The n-type doping of diamond is quite difficult, hindering the development of diamond-based electronic devices for decades. In this work, we have designed a boron–nitrogen co-doping technique to realize n-type diamonds. Basically, the activation energy of the donors has been greatly reduced by around 50%, thanks to the successful synthesis of the boron–nitrogen related donor-like complex by a fine control of the synthesis condition. Compared to the sole nitrogen doping scheme, it is found that the co-incorporation of boron elements is beneficial to a lot of aspects, including better crystalline quality, faster growth, higher nitrogen solubility, and stability. With the technique, a p-i-n diamond homojunction has been fabricated. A clear rectification behavior has been recorded, demonstrating that the current co-doping technique we proposed is a feasible path to the accessible n-type diamond.

Compared to other wide bandgap semiconductors, diamond presents unbeatable physical properties, which are particularly favorable for developing high-power electronics. Throughout the last few decades, boron-doped diamonds have been extensively studied. It has been found that the boron dopants can be easily incorporated.1–3 The boron acceptor level is located at 0.37 eV above the diamond valence band maximum (VBM) by analyzing the temperature dependence of the resistivity.4,5 According to the Hall effect measurements, the maximum mobility of boron-doped diamonds synthesized by chemical vapor deposition (CVD) has reached 1840 cm2/V s at 290 K.6 

In contrast, the n-type diamond is difficult to realize, leading to a major issue for the fabrication of diamond-based bipolar devices. Various attempts have been made to grow n-type diamond films, among which nitrogen and phosphorous are two popular selections. Regarding nitrogen, it fails to act as a shallow donor.7 For nitrogen implanted diamonds, the activation energy has been determined as 1.4–1.8 eV for a single nitrogen substitution.8 The activation energy is close to the calculated ionization energy of an isolated nitrogen in diamond, around 1.7 eV below the conduction band minimum (CBM).9 On the contrary, the phosphorus has a relatively shallow donor level of ∼0.6 eV.10,11 However, the electron density is quite low because of the low phosphorous content in the CVD diamond limited by chemical kinetics.12 On account of impracticable individual n-type dopants, some researchers have turned toward employing combinations of lighter elements arranged in clusters to make donor characteristics.13–15 This so-called co-doping technique could provide the required characteristics for useful n-type diamonds. Theoretically, the clusters, such as BN2,16 Si4N,17 and LiN4,18 that add one electron to the diamond lattice could provide donor levels even shallower than phosphorus.

More importantly, the donor density could be significantly enhanced by co-doping two elements that complement the lattice strain. In this sense, boron and nitrogen have made a perfectv co-doping combination.19,20 It is easy to understand that the strain induced by larger N ions could be relaxed by the addition of smaller B ions, leading to an enhanced solubility of both N and B atoms. Furthermore, Croot and Moussa have predicted from their calculations that the BN2 complex could be a donor with an ionization energy of 0.8–1.2 eV.18,21 Although the value is slightly deeper than phosphorous, the enhanced doping concentration could make the electron density comparable. At the same time, the crystalline quality is expected to be better than the phosphorous-doped diamond. As a result, the boron–nitrogen co-doping could be a feasible path to practical n-type diamonds. However, such a co-doping behavior is rarely investigated from experiments, especially on single-crystalline diamond (SCD) films.

In order to realize the desired complex donor, the growth experiment should be carefully designed because the forms of the B–N complex could be quite “complex.” Consequently, in this paper, we have fixed the boron dosage while changing the nitrogen one to artificially make a series of N/B ratios in the gas phase during the diamond film growth. The experiment is designed to investigate the possibility of using boron–nitrogen co-doping to realize n-type conductivity. The co-doped homoepitaxial layers are grown on a (100)-oriented type Ib CVD single-crystalline diamond substrate (3 × 3 × 1 mm3) by microwave plasma chemical vapor deposition (MPCVD). There is a same 3° off angle for all substrates. Before the substrates are introduced into the growth reactor, they are cleaned chemically in order to remove existing metallic and organic contaminations from the surface. The flow rates of CH4 and H2 both having 5N purity are kept at 20 and 500 standard cubic centimeter per minute (SCCM), respectively. The B/N doping is carried out using hydrogen-diluted diborane and nitrogen. The ratio of B2H6 to CH4 (B2/C) in the gas phase is fixed to 50 ppm for all the samples.22 The nitrogen concentration, defined as the N2/CH4 (N2/C) gas flow ratio, is kept at 0%, 2.5%, 5.0%, 7.5%, and 10%. The chamber pressure and the microwave power for all deposition are fixed at 160 Torr and 3.8 kW. The substrate temperature is set to 900 °C, as determined by an IR pyrometer.23,24 For the samples grown by the N2/C ratio of 0%, 2.5%, 5%, 7.5%, and 10%, the appropriate thickness is 6, 70, 74, 77, and 82 μm, respectively. After growth, the as-grown samples are cleaned with H2SO4 and HNO3 (1:1) mixed solutions at 250 °C for 0.5 h to remove the produced graphite on the surface. After chemical treatment, the samples are rinsed with acetone, alcohol, and deionized water.

