The Charge Transport Properties of Polycrystalline CVD Diamond Films Deposited on Monocrystalline Si Substrate

Abstract

In this work, diamond/Si heterojunctions were fabricated by synthesizing a diamond layer directly on a monocrystalline n-type Si substrate. The diamond layers were characterized using micro-Raman spectroscopy, scanning electron microscopy (SEM), and X-ray diffraction (XRD). The current–voltage (I–V) characteristics of the heterojunctions were measured at room temperature. The heterojunctions exhibited rectifying behavior, confirming their diode-like nature. Based on thermionic emission theory, key electrical parameters of the heterojunction diodes—including the ideality factor (n) and carrier mobility (μ)—were estimated from the I–V characteristics. The I–V curves revealed large ideality factors ranging from 1.5 to 6.5, indicating the presence of deep trap states with densities between 2 × 10¹⁵ and 8 × 10¹⁶ eV⁻¹·cm⁻³. These variations were attributed to differences in the structural quality of the diamond layers and the effects of surface hydrogen termination.

1. Introduction

Diamond is a wide-bandgap semiconductor (Eg ≈ 5.47 eV) known for its high breakdown voltage, excellent thermal conductivity, low dielectric constant, and superior radiation hardness, making it attractive for electronic devices operating under high temperature, high power, or radiation-intensive conditions [1]. These properties make diamond particularly promising for heterojunction fabrication, where its robustness can be combined with the mature technology of silicon. Diamond/silicon interfaces have already been investigated for applications in rectifying diodes, Schottky devices, and radiation detectors.

The electrical properties of such heterostructures depend strongly on crystalline quality, defect density, and surface termination. Previous studies have demonstrated that boron-doped or single-crystal CVD diamond exhibits excellent charge transport characteristics but requires demanding growth conditions. For example, Isberg et al. [2] reported long charge collection distances in high-quality single-crystal diamond, while Pan et al. [3] showed that polycrystalline undoped films suffer from mobility reduction due to grain boundaries and extended defects. Similarly, Pleskov et al. [4] compared undoped and boron-doped films, demonstrating that hole mobility is strongly dependent on microstructure. Other authors highlighted the role of inhomogeneities at diamond/Si interfaces in determining barrier heights and diode ideality factors [5,6]. These findings emphasize that both junction-limited and bulk-limited transport processes must be considered in polycrystalline diamond devices.

Despite these advances, most prior research has relied on relatively complex techniques such as charge collection distance (CCD) [2], current transient spectroscopy (CTS) [7], or photoconductance methods [8]. These approaches provide detailed information about trap states and recombination centers but are not always practical for routine evaluation of device-quality films. A simpler method is to analyze current–voltage (I–V) characteristics, from which transport parameters such as ideality factor, mobility, and trap density can be extracted. In particular, the slope of ln(J)–V and log(J)–log(V) plots can be used to distinguish between thermionic emission, ohmic conduction, and space-charge-limited conduction, offering insights into the relative roles of the depletion region and bulk defects in controlling transport.

Another important aspect is the effect of doping and surface chemistry. Doping diamond remains challenging due to the deep activation energies of boron (0.37 eV), nitrogen (1.7 eV), and phosphorus (0.6 eV), which limit dopant activation at room temperature [9,10]. Moreover, as-grown CVD diamond films typically exhibit a conductive hydrogen-terminated surface layer [11,12,13]. Hydrogen plays a crucial role in defect passivation and the formation of negative electron affinity surfaces, which can substantially influence charge transport at interfaces [14,15,16,17].

Building on these insights and our earlier work [18,19,20], the present study focuses on undoped polycrystalline CVD diamond films grown on silicon substrates using the hot filament CVD method. Instead of optimizing technological conditions, the aim is to establish correlations between crystalline quality, trap density, and transport properties. By combining SEM, XRD, Raman, and electrical characterization, we demonstrate how variations in structural quality affect the electronic behavior of p-diamond/n-Si heterojunctions. This approach provides a straightforward framework for understanding transport in diamond-based devices without resorting to more complex spectroscopic methods.

2. Materials and Methods

Polycrystalline diamond films (~4–6 μm thick) were deposited on (100)-oriented n-type Si substrates (3.5 Ω·cm) using hot filament CVD (HF CVD). A tungsten filament (2300 K) was positioned 6 mm above the substrate in a 2.3%–2.75% CH₄/H₂ gas mixture. Substrates were polished with 0.2 μm diamond paste, seeded in an ultrasonic bath with nano/microdiamond powders in methanol (30 min), and cleaned with alcohol and acetone (5 min). Gold contacts (5 mm diameter) were thermally evaporated onto the diamond surface and the backside of the Si substrates.

