1 Introduction

Diamond has attracted much attention for its excellent physical, optical, thermal and electrical properties, such as a high mechanical hardness of 100 GPa, a wide spectral transparency of about 70% from UV to far IR, a high thermal conductivity of 20 W/(cm·K), a high carrier mobility of 4500 cm2/V⋅s for electrons and 3800 cm2/V⋅s for holes and a large energy bandgap of 5.47 eV [1,2,3,4,5,6]. It can be applied to lathe tools, the jewellery industry, heat–sink devices, photon/particle detection and the semiconductor industry [7,8,9,10,11]. For the scarcity of natural diamonds, high pressure high temperature growth (HPHT) and microwave plasma chemical vapor deposition (MPCVD) methods have been developed as two promising ways of producing artificial diamonds [12, 13]. Gem-quality diamonds with a big size grown by the MPCVD method have been commercialized successfully [14]. The factors influencing diamond crystal growing quality, including temperature, gas flow rate, pressure, microwave power and auxiliary gas, have been discussed in detail [15,16,17,18,19]. Although diamond seeds are generally polished before growth, the surface roughness of diamond seeds is rarely mentioned affecting the crystal growth quality.

Recently, researches on growth on the surface of diamond microstructure had been proposed to improve diamond crystal growth quality. Hideo Aida et al. studied the diamond growth mechanism on diamond microneedles and achieved a stress-free layer of the diamond above the diamond microneedles [20, 21]. Tao et al. prepared nanorods on the surface of diamond seeds to grow high-quality CVD diamond layers. They noticed that the diamond seed with optimized nanorods pattern exhibited a smooth surface, lower stress and defect density [22]. Growing on the patterned microstructures surface of diamond seed could improve the diamond quality. However, does a rough surface with random patterns have a similar effect?

In the paper, three diamond seeds were prepared by laser cutting a bulk diamond for the growth experiment. Before growth, they were partially polished, totally polished and unpolished, respectively. According to the microscope and step profiler, diamond surface morphology and surface roughness were recorded. The Raman spectra, ultraviolet − visible (UV − Vis) and infra-red (IR) spectra were collected to judge long-term growth diamonds’ crystalline quality and optical properties.

2 Experimental procedures

All experiments were carried out in a homemade commercial 2.45 GHz MPCVD reactor by collaborating with Guangdong TrueOne Semiconductor Technology Co., Ltd. A bulk (100) − oriented single-crystal diamond was grown in MPCVD by a diamond seed (EDP Corporation, Japan), whose size was 7.5 mm × 7.8 mm × 3.4 mm. Three samples with a thickness of 0.5 mm were achieved by cutting the bulk diamond on a LASER SUPRA532 diamond cutting machine (STPL, India). Sample A was partially polished by a polishing machine, and half of the area was polished. Sample B was totally polished. Sample C did not suffer any polishing treatment. The three samples were cleaned with boiling concentrated sulfuric acid, followed by acetone, ethyl alcohol, and deionized water in the ultrasonic bath. Drying the samples on a heating platform was achieved at 80° centigrade for 5 min before growth. Once the MPCVD chamber gate was closed, the vacuum chamber was continuously pumped out through a molecular pump for 2 h to reach a base pressure of ~ 10−6 mbar. Before growth, the pressure in the chamber was set 160 mbar and the H2 gas flow rate was 550 sccm, the samples were etched by H plasma for 2 h to remove the surface damage. After etching, the pressure in the chamber was set 200 mbar, and 37 sccm for CH4. The whole process of this experimental procedure is shown in Fig. 1.

Fig. 1
figure 1

Whole process of the experimental procedure

Full size image

To explore the initial process of growth, the sample A grew for only 1 h. The sample B and C grew in the same chamber for 150 h with a high rate of about 7 μm/h. The growth conditions of all samples are summarized in Table 1. During the experiment, the crystal quality of the three samples was assessed by recording their Raman spectra using a HORIBA LabRAM HR Evolution spectrometer (HORIBA, France) excited by a solid-state laser at 532 nm with a spot size of 2 µm. The surface profile data were collected using a Dektak XT probe step profiler (Bruker, Germany) with a 30 uN force applied on the microneedle and a scanning distance of 5 mm. The surface morphologies of the samples were examined using a VHX − 600 digital optical microscope (Keyence, Japan) at a magnification of 500x. The UV − Vis transmission spectra of all samples were recorded using a SolidSpec-3700 spectrophotometer (Shimadzu, Japan) over a scan range of 200 − 800 nm. The IR transmission spectra of all samples were recorded using a Nicolet iS50R Fourier transform infrared spectrometer (Thermo Scientific, America), with a scanning range of 2000 cm−1 to 6000 cm−1.

