1 Introduction

Due to their high repletion, low cytotoxicity, and low aggregation, spherical particles are expected to be used in a wide variety of applications, including photonic crystals [1, 2], lasers [3, 4], drug delivery [5, 6], sensing [7, 8], catalysis [9], and solar cells [10–13]. Recently, Koshizaki’s group developed a laser melting in liquid method [14] for fabricating submicron spherical particles from colloidal nanoparticles and prepared various composition spheres [14–24].

The laser melting in liquid method is a technique for producing submicron spherical particles by irradiating colloidal nanoparticles with a non-focused moderately pulsed laser (about 10–300 mJ/cm2). In this technique, the irradiation causes aggregates in the colloidal solution to melt and form droplets; rapid cooling at the end of each pulse causes the droplets to form spherical particles. Submicron spherical particles obtained by this technique are expected to have application in medical materials because nanoparticles have cytotoxicity [25, 26] and submicron particles exhibit the enhanced permeability and retention (EPR) effect [27] in tumor tissues. In optics, various applications of submicron spherical particles are also expected because the sphere sizes are close to the visible light wavelength.

Crystalline spherical particles can also be produced by laser melting in liquids; in contrast, spheronization of crystalline materials, such as semiconductors and metals, is difficult [16]. Therefore, we usually use a method for obtaining spherical particles from aggregates of crystalline nanoparticles [28, 29]. Although this approach has potential in a variety of applications, researchers are concerned that the approach may impair performance in electric, magnetic, photoelectric, and a thermoelectric applications due to the numerous interfaces and poor adhesion of nanoparticle aggregates [16]. Therefore, to fully demonstrate the potential of these materials, it is desirable that each sphere be a single particle. To achieve this, laser melting in liquids is an effective production method. However, the materials used in laser melting in liquids should not chemically react with the solvent, and they should be unaffected by the laser light absorbance. Therefore, applicable materials are currently restricted. Production of metal, oxide, carbide, and alloy spheres has been reported [14–24], but no reports have appeared on nitride production.

Most nitrides have high chemical stability, high hardness, and high melting points. Therefore, they are used in nitriding treatments and cermet materials. Due to their good luminescent properties, high saturation electron speeds, and high breakdown fields, the III–V nitrides are used as semiconductors and luminescent materials. Among the various nitride materials, TiN is particularly important and has broad use. It has high hardness and a high melting point, and it shows biological stability and significant resistance to chemical corrosion. For these reasons, TiN is used for high-hardness machine parts, sliding portions of prostheses, and dental materials. Another important property is that the electrical conductivity of TiN is higher than that of titanium metal. TiN is used in very large-scale integration (VLSI) devices as a highly conductive diffusion barrier. In a related material, TiOxNy, the electrical conductivity can be adjusted by changing the x parameter [30], and it has structural and chemical compatibility with TiN. Therefore, TiOxNy is expected to be used as a new generation of nonvolatile memory structures [31]. TiOxNy has also been studied as an oxygen reduction reaction catalyst [32]. Although TiN has many advantages, there are few reports on the production of TiN spherical particles. Spheronization of TiN should promote its use in situations that can exploit the above-mentioned features.

In this study, the preparation of TiN spheres by laser melting in liquid was performed, and we analyzed the effects of laser fluence, irradiation time, and raw nanoparticle concentration. This is the first attempt at using a nitride as the raw material in this particle production method.

