1 Introduction

Silicon vacancy centre (SiV) in diamond is one of the many point defects that has sparked interest among researchers because of its potential to produce indistinguishable single photons [1,2]. They are preferred over their competitor nitrogen vacancy (NV) because of their superior spectral properties, such as narrow zero phonon line (ZPL) (< 6 nm at room temperature), bright emission and a weak phonon side band [1]. Although similar to NV in applications and properties, SiVs are structurally different from NV. SiVs occur in a split-vacancy configuration, with a silicon atom sitting between two adjacent vacancies of carbon atoms [3]. This gives SiV an inversion (D3d) symmetry [3]. SiV is a spin-1/2 system with its ground and excited states within the diamond band gap [4]. This results in a narrow zero phonon line for SiV at 738 nm [3].

Different methods, such as ion implantation [5], in situ doping [6], laser ablation [7], electron irradiation [8], etc. have been attempted to create colour centres in diamond. For the generation of SiVs, the preferred method is in situ doping [9] since implantation creates damages to the diamond lattice [10]. In situ doping of silicon in diamond, which leads to the formation of SiVs can be achieved in two different ways. Silane gas is passed along with hydrogen and methane to facilitate silicon doping during chemical vapour deposition (CVD) diamond growth [11]. The alternate method, which utilizes a solid-state source placed in the plasma [9,12] is preferred because of better control on doping concentration and better safety [13]. However, SiVs are routinely found in CVD-grown diamonds without any intentional dopants added [14,15]. At the initial stages, this was attributed to the silicon substrate on which the diamond films were grown [15].

Rzepka et al [16] proposed that the diffusion of silicon atoms into diamond through the substrate–film interface was the reason for the formation of SiVs. However, later calculations revealed that the diffusion coefficient for silicon in diamond is too low even at elevated temperatures for this to happen [17]. SiVs have also been attributed to the residual silicon resulting from the plasma exposure to the silicon-containing components in the chamber [15].

Through this work, we attempt to explain the different factors that lead to the formation of SiVs in diamond. Effect of the substrate material and plasma is studied and photoluminescence (PL) mapping has been used to divulge the factors that contribute to the formation of silicon vacancies.

2 Experimental

Diamond thin films were grown on different substrates using a microwave plasma chemical vapour deposition (MPCVD) system supplied by IMAT-CVD India. Microcrystalline diamond (MCD) films were grown on silicon (100), platinum-coated silicon (100), sapphire (0001), polycrystalline alumina and molybdenum substrates. First, the substrates were kept in hydrogen plasma for 10 min to confirm that they stayed stable within the plasma. The substrates were then cleaned and seeded by micro-abrasive seeding using diamond powder. For the growth of micro-crystalline diamond, hydrogen and methane gases were used at a flow rate of 480 sccm and 20 sccm each. One nanocrystalline diamond (NCD) film was grown on silicon (100) substrate by flowing 10 sccm of nitrogen along with 470 sccm of hydrogen and 20 sccm of methane. The method for growth of films was fixed based on another study, and the role of nitrogen in the formation of nanocrystalline films will be discussed elsewhere. Depositions were done for the duration of 2 h each by keeping the substrate temperature at 950°C and microwave power of 3.5 kW. For Pt–Si, the substrate was composed of a polycrystalline platinum layer of 150 nm, on a TiO2 layer of 20 nm, on a thermally grown SiO2 layer of 300 nm (Pt/TiO2/SiO2/Si). For this substrate, the temperature was kept < 750°C and microwave power of 2.5 kW so that the platinum interlayer would not melt. After deposition, the films were cooled by reducing the microwave power in phases. However, since there is no additional heater attached to the system and the heating is provided by the microwave itself, the cooling was natural and it took around 20 min for the substrates to cool down to room temperature. However, for molybdenum and alumina substrates, the film started peeling off when the cool-down process started and only small patches of film remained on the substrate. For other substrates, such an issue did not occur. Morphology of the films was studied with the help of a field emission scanning electron microscope (FESEM, FEI Inspect F50) operated at 20 kV. Structural properties were studied by X-ray diffraction (XRD) using a Rigaku-Smartlab spectrometer operated at 9 kW (CuKα). Fluorescence mapping and spot measurements were performed at room temperature and at low temperature, using a Witec alpha 300 R spectrometer with laser excitation of 532 nm and a 50× microscope objective. The fluorescence lifetime measurements were done using a streak camera detector with a laser excitation wavelength of 532 nm and a 100× microscope objective.

