The Amorphous Carbon Layers Deposited by Various Magnetron Sputtering Techniques

Abstract

This study investigates the synthesis and characterization of amorphous carbon (a-C) layers using three magnetron sputtering (MS) techniques: Pulsed MS (PMS), Gas Injection MS (GIMS), and High Power GIMS (HiPGIMS). The primary objective was to understand how these methods influence the sp³/sp² hybridization ratio, a critical parameter for tailoring the properties of amorphous carbon. Plasma diagnostics via Optical Emission Spectroscopy revealed distinct discharge characteristics, with HiPGIMS exhibiting the highest current density and plasma ionization. Structural and compositional analyses using Raman Spectroscopy and X-ray Photoelectron Spectroscopy (XPS) demonstrated a clear trend: sp³ content increased significantly from PMS to GIMS to HiPGIMS, reaching up to 50% (Raman) and 39% (XPS). This enhancement is attributed to the higher plasma density and more energetic ion bombardment in HiPGIMS, which promotes the formation of sp³ bonds. Ellipsometric spectroscopy further supported these findings, showing that HiPGIMS produced layers with the widest bandgap, indicative of higher sp³ content. The research highlights the effectiveness of advanced MS techniques, particularly HiPGIMS, in precisely controlling the sp³/sp² ratio and thereby the electrical, optical, and mechanical properties of a-C layers for various applications.

1. Introduction

Amorphous carbon (a-C) layers are incredibly versatile materials that are important in industry and scientific research [1]. Their unique properties, including exceptional hardness [2], low friction [3], excellent chemical inertness [4], and biocompatibility [5], make them ideal for a wide range of applications. a-C layers are extensively used to enhance the performance and lifespan of mechanical components, tools, and medical implants.

Scientifically, the tunable nature of a-C allows for the exploration of novel electronic and optical properties [6]. This tunability stems from the varying ratios of sp² and sp³ hybridized carbon atoms within the amorphous structure [7]. This control enables the development of advanced materials for sensors [8,9], microelectronics [6,10], and energy storage devices [11,12]. Furthermore, their inertness makes them valuable as protective layers in harsh chemical environments [13,14] and as substrates for advanced biological studies [15,16].

The ability to tailor their characteristics makes amorphous carbon layers a cornerstone for innovation across diverse scientific and industrial frontiers. The properties of a-C layers are profoundly influenced by the relative proportion of sp³/sp² bonding between carbon atoms [17]. The sp³ tetrahedral bonding dictates the layer’s hardness, density, and high elastic modulus. Conversely, sp²-bonded carbon contributes to the layer’s graphitic character, which in turn influences its electrical conductivity and optical absorption. Therefore, the sp³/sp² ratio is a critical parameter in tailoring a-C layers for specific applications, allowing for the precise control of their mechanical, electrical, and optical characteristics, ranging from hard, diamond-like layers to more flexible, graphite-like layers.

The a-C layers can be deposited using various methods, each offering unique advantages and characteristics. Among the most common techniques are Magnetron Sputtering (MS) [18,19], Plasma-Enhanced Chemical Vapor Deposition [20,21], Laser Ablation [22,23], Filtered Cathodic Vacuum Arc [24,25], and Arc Deposition [26,27], among others. There is noticeable interest in using the MS techniques for a-C layers synthesis. This is due to the high scalability of the MS for industrial production, the high uniformity of the layers, and its high sensitivity to deposition conditions. This sensitivity enables the tailoring of the layers’ characteristics, chemical composition, phase composition, and structural features. In the context of a-C layers, controlling the sp³/sp² ratio is crucial. Several technological MS parameters control the sp³/sp² ratio, including sputtering pressure [28,29], substrate temperature [30,31], substrate bias [32,33], sputtering gas [34,35], and power density [36,37], among others. The explanation behind the sp³/sp² ratio is the active role of ions during the phase and structure formation of the layer. A common concept in the literature concerns the shallow implantation of energetic species (up to 100 eV) in an amorphous sp² carbon matrix. This model assumes that energetic species transfer their kinetic energy into reconfiguring the surrounding atoms during penetration, creating the sp³ hybridized volume of the matrix [38,39]. This mechanism is primarily expected to occur under conditions favoring rather low dissipation of the kinetic energy in inelastic collisions with cold gas molecules and/or with the substrate under negative biasing. The other mechanism explaining the formation of the sp³ phase is the homogeneous nucleation on ions as the crystallization centers [40,41]. This mechanism suggests that valence electrons in carbon clusters can be significantly excited through non-elastic collisions with plasma electrons. The excitation of these electrons favors the creation of sp³ bonds at the expense of sp² bonds. The ultrasmall clusters prevent phonon excitations, so the energy exchanged can be used to create the metastable phase. After reaching the surface of the substrate, the clusters form a sp³-rich carbon film.