Table 1. The CVD samples growth process parameters.

Sample	Filament Temp. [K]	Gas Flow Rate [sccm]	Substrate Temp. [K]	CH4/H2 Ratio [%]	Deposition Pressure [mbar]
PDF34	2300 ± 50	100 ± 5		2.60	20
PDF27				2.60	80
PDF33				2.30	100
PDF20			980 ± 30	2.35	40
PDF23				2.55	60
PDF15				2.35	20
PDF22				2.75	60

summarizes the synthesis parameters, including filament temperature, gas flow rate, CH₄/H₂ ratio, substrate temperature, and deposition pressure.

The deposition rate was approximately 0.4–0.5 μm/h, consistent with values reported in the literature [21].

Characterization Methods:

•

Morphology: Scanning Electron Microscopy (SEM, JEOL JSM-820, 25 kV).

•

Structural Analysis: Raman spectroscopy (Renishaw inVia, 488 nm laser, ±2 cm⁻¹ resolution) and X-ray diffraction (XRD, DRON-4a Θ–2Θ diffractometer, Cu Kα radiation, 32 kV, 12 mA).

•

Electrical Measurements: I–V characteristics were recorded at room temperature using an Oxford Optistat cryostat. A 4–20 V rectangular waveform was applied (Rigol DG1022A), and current/voltage data were collected using a Keithley 6485 picoammeter and Fluke 8505A multimeter.

3. Results

3.1. Structural Characterization (SEM, XRD)

The layer thickness ranges from 4–6 μm, confirmed by both cross-sectional SEM (Figure 1c, blue line indicates Si substrate) and gravimetric analysis. PDF34 and PDF15, shown in Figure 1, exhibit the largest and smallest surface microcrystals, respectively. The remaining samples show very similar morphologies from the SEM point of view. Crystallite sizes were estimated from the X-ray diffraction (XRD) patterns (Figure 2a), where all samples displayed characteristic diamond reflections: (111), (220), and (331). In each case, the (220) reflection was the strongest in each case. Based on the full width at half maximum (FWHM), the dimension of the crystallites can be determined to be 57–71 nm, using the Scherrer formula [9].

Table 2. Structural parameters of diamond samples.

Sample	FWHM [cm−1]	Diamond Quality Q [%]	L(220) [nm]
PDF34	7.6	97.65	71
PDF27	8.65	98.06	67
PDF33	8.49	97.89	66
PDF20	8.03	98.64	59
PDF23	7.51	98.53	63
PDF15	6.83	98.90	57
PDF22	9.26	98.30	65

L₍₂₂₀₎—crystallite sizes in <220> direction.

provides individual crystallite sizes L(220) for each sample. SEM images reveal large grains, while XRD indicates smaller crystallite domains within these grains. Grain boundaries likely contain defects such as dislocations, microvoids, and hydrogen inclusions, especially near the diamond/silicon interface, as seen in the cross-section.

3.2. Structural Characterization (Raman)

The quality of the diamond structure was checked by Raman spectroscopy, shown in Figure 2b. The Raman spectra for all samples show similar features, i.e., sharp Raman lines peaked at 1331.5 cm⁻¹ with the FWHM ranging from 7.5 to 9.5 cm⁻¹. Moreover, all Raman spectra exhibit a weak broad band at around 1530 cm⁻¹, even in the case of the PDF15 sample. It is attributed to amorphous carbon phases, which are likely distributed along grain boundaries [22]. This peak is the so-called G-band.

In order to estimate diamond quality, the simplest way is to use equation [23]:

$$Q={\displaystyle \frac{I_D}{I_D+\frac{I_G}{50}}},$$

(1)

where: I_D and I_G are integral intensities of the diamond and the G-band respectively.

The G-band plays a key role which is a prominent feature typically observed in graphitic carbon materials, including graphene, graphite, carbon nanotubes, and nanocrystalline diamond. It corresponds to the E_2g phonon mode at the Brillouin zone center, involves in-plane stretching of sp² carbon bonds. The position and intensity of the G-band can be influenced by several key factors: strain shift (up/down), doping shift (can increase or decrease), disorder possible shift, or substrate indirect effect [16,23,24].