2 Experimental

TiN nanoparticles (Wako Pure Chemicals, c.a. 50 nm, powder form) were dispersed by ultrasonication in a glass vessel containing 16 ml of demineralized water. The TiN colloidal solution was then irradiated with a Nd:YAG pulse laser (532 nm, SHG) operated at 10 Hz with a pulse width of 13 ns. The irradiation was performed at various fluences and for various irradiation times. During irradiation, the dispersion liquid was stirred with a magnetic stirrer. The laser fluence, irradiation time, and raw nanoparticle concentration were 28–350 mJ/cm2, 5–15 min, and 0.25–1.0 mg/ml, respectively. Secondary particle sizes of the raw material colloidal solution were measured by dynamic light scattering (DLS, Sysmex Co. Zetasizer Nano). The size and morphology of product particles were measured by field emission scanning electron microscopy (FE-SEM, Hitachi S4800). The compositions of the spherical particles were evaluated by X-ray diffraction analysis (XRD, Rigaku Ultima IV/PSK), X-ray photoelectron spectroscopy analysis (XPS, Ulvac-Phi Inc. PHI Versa Probe), and scanning transmission electron microscopy energy-dispersive X-ray spectroscopy (STEM-EDS, JEOL Ltd. JEM-2100F). Absorption spectral properties were evaluated with a UV–Vis spectrophotometer (Shimazu MultiSpec1500).

3 Results and discussion

3.1 Effects of laser fluence on particle size and composition

Figure 1 presents SEM images of the as-prepared particles taken under different fluences. Laser fluence, irradiation time, and initial TiN particle concentration were set to 28–350 mJ/cm2, 30 min, and 0.25 mg/ml, respectively. Particles irradiated with 28 mJ/cm2 (Fig. 1b) and laser energy were almost cubic, which was the same shape as the raw TiN (Fig. 1a). This was because the applied laser energy could not completely melt the TiN particles. At or above 77 mJ/cm2 (Fig. 1c–i), spheres were obtained and sphere diameters increased with the fluence. However, at 350 mJ/cm2 (Fig. 1i), the production of large particles decreased and the generation of nanoparticles (NPs) was observed. This was caused by ablation or evaporation due to irradiation by the high-energy-density laser [33]. Sphere diameters were measured from SEM images, and a particle size distribution was created (Fig. 2). The sizes of the particles were divided into two regions: 50–250 nm and >250 nm. Spherical particles seemed to be generated by the following mechanism. First, spherical particles formed with diameters of 50–250 nm from the melting of the aggregated TiN raw material. This assumption was supported by the fact that secondary particle diameters in the raw TiN colloidal solution, as measured by DLS (Fig. 3), were distributed over 50–300 nm. Next, the spherical particles and raw material aggregated and melted to form larger spherical particles with diameters greater than 250 nm. The above mechanism is well known. There are many reports that aggregates in the liquid are essential for spherical particles formation [34–37]. The increase in spherical particle diameters with the laser fluence is explained in Ref. [38].

Fig. 1
figure 1

SEM images of irradiated particles at laser fluences of a raw particles, b 28 mJ/cm2, c 77 mJ/cm2, d 80 mJ/cm2, e 107 mJ/cm2, f 120 mJ/cm2, g 200 mJ/cm2, h 300 mJ/cm2, and i 350 mJ/cm2

Full size image
Fig. 2
figure 2

Size distributions of spherical particles for laser fluences of a 77 mJ/cm2, b 80 mJ/cm2, c 107 mJ/cm2, d 120 mJ/cm2, e 200 mJ/cm2, f 300 mJ/cm2, and g 350 mJ/cm2. The corresponding average sphere sizes were a 131 ± 55 nm, b 114 ± 29 nm, c 188 ± 88 nm, d 138 ± 58 nm, e 262 ± 120 nm, f 282 ± 218 nm, and g 127 ± 132 nm

Full size image
Fig. 3
figure 3

Particle size distribution obtained from DLS for the raw TiN colloids

Full size image

Figure 4 shows the XRD patterns of the spheres taken under different fluences. Spheres were not generated at a fluence of 28 mJ/cm2, but spheronization did occur at 77 and 107 mJ/cm2. At each laser fluence, XRD peaks for TiN appeared. At 28 mJ/cm2, only the peak for the raw TiN material was present. However, at 77 and 107 mJ/cm2, peaks for the rutile-type TiO2 appeared. This confirms that oxidation occurred during the melting and spheronization of the raw material particles. The peak from the rutile-type TiO2 (JCPDS card No. 00-021-1276) at 107 mJ/cm2 is larger than that at 77 mJ/cm2; i.e., the amount of oxide produced increased with the fluence. In addition, a peak for the anatase-type TiO2 (JCPDS card No. 00-021-1272) appeared at the fluence of 107 mJ/cm2.