The surface morphology is characterized using Atomic Force Microscopy (AFM) (NTEGRA, NT-MDT), and the AFM images are recorded on a 20 × 20 μm2 surface. The thickness of all diamond films is measured employing a micrometer with a resolution of 1 µm. Micro-photoluminescence (PL) and micro-Raman spectra are carried out with a 5 mW green laser (514 nm) as an excitation source at room temperature, and the laser is focused on the grown layers by a 100× objective confocal lens (Olympus) with a spot diameter of ∼1 and penetration depth of ∼400 µm, and the backscattering Raman data are recorded by employing a CCD camera. For electrical measurement, Ti (30 nm)/Au (100 nm) pads for Hall or in coplanar geometry with a gap of 26 μm for photoconductivity measurements are evaporated and annealed at 500 °C for 1 h to ensure stable Ohmic contacts. The Hall measurement of all doped diamond layers is carried out in vacuum by using an HL5500PC Hall effect measurement system with a 0.6 T magnetic field at variable temperatures from 300 to 873 K. Since the identification of the conduction type could be influenced critically by the conductive inhomogeneity of the prepared samples,25–27 two configurations, namely, van der Pauw (vdP) and Hall bar, have been employed for accurate identification of the conduction type. Details of the design for each configuration could be found in Fig. S1. For the fabrication of the bar-structured sample, the nitrogen-doped epilayer is etched into a Hall bar structure using an H2 plasma in combination with a patterned Ti layer acting as a hard mask. Details of the fabrication process of the bar-structured sample of the supplementary material. The depth profiles of B and N atoms have been measured by secondary ion mass spectrometry (SIMS) on a Cameca IMS 4f instrument using Cs+ primary ions accelerated at 10 keV. The atomic bonding structure of the diamond films with different N2/C ratios is analyzed using an x-ray photoelectron spectroscopy (XPS) system equipped with an Al source.

Figure 1(a) shows the growth rate for the boron–nitrogen co-doped diamond film samples, which have been double-confirmed by cross section PL profiling, as shown in the section titled Confirmation of the growth rate and Fig. S2 of the supplementary material. Nitrogen incorporation is found to accelerate the growth rate dramatically at the initial stage, which is similar to the case of the nitrogen mono-doped diamond growth.28,29 However, for the samples with a N2/C ratio higher than 2.5%, the growth rate enhancement by nitrogen has been much less significant. The break point indicates that the boron–nitrogen co-doping mechanism should change upon nitrogen dosage.

FIG. 1.

Boron-doped and boron–nitrogen co-doped diamond films with an increase in N2/C gas ratio: (a) growth rate; (b) the intensity of NV0 + NV, where the inset is PL spectra with an excitation wavelength of 514 nm; (c) AFM images (20 × 20 μm2); and (d) rms surface roughness (Rq) and FWHM of Raman scattering spectra.

FIG. 1.

Boron-doped and boron–nitrogen co-doped diamond films with an increase in N2/C gas ratio: (a) growth rate; (b) the intensity of NV0 + NV, where the inset is PL spectra with an excitation wavelength of 514 nm; (c) AFM images (20 × 20 μm2); and (d) rms surface roughness (Rq) and FWHM of Raman scattering spectra.

Close modal

In order to evaluate the nitrogen incorporation in the samples, the PL measurement has been employed first. The inset of Fig. 1(b) shows the PL spectra of all the samples excited by a 514-nm laser. A sharp Raman peak at 552 nm labeled R is from the diamond lattice, which is used as a reference to normalize all the spectra. Besides the R peak, two peaks at 575 and 637 nm, corresponding to the NV0 and NV emissions, have been clearly observed for all the co-doped samples.30 However, the PL of the substrate and the N2/C = 0 sample is negligibly low. It clearly demonstrates that the PL related to NV centers is from the extrinsic nitrogen doping, rather than from the substrate. As shown in Fig. 1(b), the integral intensity of the NV0 + NV vs the N2 flow agrees well with the trend of the growth rate, suggesting a saturation of nitrogen incorporation in solid. Moreover, the ratio of NV to NV0 changes with the nitrogen flow rate, which implies that the charge state of the NV centers could be modulated by co-doping of nitrogen. Detailed discussion is provided in the section titled Effect of nitrogen gas flow on the NV centers charge states and Fig. S3 of the supplementary material.

