figure 7. The zero-point corrected enthalpy change values and energy barriers are tabulated in Table 3. For comparison, important geometric parameters of the optimized ground and transition state structures are presented in Table 4.

Thus, as can be seen in Figure 7, a transition state structure resembling the nucleophilic addition-elimination mechanism proposed is obtained for all the MTB precursors studied. Clearly, a C-N bond (~1.5–1.6 Å in Table 3) is formed between the nucleophilic amine and the electrophilic carbonyl carbon of MTB in the transition state. A simultaneous elimination of amide is also evident from the lengthening of one C-S bond from ~1.8 Å to ~2.7 Å. We notice also that one of the metal-sulfur bonds is shortened during this addition-elimination process towards the formation of metal-sulfur monomer.

In Table 3, we noted that the DFT predicted activation energy values are similar, (~110 kJ mol⁻¹) for all the four precursors studied. This low energy barrier is in agreement with our observation that these reactions can occur readily at room temperature. Moreover, all the reactions were predicted to give an exothermic enthalpy ranging from 4 to 33 kJ mol⁻¹. The calculations further suggested that reactions of AgTB and CuTB will be fast, partly driven by the large exothermic enthalpy change.

In order to further elucidate the role of different amines, we also performed B3LYP/LanL2DZ calculations for reactions mediated by different amines. Using a AgTB precursor as the example, transition state structures similar to that shown in Figure 7a were obtained and the calculation results are summarized in Table 5. Similarly, the formation of a new C-N bond, lengthening of the C-S bond and shortening of the Ag-S bond are predicted.

It is important to highlight that DFT calculations predicted the bond dissociation energy of AgS-COPh as 266.0 kJ mol⁻¹, while that of Ag-SCOPh was predicted as 206.4 kJ mol⁻¹. Thus, our studies suggest that the stronger S-C bond is broken instead of the weaker Ag-S bond in the presence of amine. In Table 5, variations between different amines was predicted to be rather similar, except that the energy barrier is slightly smaller and the exothermic enthalpy is larger in the case of OA and had, as compared to DOA. All these DFT results provided good support to our experimental observations and the proposed mechanism.

In order to further support the proposed mechanism, we monitored the reaction at room temperature using FTIR analysis. Reaction mixtures of the respective MTB precursor with OA were concentrated through rotary evaporation and their FTIR spectra were measured. Figure 8 shows a typical FTIR spectrum of such reaction mixture, together with the spectra of pure OA and the precursor for comparison.

The FTIR spectrum of pure AgTB shows two strong bands at 1602 and 1573 cm⁻¹, arising from C=O stretching. These are slightly lower in frequency as compared to υ(C=O) in thiobenzoic acid, due to its strong coordination to the metal ions [41]. Lower frequency peaks at 915, 901 and 647 cm⁻¹ resulting from the C-S bending vibration and O-C-S deformation were also observed in the pure AgTB spectrum. On the other hand, the FTIR spectrum of pure OA shows a pair of fairly strong asymmetric and symmetric stretching of primary N-H bands at 3376 and 3292 cm⁻¹, and also another N-H bending mode at 1610 cm⁻¹.

In comparison, the reaction mixture AgTB + OA gives a IR peak of the C=O group at 1641 cm⁻¹ (Table 6). This position is characteristic of the −C=O bond stretching in amides [41]. Besides, secondary amides often exhibit a relatively strong bending band at about 1543 cm⁻¹, attributed to a combination of a C-N stretching band and an N-H bending band. Meanwhile, the doublet of N-H is replaced by a single sharp band at 3310 cm⁻¹, typical of the N-H stretching frequency of a secondary amine or an amide. Clearly, the N-H bending mode, existing at 1610 cm⁻¹ in pure OA, has disappeared in the reaction mixture.

Thus, the above FTIR analysis confirms the attack of the OA molecule onto the carbonyl carbon of the AgTB precursors to form an intermediate secondary amide adduct. In addition, it was also found that the vibrational peaks of C-S bending and O-C-S deformation detected in AgTB had disappeared from the mixture. This further confirms the breaking down of these bonds as suggested by DFT calculations in Section 3.3.2.