3 Results and discussion

SEM images of the films grown on different substrates are shown in figure 1a–f. Figure 1a and b corresponds to the SEM images of diamond films grown on silicon (100) substrate (referred as D1 and D2 from now onwards). The micrograph of D1 shows continuous growth of the film with large grain sizes of the order of 1 µm, which can be categorized as microcrystalline diamond. Diamond film D2, which was grown by adding nitrogen during growth shows smaller grain size, which is of the order of nanometres. Figure 1c, which shows the film grown on a platinum-coated silicon substrate (D3) reveals a film that is not continuous. On a sapphire substrate (D4), instead of continuous film growth, isolated spheres of diamond particles with diameters between 5 and 10 μm are resulted, as shown in figure 1d. The presence of different facets and grains on each of these particles means that they are not single crystals, but polycrystalline spheres of diamond. The films grown on alumina and molybdenum substrates (D5 and D6) peeled off during the cooling down after deposition because of the mismatch between the thermal expansion coefficient of the substrate and film [18,19]. For the alumina substrate, the film chipped away as large pieces and only small regions of diamond remained on the substrate, which is by themselves continuous as seen in figure 1e. Meanwhile, the films grown on molybdenum substrate cracked on the substrate with most of the film staying on the substrate. SEM image of D6 (figure 1f) shows one such crack in the film. In short, diamond films were grown on five different substrates so that the substrate effect on the formation of SiVs can be studied.

Figure 1
figure 1

SEM micrographs of diamond films grown on different substrates. (a) MCD film and (b) NCD film on silicon substrate, MCD on (c) platinum-coated silicon, (d) sapphire, (e) alumina and (f) molybdenum.

Full size image

X-ray diffraction (XRD) pattern and PL spectra of the films D1–D6 are shown in figure 2a and b, respectively. For all the films, XRD pattern revealed the diamond (111) peak, which occurs at 43.5°. However, the other peaks of diamond at 75.9, 93.7 and 119.5° for the (220), (311) and (400) planes were not visible in some of the samples. This is because of the preferred (111) orientation, which is commonly seen in CVD-grown polycrystalline diamond samples [20]. Since the XRD measurements were done at 2.5° off-axis from the substrate normal, the substrate peaks could be avoided for samples D1–D4 [21]. For discontinuous films D5 and D6 on polycrystalline substrate, the substrate peak could not be avoided and is visible in the XRD pattern. The PL spectra of the samples show a sharp peak at 738 nm, which is the zero phonon line of SiV, revealing the presence of SiV in the samples [3]. However, sample D2, which is a nanocrystalline film on silicon did not show such a peak. SiV is present in all the films irrespective of whether their substrates contain silicon or not. The PL measurements reveal that the source of silicon that is responsible for the formation of SiVs in MPCVD is the residual silicon present in the chamber, which may originate from the quartz components that are used to control the shape of the plasma. The absence of SiV in the nanocrystalline sample may be due to the presence of nitrogen in the plasma, which has been reported to reduce the etch rates of hydrogen plasma [22].

Figure 2
figure 2

(a) X-ray diffraction pattern of the films D1–D6. The diffraction angles corresponding to the position of diamond are marked with dotted grey lines. (b) Photoluminescence (PL) spectra of the films recorded with a laser excitation of 532 nm.

Full size image

To understand the effect of substrate material, cross-sectional PL and Raman mappings were done on films D1 and D2 after making a freshly cut cross-section. The cross-sectional map of sample D1 corresponding to the intensity of the first-order Raman peak in diamond (1332 cm−1) is shown in figure 3a [23].

Figure 3
figure 3

(a) Raman mapping of the cross-section of film D1 for the 1332 cm−1 peak (substrate to film is left to right). PL maps of the films (b) D1 and (c) D2 for the 737 nm peak of SiV. The excitation wavelength was 532 nm for all the measurements. Scale bar of the intgrated intensity is shown below each figure.