Figure 1(c) shows the AFM images. The boron mono-doped sample presents a typical growth mode related to the etching effect of boron. However, the small particles may be caused by the etching effect of boron shown on the step flow at the lower nitrogen concentrations. When nitrogen is further co-incorporated, a clear step bunching growth mode is shown gradually while the small particles disappear slowly. The change in surface morphology indicates the suppression of the boron etching effect by nitrogen addition. In addition, the step height gradually decreases from 114.5 to 30.2 nm by analyzing the cross-sectional view of steps with Nova software (see the section titled “Cross-sectional view of the AFM images” for all the samples and Fig. S4 of the supplementary material). It could be due to the enhancement of surface atomic migration ability, leading to the increase in the surface smoothness for step flow with the N2/C ratio from 2.5% to 10.0%.

Moreover, both the full-width at half maximum (FWHM) of the Raman scattering spectra (Fig. S5) and the extracted roughness of all samples show “Λ” shapes with the increase in the N2 flow rate [Fig. 1(d)]. A break point at the N2/C ratio of 2.5% is also observed, indicating that the enhancement of the growth rate and the suppression of the boron etching effect by nitrogen addition mutually affect the surface and crystalline quality. The worse crystalline quality is a result of a significant enhancement of the growth rate by adding a small amount of nitrogen. An excessive addition of nitrogen (N2/C > 2.5%) is found to improve the surface and crystalline quality due to the suppression of the boron etching effect.

More interestingly, electrical properties of the samples show a conduction type change between the N2/C ratio of 2.5% and 5%, having a similar break point as presented above. The temperature dependence of the electrical parameters (resistivity, mobility, and carrier density) for all the samples is shown in Figs. S6 and S7(a)–S7(c),31 respectively. In order to verify the n-type conduction, the data for the sample grown by the N2/C ratio of 5.0% are double-checked by the Hall bar configuration, while the data for the sample grown repeatedly by the N2/C ratio of 10.0% in a 2-month time interval are measured to check the stability and repeatability of the conduction type. Figure S7d shows the Hall voltage measured vs the applied magnetic field at different temperatures for the sample grown by the N2/C ratio of 10.0%. The negative slope double-confirms the n-type conduction for the sample. In the supplementary material, we have also explained the accuracy of the Hall measurement for a very high resistivity, which is ascribed to the high-resistivity model equipped on our Hall system (see the section titled “Explained the accuracy with Hall measurements for a very high resistivity” of the supplementary material). In addition, the resistance measured between the pads for all the n-type samples has been shown in Table S1. As can be seen, the samples are quite homogeneous, which reduces the possibility of wrong conduction type identification. All the above attempts have guaranteed the authenticity of the conduction type confirmation.

As the data are quite stable, here, we take the temperature point at 773 K, as depicted in Fig. 2, for further analysis. On increasing the nitrogen gas flow, the resistivity of the samples rises monotonically from 10−1 to 2 × 106Ω cm. The corresponding Hall mobility and carrier density are also shown in Fig. 2. As can be seen, the much higher resistive samples are attributed to the reduced carrier density. Utilizing the electrical charge balance equation (for details, see the section titled “Simulations of the temperature dependence of the Hall density for the doping concentration” of the supplementary material), the doping concentration for donors and acceptors has been extracted from the temperature-dependent Hall carrier density. Results including the fitting curves and the extracted doping and compensation concentrations are shown in Fig. 3. As seen from Fig. 3(c), the simulated doping concentrations for major carriers are consistent with the measured Hall density, as shown in Fig. 2. However, the electron carrier density is much lower than the donor doping concentration, which indicates that the activation energy of the corresponding donor is relatively deep.

FIG. 2.

The resistivity, mobility, and carrier concentration at 773 K of boron-doped and boron–nitrogen co-doped diamond films.

FIG. 2.

The resistivity, mobility, and carrier concentration at 773 K of boron-doped and boron–nitrogen co-doped diamond films.

Close modal
FIG. 3.

Typical temperature dependence of (a) the hole concentration and (b) the electron concentration and (c) doping concentration and (d) compensation concentration of boron-doped and boron–nitrogen co-doped diamond films.

FIG. 3.

Typical temperature dependence of (a) the hole concentration and (b) the electron concentration and (c) doping concentration and (d) compensation concentration of boron-doped and boron–nitrogen co-doped diamond films.

Close modal

In order to obtain the energy level of the dopants, temperature-dependent resistivity has been measured in the range 1.0 < 1000/T < 3.5 K−1, as shown in Figs. 4(a) and 4(b). From the slope of the ln ρ ∼ 1000/T curves, the activation energy of the acceptors is quite small, as shown in Fig. 4(a), which is attributed to the boron acceptors. The activation energy for the N2/C = 2.5% sample is slightly higher than the 0% one, implying that there might be another acceptor form with higher activation energy existing in the N2/C = 2.5% sample, while for donors for the n-type samples, the activation energies are extracted to be 0.8–0.9 eV below the CBM, as shown in Fig. 4(b). These values are much shallower than those of the single substitutional nitrogen (Ns), implying that the form of the donor must be different to the Ns. The activation energy of the donor is further confirmed by the photoconductivity measurement employing laser diodes with different wavelengths. The photocurrent as a function of the photon energy is shown in Figs. 4(c) and 4(d) for all the samples. Abrupt absorption can be observed at the photon energy around 0.7–0.8 eV for the n-type samples, which is in good agreement with the values obtained by fitting the temperature-dependent resistivity and electron concentration, as shown in Fig. 5. Moreover, no obvious absorption can be observed on the p-type samples [Fig. 4(c)], which confirms that the activation energy around 0.8 eV is related to the donor. The photocurrent drops sharply for the N2/C = 7.5% sample compared to the 5% one mainly due to the increased resistivity, making the low-resistance contact much harder to form.