Table 1 Growth conditions of all samples
Full size table

3 Results and discussion

3.1 Surface morphology and surface roughness

The optical image of the surface depicted in Fig. 2a, b illustrates Sample A prior to its growth. Figure 2a displays the cut area, while Fig. 2b displays the polished area. These figures provide a visual representation of the initial state of Sample A. Figure 2a reveals the presence of several strips resulting from laser cutting. In contrast, Fig. 2b appears dark, as the polished surface reflects light with ease. These observations suggest that the polishing process has effectively altered the surface properties of the material. Upon completion of the growth process, a surface image of the interface between the growth area and the cut area was obtained after 1 h, as depicted in Fig. 3c. Notably, the figure on the right side of Fig. 3c illustrates the characteristic step-flow growth mode of CVD diamond, which is observed in the experiment [23]. Nevertheless, the left area remained as a cut surface, and it appeared that the partially polished diamond sample had a preference for growth from the polished side. This observation suggests that the polishing process has a significant impact on the growth characteristics of the diamond material. The profiles of Sample A before and after 1 h of growth are presented in Fig. 3a. The figure clearly depicts significant peak-to-valley profiles resulting from laser cutting between 0 and 2 mm. Moreover, the tilt of the polished area observed in the figure is consistent with the actual polishing condition. To facilitate a comparison of the thickness before and after growth, the average thickness calculated from 0 to 3.5 mm is established as the uniform baseline. Notably, the difference in the average thickness calculated between 3.5 and 5 mm is found to be 7.39 μm. It is apparent that the polished region of the diamond sample exhibits a preference for growth. This phenomenon can be attributed to the epitaxial lateral overgrowth (ELO) mechanism of diamond [24]. The rough surface of a diamond tends to undergo epitaxial lateral overgrowth initially, followed by the stacking of diamond layers, after the uneven diamond surface is connected to form a smooth plane. Hence, after 1 h of growth, the cutting areas on the surface of Sample A do not exhibit the characteristic step-flow growth mode, while the polished areas of the surface show the characteristic step-flow growth mode.

Fig. 2
figure 2

a Surface morphology of the cut area of sample A before growth; b surface morphology of the polished area of sample A before growth; c surface morphology of an interface between the growth area and cut area of sample A after growth 1 h; d surface morphology of sample B after growth; e surface morphology of sample C after growth

Full size image
Fig. 3
figure 3

a Surface profile of sample A before and after growth; b surface profile of sample B before and after growth; c surface profile of sample C before and after growth

Full size image

After 150 h of growth, the surface morphologies of Samples B and C are examined and are presented in Fig. 2d and e, respectively. After a thorough analysis of the figures, it is apparent that they exhibit the characteristic step-flow growth pattern of CVD diamond. The step-flow pattern observed in Fig. 2d appears to be denser and more uniform in comparison with that shown in Fig. 2e. Figure 3b, c exhibits the surface profiles of sample B and C, respectively, both before and after the growth process. Ra and Rz are used to characterize the surface roughness of sample B and sample C, where Ra represents the arithmetic mean deviation of the surface profile, Rq represents the root mean square of the surface profile, while Rz represents the maximum height difference of the surface profile. The mathematical models used to calculate all evaluated parameters are shown from Eqs. (1), (2) and (3) [25]:

$$\mathrm{Ra}=\frac{1}{l}{\int}_{0}^{l}\left|Z(x)\right|dx$$
(1)
$$\mathrm{Rq}=\frac{1}{l}\sqrt{{\int}_{0}^{l}{Z}^{2}(x)dx}$$
(2)
$$\mathrm{Rz}={Zp}_{s}+{Zv}_{s}$$
(3)

where Z(x) represents the values of the ordinates of the profile points in the sampling length l. Zps represents the maximum height of the profile peaks, while Zvs represents the greatest depth of the profile valleys. These values are both determined within the sampling length l. Table 2 presents the Ra, Rq, and Rz values acquired for sample A and B, utilizing l = 5 mm (with profiles collected via probe step profiler) and dx = 0.001 mm parameters. The Ra, Rq, and Rz values for the polished surface of sample B before growth were 0.007 μm, 0.004 μm, and 0.05 μm, respectively. In contrast, the unpolished surface of sample C exhibited Ra, Rq, and Rz values of 1.76 μm, 0.99 μm, and 10.09 μm, respectively. It can be observed that the surface of diamond that has been polished is smoother compared to the surface which has not been polished. Following a growth period of 150 h, the Ra, Rq and Rz values for sample B were found to be 0.72 μm, 0.38 μm and 3.83 μm, respectively, and for sample C, the values were 0.63 μm, 0.33 μm and 3.07 μm, respectively. By comparing the Ra, Rq, and Rz values of the diamond surfaces before and after growth, it can be seen that sample B, which had been polished, became rougher after 150 h of growth, while sample C, which had not been polished, became smoother after 150 h of growth. Despite the Ra, Rq and Rz values for the grown sample C being lower than those of the grown sample B, the difference is insignificant. It appears that after a certain duration of growth, the surfaces of diamond samples, which had differing initial roughness, reached a similar level of roughness. It can be assumed that, when the surface of the diamond sample is relatively flat, the diamond undergoes growth in a layer-by-layer step-flow pattern. When the surface of a diamond is rougher, its initial growth proceeds in accordance with the ELO mechanism until arriving at a sufficiently smooth, flat plane. At this point, the diamond transitions to using the stacking step-flow growth mode. This tendency represents a common pattern in diamond growth, observable particularly when samples differ in initial surface roughness. Therefore, the surface roughness of the diamond seeds do not impact the surface roughness of the diamonds produced during long-term growth, which is generally linked to the growth conditions under which the diamonds are produced [26].