Full size image

This map shows an increase in intensity along the growth direction of diamond (Film is kept vertically with the substrate on the left and thus, the film growth direction is left to right.), which reveals an improvement in diamond quality as it grows [23]. Figure 3b and c shows the PL map of the films D1 and D2, corresponding to the 737 nm ZPL of SiV. The cross-sectional PL mappings reveal the effect of the two contributors to the formation of SiV in diamond. The first source, the substrate itself, causes the formation of SiV near the silicon–diamond interface, which results in a bright line at the interface. The formation of this layer can be attributed to the hydrogen plasma etching silicon from the substrate during the initial stage of diamond growth until a continuous layer of film grows on the substrate. In nanocrystalline film D2, this is the only contributor to the formation of SiV, which is why there is no PL signal from the defect, which is closer to the surface of the film. This confirms the earlier observation in figure 2b, where the PL peak corresponding to the SiV was not observed in the film D2. Even though there is SiV centres present in the film D2, they are at the interface, and not at the surface, and as a result, not observed in the plan-view spectra from the surface. For the MCD film, there is a second contributor, which causes a gradual increase in intensity along the diamond growth direction. This is the residual silicon from the quartz components exposed to the plasma, results in the formation of SiV throughout the film [15]. The gradual increase in intensity can be attributed to the increase in diamond crystalline quality, which is revealed from the Raman map. On the other hand, the absence of SiV in nanocrystalline diamond can be a result of nitrogen addition in the plasma. Since the samples D3–D6 do not contain a continuous film, cross-sectional measurements are not possible. However, since samples D4–D6 do not contain any Si and the Si in D3 is not exposed to the plasma because of the Pt interlayer, the sole contributor for SiVs in these samples is the residual silicon in the chamber.

Figure 4a shows the PL area map of a 50 × 50 μm sample D1, which reveals peaks corresponding to the position of SiV on the sample. The intensity variation of the signal emanates from the sample can result from a combination of grain size [24], grain orientation [25] and the position of SiV in the grain [26]. The fluorescence lifetime data shown in figure 4b is taken with a pulsed excitation of 532 nm and was fit using a double exponential. The shorter lifetime corresponding to the background fluorescence was 0.52 ns, whereas the longer lifetime, which corresponds to the SiV was 4.7 ns [27]. Temperature dependence of the fluorescence was measured from room temperature (320 K) to 80 K using a liquid nitrogen-based cryostat at intervals of 30 K. The normalized data revealed a blue shift in the PL spectra with the peak becoming sharper. At 80 K, as shown in figure 4c, the zero-phonon line occurs at 737.2 nm with a phonon sideband at 766.7 nm. The temperature dependence of the PL peak position showed a linear relationship and is shown in figure 4d. The slope of the linear fit reveals a temperature dependence of 0.0359 nm K−1 for the peak position.

Figure 4
figure 4

(a) PL map of the film D1 for the 737 nm peak of SiV taken in plan-view. (b) Transient fluorescence spectum of the film D1 excited at 532 nm. (c) PL spectum of SiV taken at 80 K. (d) Temperature dependence of the peak position of SiV.

Full size image

4 Conclusions

In this work, we have grown diamond thin films on different substrates to understand the formation of silicon vacancies in diamond during growth. The films grown on silicon substrate were uniform and had good adhesion with the substrate after cooling down. For the films grown on Pt-coated silicon and sapphire substrates, continuous films were not obtained. While the films on platinum-coated silicon contained patches that were connected with one another, spherical micron-sized particles were observed on the sapphire substrate. The film growth was uniform on alumina and molybdenum substrates, but the films peeled off during the cooling process because of the mismatch in the thermal expansion between the substrate and film.

All the films grown in this study, except the nanocrystalline diamond on silicon, showed the presence of SiV in the PL spectra. However, cross-sectional PL mappings revealed the presence of SiV at the silicon–diamond interface for the nanocrystalline film. The Raman spectra, combined with the cross-sectional PL mappings revealed the mechanism behind the formation of SiVs in the diamond. While silicon originating from the substrate is a contributing factor for the formation of SiV at the silicon–diamond interface, the presence of SiVs throughout the thickness of the film is because of residual silicon in the chamber. The residual silicon originates from the quartz components in the chamber that are exposed to the plasma and can be controlled by varying the growth conditions. The lateral uniformity, fluorescence lifetime and temperature dependence were studied to look for possible applications.