FIG. 4.

The fitting of activation energy by the relation between resistivity and temperature: (a) N2/C < 2.5% and (b) N2/C > 2.5%. Photocurrent spectroscopy with illumination wavelength between 780 and 1650 nm in a log scale. The photocurrent spectra are normalized by the light intensity: (c) N2/C < 2.5% and (d) N2/C > 2.5%.

FIG. 4.

The fitting of activation energy by the relation between resistivity and temperature: (a) N2/C < 2.5% and (b) N2/C > 2.5%. Photocurrent spectroscopy with illumination wavelength between 780 and 1650 nm in a log scale. The photocurrent spectra are normalized by the light intensity: (c) N2/C < 2.5% and (d) N2/C > 2.5%.

Close modal
FIG. 5.

The activation energy from fitting by resistivity and carrier concentration.

FIG. 5.

The activation energy from fitting by resistivity and carrier concentration.

Close modal

The above results have demonstrated the realization of n-type conductivity in diamond thin films by the boron–nitrogen co-doping technique. The donor level is around 0.8 eV below the CBM. In order to elucidate the form of the donor and the nitrogen regulation mechanism on boron doping behavior, XPS has been meticulously analyzed. The carbon, boron, and nitrogen 1s core level XPS signals have been shown in Fig. 6. The carbon 1s line is quite smooth and broad, so it is reasonable to deconvolute the peaks into a few components [Fig. 6(a)]. For the boron mono-doped sample, two peaks have been deconvoluted, and the 285.0 eV-peak is ascribed to the C–C bond from the diamond lattice, while the other small shoulder at 283.6 eV is assigned to the C–B bond. After nitrogen incorporation, the main component related to the C–C bond shifts to a lower binding energy (BE) at 284.0–284.1 eV.32 The shift of the BE is associated with combined effects of an upward band bending and a chemical shift due to the charge transfer.33 Besides the component related to the C–B bond, another peak at around 285.0 eV must be involved to fit the curve. This peak is ascribed to the various defect-induced C–N bonding states.34 

FIG. 6.

(a) C 1s, (b) B 1s, and (c) N 1s XP spectra of boron-doped and boron–nitrogen co-doped diamond films with N2/C = 0, 2.5%, 5.0%, 10.0%.

FIG. 6.

(a) C 1s, (b) B 1s, and (c) N 1s XP spectra of boron-doped and boron–nitrogen co-doped diamond films with N2/C = 0, 2.5%, 5.0%, 10.0%.

Close modal

For boron and nitrogen elements, the XPS signals [Figs. 6(b) and 6(c)] are much weaker due to their relatively low dosage. However, after long time integration, some changes due to nitrogen co-doping can be marked to some extent. For the boron 1s lines, only one peak at 189.4 eV is detectable, while two components can be deconvoluted when nitrogen is added. For the nitrogen 1s lines, two obvious components can be seen. According to the possible chemical environment and some previous literature,34–36 the two peaks of the B 1s lines are ascribed to the B–C and B–N bonds, while the two peaks of N 1 s lines are assigned to the N–B and N–C bonds, respectively.

The XPS is also known as a power tool to quantify the doping concentration of the dopants by comparing the relative area ratio among components by considering the sensitivity factor of each component.36 Therefore, in order to see the relative doping concentration of boron and nitrogen, we have made a ratio of the C–B components over the C–N ones in the C 1s XPS lines and a ratio of the B 1s lines over the N 1s lines. The results have been drawn in Fig. 7(a). As can be seen, the relative stoichiometry of B to N is quite consistent between the two ways of estimation. Generally, the trend of the result is essentially the same, indicating that the relative doping concentration of boron over nitrogen monotonically decreases as the nitrogen dosage increases. It should be noted that the calculated relative B/N doping concentration ratio is slightly different, which has been reported in some previous works.34 Furthermore, as the N2/C ratio increases from 2.5% to 5% in the gas phase, the B/N ratio decreases from >1 to <1, as shown in Fig. 7(a). It reveals that the growth condition has changed from boron-rich to nitrogen-rich, which could be connected to the conduction type change.

FIG. 7.

The various ratios as measured with XPS evolution with an increase in N2/C ratio: (a) the ratio of C–B/C–N in C 1s as well as B/N in B1s and N1s for boron–nitrogen co-doped diamond films, (b) B atoms and N atoms compared to C atoms from XPS, (c) the changes in boron concentration and nitrogen concentration from SIMS with an increase in N2/C ratio, and (d) the ratios of B–N/B–C in the boron peak and N–B/N–C ratio in the nitrogen peak from XPS for boron–nitrogen co-doped diamond films.