Fig. 4
figure 4

XRD patterns of products obtained from laser irradiation with 28 mJ/cm2, 77 mJ/cm2, and 107 mJ/cm2

Full size image

Figure 5 shows the EDS mapping images of a sample irradiated with a laser fluence of 80 mJ/cm2. Compared with Ti and N atoms, O atoms appear more often at the edge of the particle; this indicates that surface oxidization occurred.

Fig. 5
figure 5

EDS compositional mappings (Ti, N, O, and N/O overlay) of sample after laser irradiation with 80 mJ/cm2

Full size image

Figure 6 shows the XPS spectra of a sample irradiated with a laser fluence of 77 mJ/cm2. The large peak in the region 454–462 eV is derived from a Ti 2p orbital and comprises several peaks. The binding energies are 459.34, 456–459, and 455.8 eV for Ti–O in TiO2, Ti–O–N, or O–Ti–N originating in TiOxNy, N-doped TiO2, and Ti–N in TiN, respectively [39, 40]. The Ti–N peak was smaller than the others, suggesting that the surfaces of the spheres were oxidized. The XPS characterization was consistent with the results from the EDS mapping and supports the conclusion that surface oxidation occurred. As sources for oxygen, the water used as the solvent, dissolved oxygen in the water [41], and dissolved oxygen in TiN can be considered. Further investigation is needed to establish a detailed mechanism for the surface oxidization.

Fig. 6
figure 6

Ti 2p core level XPS spectra of sample after laser irradiation with 77 mJ/cm2

Full size image

In the UV–Vis spectrum (Fig. 7), as the laser fluence was increased, absorption at 500–800 nm decreased and absorption at 200–400 nm increased. Absorption at visible wavelengths was due to TiN, and that at ultraviolet wavelengths was due to TiO2 [42]. Therefore, the UV–Vis spectrum shows that increasing the laser fluence depleted TiN and generated TiO2 due to oxidation.

Fig. 7
figure 7

UV–Vis absorption spectra of samples obtained from laser irradiation with different laser fluences

Full size image

Figure 8 presents the average sphere diameters measured from SEM images and the phase diagram of TiN particles, as calculated by the heating–melting–evaporation (H–M–E) model [38]. Table 1 presents the data of the optical and thermodynamic properties of TiN which were used to calculate by H–M–E model. This model provides the energy needed for melting and evaporation based on properties that depend on particle size, such as thermal capacity, specific latent heat, and adsorption coefficients. This model assumes that all energy absorbed by a particle from a laser pulse is expended in the particle H–M–E process. If the amount of absorbed energy is expressed by the following formula, then the particle melting starts.

Fig. 8
figure 8

Phase diagram calculated for TiN at 532 nm laser irradiation. Solid line (J1) locates the start of melting; dashed line (J2) indicates the completion of melting. Markers indicate average size and standard deviation of spheres obtained by laser irradiation with corresponding fluences

Full size image
Table 1 Optical properties and thermochemical parameters of TiN
Full size table
$$J\sigma_{\text{abs}}^{\lambda } = m_{\text{p}} \mathop \int \limits_{{T_{0} }}^{{T_{\text{m}} }} c_{p(\rm T)}^{\text{s}} {\text{d}}T$$
(1)

If the irradiated energy increases and the amount of absorbed energy is expressed by the following formula, then the particle is completely melted.

$$J\sigma_{\text{abs}}^{\lambda } = m_{\text{p}} \left[ {\mathop \int \limits_{{T_{0} }}^{{T_{\text{m}} }} C_{p(\rm T)}^{\text{s}} {\text{d}}T + \Delta H_{\text{m}} } \right]$$
(2)