In this work, we wanted to address the issue of synthesizing a-C layers using various magnetron sputtering techniques but utilizing common apparatus components. Thus, it is important to have the ability to analyze the results of layer deposition resulting from different methods of plasma excitation and their propagation conditions. For this purpose, we used the following techniques:

• Pulsed Magnetron Sputtering (PMS)—a technique that can be considered a technological standard of MS, characterized by a low degree of plasma component excitation and a power density in the range of 10⁰–10² W/cm² [42,43];
• Gas Injection Magnetron Sputtering (GIMS)—a technique that uses power pulses with a density characteristic of the PMS technique but coupled with the operation of pulsed valves supplying working gas [44,45]. Thanks to this, the generated plasma pulses have a chance to propagate in an environment with a lower concentration of gas molecules, which leads to limiting the dissipation of plasma particle energy [46,47];
• High Power Gas Injection Magnetron Sputtering (HiPGIMS)—pulsed technique using short power pulses with a density of around 10²–10³ W/cm², coupled with the operation of pulsed valves supplying working gas [37] and characterized by a high degree of excitation of plasma components [48,49,50].

All MS versions were implemented in a common vacuum chamber, the same pumping system, gas dosing apparatus, magnetron gun, and graphite target. Glow discharges of each SM process were diagnosed to evaluate the plasma excitation state. The deposited layers were examined both structurally and in terms of phase composition, with a particular focus on determining the sp³/sp² ratio. Additionally, optical properties were examined to estimate the electronic structure of the fabricated a-C layers.

2. Materials and Methods

2.1. Sputtering the Graphite Targets

Figure 1 presents a diagram of the three different MS techniques used in the experiment: PMS, GIMS, and HiPGIMS. All the process parameters are gathered and presented in

Table 1. Technological parameters of sputtering processes.

	PMS	GIMS	HiPGIMS
tAr (ms)	-	4	4
T (Hz)	-	2	2
D (%)	100	10	0.2
Jdmean (mA/cm2)	14	23	74
Pdmean (W/cm2)	10	18	6
pAr (Pa)	0.4 (steady)	10−3–10−1 (oscillating)	10−3–10−1 (oscillating)
tpulse (ms)	-	50	1
tproc (min)	10	120	120

. The 5 × 10⁻⁴ Pa base pressure in the chamber was established by a vacuum system comprising turbomolecular, Roots, and rotary pumps. The constant Ar atmosphere was created to perform the PMS process of a-C layers deposition. The mid-frequency DPS power supply (PS) (Dora Power Systems, Wroclaw, Poland) [51] (Posadowski et al., 2008) was used for PMS sputtering of a 5 cm diameter and 4 mm thick graphite target (purity 99.97%). The target was sputtered with an average current density of 14 mA/cm² and a power density of 10 W/cm².

The fast impulse gas valve (GV) system coupled with DPS PS was used to perform the GIMS process. The Ar was used as a sputtering gas and delivered through the outlet directed at the magnetron target. The GV operation parameters were adjusted by the electronic controller and are listed in Table 1: the opening time (t_Ar) and frequency (T). The controller was also coupled with PS, allowing for setting the power pulse ignition and turning it off. For this experiment, the 50 ms power pulses (t_pulse) were used to deposit a-C films.

The duty cycle D of GIMS was 10%. Jd_mean was calculated and equaled 23 mA/cm² while Pd_mean was 18 W/cm². The pulse manner of gas distribution, with parameters described above, resulted in periodic pressure fluctuations (p_Ar) ranging from 10⁻³ to 10⁻¹ Pa, in contrast to the PMS process’s pressure, which is fixed at 0.4 Pa.