FIG. 7.

The various ratios as measured with XPS evolution with an increase in N2/C ratio: (a) the ratio of C–B/C–N in C 1s as well as B/N in B1s and N1s for boron–nitrogen co-doped diamond films, (b) B atoms and N atoms compared to C atoms from XPS, (c) the changes in boron concentration and nitrogen concentration from SIMS with an increase in N2/C ratio, and (d) the ratios of B–N/B–C in the boron peak and N–B/N–C ratio in the nitrogen peak from XPS for boron–nitrogen co-doped diamond films.

Close modal

In addition, the area ratio of the boron and nitrogen 1s lines over the carbon 1s lines has been calculated and plotted in Fig. 7(b). This result represents the relative doping concentration of boron ([B]) and nitrogen ([N]) upon the N2/C ratio in the gas phase. When compared to Fig. 7(a), although the B/N ratio reduces monotonically, the doping concentration of boron has a sharp increase when a small amount of nitrogen is added. Further addition of nitrogen will cause a suppression effect on boron incorporation in diamond, as reported in Ref. 37. For nitrogen, the trend accords well with those of the growth rate, the nitrogen-vacancy related emission, and the surface morphology, as previously shown in Figs. 1(a)1(c).

Since the XPS can only give a relative changing trend for [B] and [N], for accurate measurement of the B and N atomic concentration, the SIMS depth profiles of boron and nitrogen have been measured and the results are shown in Fig. S8. Figure 7(c) plots the extracted [B] and [N] from the SIMS. Note that the SIMS profile for the N2/C = 0 sample is very noisy, so we speculate that the value of ∼1016 cm-3 reaches the detection limit of [N]. The trend accords quite well with that in Fig. 7(b), indicating that the estimation of [B] and [N] from XPS is also reasonable. This consistency also double-confirms the authenticity of the doping concentration of the dopants indirectly extracted from the various characterization methods.

Nitrogen addition on the suppression of the boron dopants has been reported before.37 The interactions between BH and N2 radicals in the gas phase would lead to less boron incorporation. However, the spike of the boron incorporation at the N2/C ratio of 2.5% could be due to the formation of the B–N complexes in solid. Smaller ionic radius of boron results in tensile strain in the diamond lattice, which could be compensated and balanced by introducing nitrogen with larger size. Meanwhile, according to the first-principles calculation by Croot, the N–B complex is quite easy to form due to its negative formation energy.21 Therefore, although the boron atoms in the gas phase could be partially consumed by nitrogen addition, the efficiency of the boron incorporation in solid could be enhanced by the formation of the B–N complex. Experimentally, the nitrogen and boron doping concentrations have increased sharply when the N2/C ratio increases from 0% to 2.5%, as shown in Figs. 7(b) and 7(c), which could be evidence for the formation of the B–N complex.

When the N2/C ratio increases further above 2.5%, the suppression effect of nitrogen to boron incorporation becomes dominant and the boron concentration keeps decreasing. However, the nitrogen concentration does not decrease but gradually increases to saturation. This result could be related to the different forms of the B–N complexes in the samples grown by different N2/C ratios. Under the boron-rich condition, the B–N complexes might be in the form of B–N–B, while under the nitrogen-rich condition, the complexes could be in the form of either symmetric N–B–N or asymmetric N–N–B.21 Electrically, the B–N–B cluster acts as an acceptor, while the N–B–N and N–N–B clusters act as donors with relatively shallow and deep activation energy, respectively. As discussed above, the infection of the B/N ratio from >1 to <1 is between the N2/C ratio of 2.5%–5% [Fig. 7(a)], and it is thus reasonable that the sample grown with the N2/C ratio of 2.5% is boron-rich, while the other co-doping samples are nitrogen-rich. As a result, in Figs. 7(b) and 7(c), although the boron concentration decreases, due to the transformation of B–N–B to N–B–N and/or N–N–B, the nitrogen concentration does not decrease.

The ratios of the B–N/B–C in the boron peak and the N–B/N–C in the nitrogen peak also give some information supporting the formation of B2N and N2B complexes. We have plotted the B–N/B–C ratio extracted from the B 1s XPS in Fig. 7(d). The ratio gradually increases, indicating that the B–N complex may change from the B-rich form (B–N–B) to N-rich form (N–B–N). For the N–B/N–C ratio in the N 1s XPS, it also increases with the N2/C ratio, which seems opposite. However, when looking at Fig. 6(c), the N–C peak is quite broad for the N2/C 2.5% sample, implying that nitrogen in the N–C bonds may have multiple chemical states, such substitutional nitrogen and/or NV centers. If one deconvolutes the 400.3 eV broad peak and takes the 399.3 eV component to calculate the N–B/N–C ratio, it would decrease sharply when the N2/C ratio changes from 2.5% to 5%. It is also an indication of a possible change in the B–N complex form.