If the absorbed energy is even higher, then the particle evaporating starts.

$$J\sigma_{\text{abs}}^{\lambda } = m_{p} \left[ {\mathop \int \limits_{{T_{0} }}^{{T_{\text{m}} }} C_{\text{p(T)}}^{\text{s}} {\text{d}}T + \Delta H_{\text{m}} + \mathop \int \limits_{{T_{\text{m}} }}^{{T_{\text{b}} }} C_{\text{p(T)}}^{\text{l}} {\text{d}}T} \right]$$
(3)

Here, J = E 0/S 0 is the laser fluence of laser beam with pulse energy, E 0, and cross section, S 0; σabs is the particle absorption cross section; \(c_{\text{p}}^{\text{s}}\) and \(c_{\text{p}}^{ 1}\) are the particle heat capacities in solid and liquid states; m p = ρ p ((πd 3p )/(6)) is the particle mass; T 0 is the particle initial temperature; T m is the melting temperature; T b is the boiling temperature; ΔH m is the heat of melting; and ΔH ev is the heat of evaporation. Therefore, in Fig. 8, the curve J1(dp) corresponds to the start of melting [formula (1)], and curve J2(dp) corresponds to the complete melting of the particle [formula (2)]. Accordingly, the region below the J1 curve represents the solid phase (S), and the domain above J2 is the liquid phase; the domain between J2 and J1 represents states containing both liquid and solid. In plots such as in Fig. 8, a curve corresponding to the start of evaporation (J3) and another corresponding to complete evaporation are usually shown, but these curves were not calculated here because enthalpy of vaporization of TiN is not available to model the evaporation of TiN. In Fig. 8, the actual particle diameters at 200 and 300 mJ/cm2 show that spheronization occurred in the region below J1. In this region, spheronization does not usually occur because the laser fluence is lower than the energy necessary for melting. Furthermore, it is difficult to believe that melting occurred in areas of high local fluence. At 300 mJ/cm2, particles with diameters of 500 nm or greater were present; however, J1 for the particles is around 400 mJ/cm2. This contradicts previous experimental results that nanoparticle generation dominates at an irradiation of 350 mJ/cm2 or more. Therefore, it seems that substances other than TiN and TiO2 generated by oxidation were participating in sphere formation, and these other substances need to be included in the H–M–E model.

Based on our overall results, we propose the following mechanism for spheronization. Initially, TiN nanoparticles existed as aggregates of about 50–300 nm. During laser irradiation, the aggregate melted and droplets formed with diameters of about 50–250 nm. Subsequently, spherical particles formed by rapid cooling between laser pulses. During heating and melting, spherical surfaces were oxidized. Furthermore, spheres aggregated with other TiN nanoparticles and spherical particles, and the aggregates were irradiated by the next laser pulse. TiO2 did not absorb light at 532 nm, but TiN in the spheres was heated by the laser beam. TiO2 received some of this heat by conduction, so the entire sphere is heated. The melting point of TiO2 is 2130 K, which is lower than the melting point of TiN (3218 K) [43]. Therefore, TiO2 melts when the laser fluence is lower than J1 for TiN, and the spheres grow by incorporating aggregated particles. Hence, a spherical particle forms with a diameter that is not obtained only to the TiN presence. The cause of nanoparticle generation would be ablation or evaporation by laser irradiation at high fluence. Spheres would evaporate because the boiling point of TiO2 is 2500–3000 °C [44], which is nearly the melting point of TiN. Nevertheless, a detailed investigation is required on the mechanisms involved in the melting and creation of particles.

As regards the oxidation, it is considered that the oxidation of TiN occurred by dissolved O2, by water as solvent, or by oxygen atoms included in raw TiN. In addition, the oxidation of the Ti, which is generated by decomposition of TiN, may be occurred. Considering the amount of generated TiO2, the main cause of oxidation of the particles would be dissolved oxygen and water. There are some problems that will be caused by the formation of TiO2 when the product will be used in applications. For example, in the application to the electrodes, the particle’s electrical conductivity is decreased by oxidation, so spheres cannot use as electrodes material. In terms of mechanical properties, the hardness of the spherical particles is degraded by oxidation. Accordingly, we are currently performing an experiment using liquid nitrogen as the solvent for the production of pure TiN spheres and elucidation of the mechanism for sphere production.