The structural parameters derived from X-ray and Raman spectroscopy are collected in Table 2.

Interestingly, despite having the smallest crystallites, the PDF15 sample exhibits the highest structural quality, as indicated by its Q factor and the narrowest Raman FWHM.

The improved quality of PDF15 can be attributed to reduced extended defects within the crystallite domains, despite its smaller size. This suggests that Raman FWHM and the Q factor are more sensitive to internal crystal perfection than to crystallite size alone.

This microstructural complexity significantly influences the electronic properties of the diamond/silicon heterostructures discussed later.

3.3. Electrical Transport Properties of the p-Diamond/n-Si Heterojunctions

Figure 3 shows the current–voltage (I–V) characteristics of the fabricated p-diamond/n-Si heterojunctions measured at room temperature (RT). Among the tested devices, the PDF15 sample exhibits the strongest rectifying behavior, confirming its superior junction quality.

The current transport across the junction evolves through distinct regimes depending on the applied forward bias voltage. These regimes are described by thermionic emission theory, ohmic conduction, and space-charge-limited conduction (SCLC), each associated with a characteristic voltage range and corresponding analytical expression.

3.4. Thermionic Emission Regime (Junction-Limited Transport)

At small forward bias, the conduction is governed by thermionic emission across the junction barrier, described by [25]:

$$\mathrm{J}={\mathrm{J}}_0\left[\mathrm{e}\mathrm{x}\mathrm{p}\left({\displaystyle \frac{\mathrm{q}\mathrm{V}}{\mathrm{n}\mathrm{k}\mathrm{T}}}\right)-1\right]$$

(2)

where J₀ is the reverse saturation current density, n is the ideality factor, and the other symbols have their usual meaning.

For qV >> kT, Equation (2) predicts an exponential rise of current, and a linear relation between ln(J) and V is obtained. The slope of this region provides the ideality factor n, which reflects junction quality and recombination processes.

Here’s a schematic voltage-regime map that visually connects the three conduction mechanisms:

3.5. Ohmic Region (Bulk-Limited Transport, Very Low Bias: V < V_T)

When the applied voltage is below a threshold voltage V_T, the conduction through the diamond layer is limited by the drift of thermally activated holes. In this region, the slope of ln(J) versus ln (V) is ~1 (Figure 4), indicating a linear J–V dependence.

The current density follows the ohmic law [26]:

$$\mathrm{J}={\mathrm{p}}_0\mathrm{q}{\mathsf{\mu }}_{\mathrm{p}}{\displaystyle \frac{\mathrm{V}}{\mathrm{d}}},$$

(3)

where: p₀ is the equilibrium hole concentration, μ_p is the hole mobility, and d is the diamond layer thickness.

This represents trap-free conduction, where injected carriers are negligible and the electric field is uniform across the diamond.

3.6. Space-Charge-Limited Conduction (Trap-Controlled Transport, V > V_{T )}

For voltages beyond V_T, injected carriers dominate over the equilibrium concentration, and the transport shifts to the space-charge-limited current (SCLC) regime. In the log–log plot, the slope increases to ~2 or higher, characteristic of quadratic dependence on applied voltage.

For a single dominant trap level, the current density is given by [27]:

$$\mathrm{J}={\displaystyle \frac{9}{8}}{\mathsf{ϵ}}_0\mathsf{ϵ}\mathsf{\theta }\left(\mathrm{T}\right){\mathsf{\mu }}_{\mathrm{p}}{\displaystyle \frac{{\mathrm{V}}^2}{{\mathrm{d}}^3}},$$

(4)

where ε is the vacuum permittivity, ε = 5.7 is the dielectric constant of diamond, and θ(T) is the trapping factor.

Forward current is limited by bulk traps in the diamond layer, as evidenced by SCLC behavior. Thus, the depletion region is Si-dominated, while current limitation is diamond-dominated.

This relation is confirmed experimentally by the linear dependence of J on V² (Figure 5). The slope of these curves provides direct access to carrier mobility and trap parameters.

3.7. Trap Density and Carrier Mobility

The trapping factor θ(T) is linked to the trap density N_t and trap depth E_t by [28]:

$$\mathsf{\theta }\left(\mathrm{T}\right)={\displaystyle \frac{{\mathrm{N}}_{\mathrm{V}}}{{\mathrm{N}}_{\mathrm{t}}\left(\mathrm{T}\right)}}={\displaystyle \frac{{\mathrm{N}}_{\mathrm{V}}}{{\mathrm{N}}_{\mathrm{t}}}}\mathrm{e}\mathrm{x}\mathrm{p}\left(-{\displaystyle \frac{{\mathrm{E}}_{\mathrm{t}}}{\mathrm{k}\mathrm{T}}}\right),$$

(5)

with N_V ≈ 10¹⁹ cm⁻³ as the effective valence band density of states.