Moreover, as characterized by the nitrogen-vacancy related emission in Fig. 1(b), the trend of the nitrogen concentration variation upon the N2/C ratio could also be partly attributed to the enhanced incorporation of substitutional nitrogen on carbon sites (NS). Consequently, the N–C component in N 1s XPS lines should include the contribution from both the NS and the N–B complexes. As can be seen in Fig. 6(c), the binding energy of the N–C component shifts from 400.3 to 399.3–399.4 eV when the growth condition shifts from boron-rich to nitrogen-rich. This shift could be due to the gradual formation of the NS defects and the reduced concentration of the B–N complex. Meanwhile, a slight shift of the B–C component from 187.1 to 187.6 eV is also observed in the B 1s XPS lines [Fig. 6(b)]. This shift could be understood by the fact that the binding energy of the B 1s electrons in the N–B–C3 configuration (corresponding to boron in the B–N–B complex) would be a little lower than that in the NC–B–CN one (corresponding to nitrogen in the N–B–N complex) since carbon (2.55) is less electronegative than nitrogen (3.04). This result is also evidence of the transformation of the B–N complex forms from the N2/C ratio of 2.5% to 5%.

Such a transformation leads to the conduction type conversion eventually. For the two p-type samples, although boron concentration has been enhanced by nitrogen addition, the formation energy of the NS is rather small as the Fermi level is close to the VBM (the p-type case).38 The formation of the NS could critically compensate the B–N–B complex acceptors. Lower hole concentration with lower mobility and worse crystalline quality is observed for the co-doped p-type sample as compared to the boron mono-doped case. For the n-type samples, the electron density monotonously decreases. It could be due to the reduced concentration of the N–B–N complexes since the boron incorporation is inadequate to provide more N–B–N formation. Meanwhile, due to the decrease in the electron density, the Fermi level moves down and the formation of the NS becomes relatively favored.38 The reduction of the N–B–N donors and the rise of the NS defects, as evidenced above by the N–C peak shift in N 1s XPS, would lead to a stable concentration of nitrogen but much fewer free electrons in the conduction band.

Utilizing the developed B/N co-doping technique, we have fabricated a p-i-n diamond homogeneous diode on the Ib (100) substrate. After the epilayer growth, an etching to the boron-doped p-type layer is done by the inductive coupled plasma etching machine. Figure S9 shows the electrical properties of the p+-type doped layer, giving a net acceptor concentration at 2.3 × 1020 cm-3. Two titanium-gold pads on the alleged n-type and p-type layers are patterned and fabricated by photolithography and e-beam evaporation. Then, Ti (30 nm)/Au (100 nm) electrodes of 200 μm diameter are annealed at 500 °C for 1 h to ensure stable Ohmic contacts. The schematic is shown in Fig. 8(a), and the thickness of each of these layers is estimated by the growth rate. An i-layer with a thickness of ∼0.3 μm is formed between the p- and n-layers. The B concentration and thickness are 2.3 × 1020 cm-3 and ∼0.5 μm for the p+-layer with B2/C = 150 ppm and 1.1 × 1018 cm-3 and ∼1.5 μm for the p-layer with B2/C = 50 ppm, respectively. The donor concentration and thickness are 1.7 × 1015 cm-3 and ∼1.7 μm, respectively, for the n-layer with N2/C = 5.0%. Figure 8(b) shows the measured I–V curve for the device at 300 K. A clear rectification behavior with an ON/OFF ratio of 104 could be measured. The characteristic demonstrates the n-type conduction of the B/N co-doped layer.

FIG. 8.

Schematic structure of (a) diamond PIN and (b) I–V characteristics.

FIG. 8.

Schematic structure of (a) diamond PIN and (b) I–V characteristics.

Close modal

In conclusion, nitrogen modulation on the boron doping behavior for n-type diamonds by MPCVD has been systematically studied. On increasing the nitrogen concentration in the gas phase, the form of the boron dopants changes from boron-rich acceptor-like B–N–B to nitrogen-rich donor-like N–B–N or N–N–B. This transformation of the B–N complex form is responsible for the observed conduction type conversion. An appropriate nitrogen dosage has led to a measurable n-type conductivity at 773 K. The activation energy for the corresponding donor is confirmed to be around 0.8 eV, which is ascribed to the complex donor in the form of N–B–N. The compensation effect from the boron atoms, which is considered to be highly detrimental for efficient n-type doping in diamond, could be suppressed by such a nitrogen co-incorporation technique. The current study has provided some feasible clues to realize room temperature effective n-type diamonds in the future.