Table 2 Ra, Rq, and Rz values acquired for sample A and B, utilizing l = 5 mm (with profiles collected via probe step profiler) and dx = 0.001 mm parameters
Full size table

3.2 Crystalline quality and optical property

The crystalline quality of all diamond samples was compared by Raman spectra, and the optical property of all samples was judged by UV − Vis and IR transmission spectra. The sampling position of sample A was the polished area and growth area. The Raman spectra of all diamond samples are shown in Fig. 4a. The full width at half maximum (FWHM) and Raman peak are summarized in Table 3. The internal stress of CVD diamond layers can be estimated by the following Raman spectra analysis:

Fig. 4
figure 4

a Raman spectra of all samples before and after growth; b UV − Vis transmission spectra of all samples before and after growth; c IR transmission spectra of all samples before and after growth

Full size image
Table 3 FWHM and Raman peak of all samples
Full size table
$$\begin{aligned}\mathrm{P}=0.34\frac{Gpa}{{cm}^{-1}}\Delta v\\{\rm with} \, \Delta v=v-1332.5 {cm}^{-1}\end{aligned}$$

where Δν is the shift of the Raman peak of the diamond layers with respect to the unstressed Raman peak of the diamond [27, 28]. Table 3 shows that the three diamond seeds prior to growth exhibited a very similar tensile stress of 0.01 GPa, as indicated by the Raman peak measured at 1332.47 cm−1. Their respective FWHM values were 1.821 cm−1, 1.843 cm−1, and 1.858 cm−1. These preliminary findings provide a useful baseline for assessing any changes in stress and crystalline structure that may result from subsequent growth processes. It is worth noting that the FWHM values of the three samples were all very close to each other, with minimal variation observed. This consistency suggests a high degree of structural uniformity among the samples. For the three samples after growth, the respective Raman spectra peak were at 1332.74 cm−1, 1332.02 cm−1 and 1332.74 cm−1, indicating a compression stress of 0.08 GPa, a tensile stress of 0.16 GPa and a compression stress of 0.08 GPa, respectively. The FWHM of the three samples after growth were 1.916 cm−1, 1.950 cm−1 and 2.001 cm−1, respectively. In comparison, the internal stress of sample C was lower than that of sample B; however, the FWHM of sample C was wider than that of sample B. In general, CVD diamonds have a certain tensile stress [29], but the unpolished sample C after growth suffered compression stress. It indicated that a certain rough surface of diamond seed could improve the internal stress of the diamond, but reduce the crystalline quality slightly. The view of improving internal stress was also recognized by other researchers [21, 22].

The UV − Vis and IR transmission spectra are shown in Fig. 4b, c. From Fig. 4b, a strong absorption occurred between 200 to 230 nm, and the transmission of all samples between 400 and 800 nm reached approximately 71.2% (theoretical value) [30], indicating that all visible light could transmit three samples. From Fig. 4c, the strong bands at 1900 cm−1 and 2300 cm−1 were the inherent two − phonon lines of a diamond with C–C bonds [31]. In addition, the type absorption lines such as 1282 cm−1, 1370 cm−1 and 1330 cm−1 were invisible, suggesting that the nitrogen content in all samples was relatively low. Similarly, the transmission of all samples between 4000 cm−1 and 6000 cm−1 reached approximately 71.2%. Therefore, the optical properties of all samples before and after growth mainly were the same. The surface roughness had almost no impact on the optical properties of the diamond after growth.

4 Conclusion

In summary, three samples were prepared by cutting a bulk diamond and subsequently suffered partially polished, totally polished and unpolished, respectively. After a period of growth, the growth of the partially polished sample A started from the low areas with a high rate of about 7 μm/h. The surface roughness of long-term growth totally polished sample and unpolished sample were almost the same. As a result, the surface roughness of diamond seed made no effect on the surface roughness of long-term growth diamond. Compared by the Raman spectra, a certain rough surface of diamond seed could improve internal stress of diamond, but reduce the crystalline quality slightly. The UV − Vis and IR transmission spectra were used to judge the optical properties. The results showed that the surface roughness had almost no impact on optical properties of diamond after growth. Therefore, for gem-quality diamond production, diamond seeds with low roughness after cutting could be used for direct growth without polishing.