The density of traps can also be estimated from the ideality factor n using [5,29,30,31,32]:

$$\mathrm{n}={\displaystyle \frac{\mathrm{q}}{\mathrm{k}\mathrm{T}}}{\displaystyle \frac{\mathrm{d}\mathrm{V}}{\mathrm{d}\left(\mathrm{l}\mathrm{n}\mathrm{J}\right)}}={\displaystyle \frac{\mathrm{q}{\mathrm{E}}_{00}}{\mathrm{k}\mathrm{T}}}\mathrm{c}\mathrm{o}\mathrm{t}\mathrm{h}\left[{\displaystyle \frac{\mathrm{q}{\mathrm{E}}_{00}}{\mathrm{k}\mathrm{T}}}\right],$$

(6)

where:

$${\mathrm{E}}_{00}={\displaystyle \frac{\mathrm{h}}{4\mathsf{\pi }}}{\left[{\displaystyle \frac{{\mathrm{N}}_{\mathrm{t}}}{{\mathrm{m}}^{*}{\mathsf{ϵ}}_0\mathsf{\epsilon }}}\right]}^{1/2}$$

(7)

Combining the estimates of N_t from Equation (7) and θ(T) from Equation (5), the carrier mobility μ_p is extracted using Equation (4). The obtained parameters are summarized in

Table 3. The estimated electronic parameters for all samples.

Sample	Ideality Factor n	E00 [meV]	Nt × 1016 [eV−1cm−3]	Mobility μp × 10−2 [cm2/Vs]
PDF34	6.4	159.6	8.915	0.143
PDF27	4.2	105.0	5.139	0.69
PDF33	3.6	90.1	4.850	0.501
PDF20	2.0	52.3	0.956	1.412
PDF23	2.5	63.1	2.049	1.111
PDF15	1.6	38.5	0.519	1.867
PDF22	4.1	101.2	3.571	1.061

Note: n is determined from the I–V characteristics for the voltage range 0–0.3 V.

.

3.8. Discussion of Transport Mechanisms

The extracted mobilities and trap densities are consistent with literature [3,4,6,8,24,33,34]. However, mobilities remain significantly lower than in natural single-crystal diamond, reflecting the effect of impurities, grain boundaries, and structural defects in CVD-grown films.

Figure 6a,b confirm that diamond quality strongly affects mobility and trap density: better crystalline quality corresponds to lower trap density and higher mobility. As shown in Figure 6c, mobility decreases as trap density increases, consistent with the relation [6]:

$${\mathsf{\mu }}_{\mathrm{p}}=\mathrm{L}\mathrm{q}{\left({\displaystyle \frac{1}{2\mathsf{\pi }{\mathrm{m}}_{\mathrm{p}}^{*}\mathrm{k}\mathrm{T}}}\right)}^{1/2}\exp \left[-{\displaystyle \frac{\mathrm{q}{\mathrm{N}}_{\mathrm{t}}^2}{8\mathsf{\epsilon }\mathrm{k}{\mathrm{N}}_{\mathrm{A}}\mathrm{T}}}\right],$$

(8)

where L is the crystallite size and N_A is the dopant concentration.

Finally, Figure 6d demonstrates that mobility also decreases with increasing carrier concentration, due to enhanced carrier–carrier scattering [35,36]. This confirms that the main limiting factors for conductivity in diamond films are morphology, crystalline quality, and defect-related trap states.

4. Conclusions

In this study, we analyzed the structural and electronic properties of undoped polycrystalline diamond films deposited on silicon substrates. Raman spectroscopy confirmed that the amorphous carbon content remained largely consistent across different growth conditions.

The p-diamond/n-Si heterojunctions exhibited rectifying behavior consistent with thermionic emission theory. Film quality significantly influenced the electronic properties, particularly charge carrier mobility and trap state density.

Higher structural quality corresponded to lower trap densities and higher carrier mobilities, which ranged from 0.00143 to 0.01867 cm²/V·s. Increased trap density and carrier concentration were found to reduce mobility, indicating that grain boundary trapping is the dominant mobility-limiting factor.