See the supplementary material for the sample configuration for the Hall measurements, details of the fabrication process of the bar-structured sample, the confirmation of the growth rate, the effect of nitrogen gas flow on the charge state of NV centers, the cross-sectional view of the AFM images for all the samples, the Raman scattering spectra, the electrical properties for p-type conductivity films, the relation between electrical properties and temperature of n-type films with N2/C higher than 2.5%, the accuracy with Hall measurements for a very high resistivity, the resistance measured between the pads and the measured sheet resistance for all the n-type diamond samples, the simulations for the doping concentration, the SIMS depth profiles of boron concentration and nitrogen concentration, and the boron concentration of the p+-layer fitted by the relation between temperature and hole concentration.

The authors acknowledge financial support from the National Key R&D Program of China (Grant Nos. 2018YFB0406502, 2017YFF0210800, and 2017YFB0403003), the National Natural Science Foundation of China (Grant Nos. 61974059, 61674077, and 61774081), the Natural Science Foundation of Jiangsu Province (Grant No. BK20160065), and the Fundamental Research Funds for the Central Universities.

The data that support the findings of this study are available from the corresponding authors upon reasonable request and are available within the article and its supplementary material.

1.
H.
Zeng
,
P. U.
Arumugam
,
S.
Siddiqui
, and
J. A.
Carlisle
,
Appl. Phys. Lett.
102
,
223108
(
2013
).
2.
G.
Chicot
,
T. N.
Tran Thi
,
A.
Fiori
,
F.
Jomard
,
E.
Gheeraert
,
E.
Bustarret
, and
J.
Pernot
,
Appl. Phys. Lett.
101
,
162101
(
2012
).
3.
S. J.
Cobb
,
Z. J.
Ayres
, and
J. V.
Macpherson
,
Annu. Rev. Anal. Chem.
11
,
463
(
2018
).
4.
T. H.
Borst
and
O.
Weis
,
Phys. Status Solidi A
154
,
423
(
1996
).
5.
E. A.
Ekimov
,
V. A.
Sidorov
,
E. D.
Bauer
,
N. N.
Mel’nik
,
N. J.
Curro
,
J. D.
Thompson
, and
S. M.
Stishov
,
Nature
428
,
542
(
2004
).
6.
H.
Kato
,
M.
Ogura
,
T.
Makino
,
D.
Takeuchi
, and
S.
Yamasaki
,
Appl. Phys. Lett.
109
,
142102
(
2016
).
7.
R. G.
Farrer
,
Solid State Commun.
7
,
685
(
1969
).
8.
M.
Hasegawa
,
Y.
Yamamoto
,
H.
Watanabe
,
H.
Okushi
,
M.
Watanabe
, and
T.
Sekiguchi
,
Diamond Relat. Mater.
13
,
600
(
2004
).
9.
R.
Kalish
,
C.
Uzan-Saguy
,
B.
Philosoph
,
V.
Richter
,
J. P.
Lagrange
,
E.
Gheeraert
,
A.
Deneuville
, and
A. T.
Collins
,
Diamond Relat. Mater.
6
,
516
(
1997
).
10.
M.
Katagiri
,
J.
Isoya
,
S.
Koizumi
, and
H.
Kanda
,
Appl. Phys. Lett.
85
,
6365
(
2004
).
11.
F. A. M.
Koeck
,
R. J.
Nemanich
,
A.
Lazea
, and
K.
Haenen
,
Diamond Relat. Mater.
18
,
789
(
2009
).
12.
M.
Frenklach
and
H.
Wang
,
Phys. Rev. B
43
,
1520
(
1991
).
13.
B. D.
Yu
,
Y.
Miyamoto
, and
O.
Sugino
,
Appl. Phys. Lett.
76
,
976
(
2000
).
14.
A. S.
Barnard
,
S. P.
Russo
, and
I. K.
Snook
,
Philos. Mag.
83
,
1163
(
2003
).
15.
H.
Katayama-Yoshida
,
T.
Nishimatsu
,
T.
Yamamoto
, and
N.
Orita
,
J. Phys.: Condens. Matter
13
,
8901
(
2001
).
16.
H.
Katayama‐Yoshida