3.2 Influence of irradiation time on particle size and composition

SEM images and particle size distributions for each irradiation time are shown in Fig. 9. Irradiation times of 15, 30, and 60 min were used with laser fluence and raw TiN particle concentration set to 80 mJ/cm2 and 0.25 mg/ml, respectively. Particles irradiated for 60 min (Fig. 9c) had slightly larger diameters than the others. We believe that differences in particle size were caused by increases in particle growth through TiO2 melting. An oxide’s size increases with the irradiation time, and the particles become easy to melt due to the TiO2 presence. This assumption is supported by the UV–Vis spectrum (Fig. 10). In Fig. 10c, absorbance in the visible region decreased, suggesting that the amount of TiN decreased by oxidation. However, the influence of irradiation time on particle size was not as large as the effects of laser fluence. In fact, from the studies of the mechanism, there are reports that increases in particle size require increases in both aggregate size [36, 37] and fluence; the latter must be large enough to fuse the aggregates [38].

Fig. 9
figure 9

SEM images and size distributions of spherical particles after irradiation for a 15 min, b 30 min, and c 60 min. Corresponding average sphere sizes were a 96 ± 19 nm, b 114 ± 29 nm, and c 117 ± 27 nm

Full size image
Fig. 10
figure 10

UV–Vis absorption spectra of samples obtained by laser irradiation over different irradiation times

Full size image

3.3 Influence of raw particle concentration on particle size and composition

Figure 11 shows SEM images and particle size distributions of the particles for each concentration of raw TiN particles. The TiN particle concentrations were 0.25, 0.50, and 1.0 mg/ml.

Fig. 11
figure 11

SEM images and size distributions of spherical particles obtained by laser irradiation using different raw TiN nanoparticle concentrations. Concentrations of TiN were a 0.25 mg/ml, b 0.50 mg/ml, and c 1.0 mg/ml. Corresponding average sphere sizes were a 96 ± 19 nm, b 114 ± 29 nm, and c 117 ± 27 nm

Full size image

At each concentration, laser fluence and irradiation time were set to 80 mJ/cm2 and 30 min, respectively. At the highest concentration, 1.0 mg/ml, raw TiN particles remained because many particles were not exposed to the incident laser light. Spheronization proceeds only in regions that are exposed to the laser light. The UV–Vis spectrum in Fig. 12 shows no major differences in absorbance between the ultraviolet and visible regions; therefore, the raw TiN concentration hardly affected the composition of spherical particles.

Fig. 12
figure 12

UV–Vis absorption spectra of samples obtained by laser irradiation with different TiN particle concentrations

Full size image

4 Conclusions

Using TiN nanoparticles as the raw material, we succeeded in producing submicron spherical particles with diameters of about 50–600 nm. The spheronization method was laser melting in liquid. The produced spherical particles contained TiO2, TiOxNy, and TiN, indicating that oxidization occurred during the melting process. Laser fluence had the largest effect on the size and composition of the product spheres. Spheronization began at a laser fluence of approximately 80 mJ/cm2. Particle diameters increased in the fluence range of 77–350 mJ/cm2, but the generation of nanoparticles became dominant at or above 350 mJ/cm2. The generation of nanoparticles by laser irradiation at high fluence was probably caused by the evaporation or ablation of particles. Therefore, spheres with diameters of 50–600 nm could be generated by controlling the fluence in the 80–350 mJ/cm2 range. Oxidization appears to have increased with the laser fluence. Generation of TiO2 implies that spheronization is related to oxidation. It will be necessary to investigate the mechanism of oxidization and spheronization in future studies.