,
T.
Nishimatsu
,
T.
Yamamoto
, and
N.
Orita
,
Phys. Status Solidi B
210
,
429
(
1998
).
17.
D.
Segev
and
S.-H.
Wei
,
Phys. Rev. Lett.
91
,
126406
(
2003
).
18.
J. E.
Moussa
,
N.
Marom
,
N.
Sai
, and
J. R.
Chelikowsky
,
Phys. Rev. Lett.
108
,
226404
(
2012
).
19.
J.
Isoya
,
H.
Kanda
, and
Y.
Morita
,
Phys. Rev. B
56
,
6392
(
1997
).
20.
N.
Ishii
and
T.
Shimizu
,
Phys. Rev. B
58
,
12533
(
1998
).
21.
A.
Croot
,
M. Z.
Othman
,
S.
Conejeros
,
N. A.
Fox
, and
N. L.
Allan
,
J. Phys.: Condens. Matter
30
,
425501
(
2018
).
22.
D. Y.
Liu
,
L. C.
Hao
,
Z. A.
Chen
,
W. K.
Zhao
,
Y.
Shen
,
Y.
Bian
,
K.
Tang
,
J. D.
Ye
,
S. M.
Zhu
,
R.
Zhang
,
Y. D.
Zheng
, and
S. L.
Gu
,
Appl. Phys. Lett.
117
,
022101
(
2020
).
23.
L. C.
Hao
,
Y.
Shen
,
X. D.
Yang
,
Y.
Bian
,
Q. Q.
Du
,
D. Y.
Liu
,
W. K.
Zhao
,
J. D.
Ye
,
K.
Tang
,
H. P.
Wu
,
R.
Zhang
,
Y. D.
Zheng
, and
S. L.
Gu
,
J. Phys. D: Appl. Phys.
53
,
075107
(
2019
).
24.
M.-A.
Pinault-Thaury
,
B.
Berini
,
I.
Stenger
,
E.
Chikoidze
,
A.
Lusson
,
F.
Jomard
,
J.
Chevallier
, and
J.
Barjon
,
Appl. Phys. Lett.
100
,
192109
(
2012
).
25.
A.
Kumar
,
J.
Pernot
,
F.
Omnès
,
P.
Muret
,
A.
Traoré
,
L.
Magaud
,
A.
Deneuville
,
N.
Habka
,
J.
Barjon
,
F.
Jomard
,
M. A.
Pinault
,
J.
Chevallier
,
C.
Mer-Calfati
,
J. C.
Arnault
, and
P.
Bergonzo
,
J. Appl. Phys.
110
,
033718
(
2011
).
26.
T.
Ohgaki
,
N.
Ohashi
,
S.
Sugimura
,
H.
Ryoken
,
I.
Sakaguchi
,
Y.
Adachi
, and
H.
Haneda
,
J. Mater. Res.
23
,
2293
(
2008
).
27.
O.
Bierwagen
,
T.
Ive
,
C. G.
Van de Walle
, and
J. S.
Speck
,
Appl. Phys. Lett.
93
,
242108
(
2008
).
28.
T.
Liu
and
D.
Raabe
,
Appl. Phys. Lett.
94
,
021119
(
2009
).
29.
S.
Dunst
,
H.
Sternschulte
, and
M.
Schreck
,
Appl. Phys. Lett.
94
,
224101
(
2009
).
30.
H.
Ozawa
,
Y.
Hatano
,
T.
Iwasaki
,
Y.
Harada
, and
M.
Hatano
,
Jpn. J. Appl. Phys., Part 1
58
,
SIIB26
(
2019
).
31.
J.-P.
Lagrange
,
A.
Deneuville
, and
E.
Gheeraert
,
Diamond Relat. Mater.
7
,
1390
(
1998
).
32.
Y.
Li
,
X.
Jia
,
W.
Shi
,
S.
Leng
,
H.-a.
Ma
,
S.
Sun
,
F.
Wang
,
N.
Chen
, and
Y.
Long
,
Int. J. Refract. Met. Hard Mater.
43
,
147
(
2014
).
33.
M.
Chandran
,
M.
Shasha
,
S.
Michaelson
, and
A.
Hoffman
,
Appl. Phys. Lett.
107
,
111602
(
2015
).
34.
B. J.
Matsoso
,
K.
Ranganathan
,
B. K.
Mutuma
,
T.
Lerotholi
,
G.
Jones
, and
N. J.
Coville
,
New J. Chem.
41
,
9497
(
2017
).
35.
N.
Yu
,
F.
Romero‐Borja
,
Z. H.
Zhang
,
X. T.
Cui
,
J. R.
Liu
,
L. T.
Wood
,
W. K.
Chu
,
D.
Marton
,
J. W.
Rabalais
,
K. M.
Forster
, and
R. R.
Reeber
,
Appl. Phys. Lett.
63
,
1643
(
1993
).
36.
J. H.
Yang
,
S. Y.
Lee
,
M. H.
Yum
,
Y. S.
Shin
,
C. H.
Yang
,
N.-K.
Park
,
C.-Y.
Park
,
Y.-R.
Jang
,
M. H.
Kwon
,
C.
Jo
,
J.-I.
Lee
,
C.-W.
Yang
,
J.-B.
Yoo
,
K.
Ihm
,
T.-H.
Kang
,
Y.-W.
Jin
, and
J.-M.
Kim
,
J. Korean Phys. Soc.
51
,
125
(
2007
).
37.
A.
Croot
, “
Boron and nitrogen in diamond: An ab initio simulation, plasma emission spectroscopy and material deposition and characterisation study
,” Ph.D. thesis,
University of Bristol
,
2018
.
38.
K.
Czelej
,
P.
Śpiewak
, and
K. J.
Kurzydłowski
,
Diamond Relat. Mater.
75
,
146
(
2017
).

Supplementary Material