The DPS high-power supply was used to perform the HIPGIMS process. The DPS HiPS was coupled with the GV system to create the same pressure oscillating conditions as the GIMS process: the opening time (t_Ar) and frequency (T). Under these conditions, the system generated 1 ms long plasma pulses with a Jd_mean of 74 mA/cm² and a Pd_mean of 6 W/cm². The D was 0.2%. The t_pulse was the only technological parameter different from the GIMS process, which was an effect of the electronic construction of the DPS HiPS.

The a-C layers were fabricated onto Si (100) wafers, which were mounted perpendicularly to the Z axis of the magnetron at a distance of 80 mm from the target, as a result of the (t_proc) deposition process. The wafers were cleaned by ultrasonic treatment in acetone for 15 min and then dried before being mounted in the vacuum chamber. The deposition process did not involve any intentional substrate biasing or heating. The deposition temperature did not exceed 60 °C.

2.2. Examination Methods

Plasma characterization was performed by analyzing the magnetron electric circuit’s current and voltage waveforms, as well as using optical emission spectroscopy (OES). The current and voltage waveforms were measured using a 1 A: 0.1 V Rogowski coil and a 1 kV: 1 V voltage probe (Dora Power Systems, Wroclaw, Poland), and recorded by a Rigol MSO5204 oscilloscope (Rigol Technologies Inc., Portland, OR, USA). All calculations were performed using OriginLab software (2019b ver. 9.6.5.169). OES investigations were executed using a Mechelle 900 optical spectrometer (Oxford Instruments Andor, Belfast, UK) with a mounted SensiCam Fs camera (300–1000 nm spectral range, relative spectral resolution ∆λ/λ = 900. The 100 ms (PMS and GIMS) and 10 ms (HiPGIMS) exposition frames were used to register the generated plasma’s OES spectra. Plasma spectra were registered during 100, 200, and 500 exposure frames for PMS, GIMS, and HiPGIMS, respectively. A high number of frames was crucial for achieving a satisfactory signal-to-noise ratio and for obtaining the most statistically representative plasma spectrum under the investigated conditions. Registration was triggered from a current rise measured by a Rogowski coil mounted in the magnetron cathode’s electric circuit. Exposition time was 100 (PMS and GIMS) or 10 ms (HiPGIMS) long. Spectra registration was performed by the optical collimator (Oxford Instruments Andor, Belfast, UK), which was positioned 80 mm from the cathode surface. The collimator was oriented parallel to the magnetron’s cathode surface. The intensity of registered OES spectra was further normalized considering various numbers of frames and exposure times.

The structure morphology of the 5 μm × 5 μm surface area of the a-C films was imaged using Atomic Force Microscopy (AFM) by an Innova device (Bruker Corp., Billerica, MA, USA). The investigation was conducted using the tapping mode. The analysis was performed using the Gwyddion software (ver. 2.61, GNU, Brno, Czech Republic). The surface roughness (S_a) has been calculated using Equation (1):

$$S_a= {\displaystyle \frac{1}{A}}{\iint }_A^{\infty }\left|z\left(x, y\right)\right|dxdy$$

(1)

where A and z(x, y) are the best-fitting mean plane and the height of the surface relative to A, respectively.

X-ray Photoelectron Spectroscopy (XPS) was used to examine the composition and sp³/sp² ratio of the a-C films. The film samples were investigated in an ultra-high vacuum of ≤2 × 10⁻¹⁰ mbar. A lamp with an Al K-α aluminum anode (1486.6 eV) (Prevac, Rogow, Poland) was used as a source of the excitation radiation for the spectrometer. The photoelectron energy was measured using a VG-Scienta R3000 analyzer (VG-Scienta, Uppsala, Sweden) with a ΔE = 0.1 eV step. The experimental data were calibrated with respect to the position of the C1s level and fitted to Gauss-Lorentz shapes using Casa XPS software (ver. 2.3.17, Casa Software Ltd., Teignmouth, UK).

The vibrational spectroscopy of the a-C layers was studied using a 532 nm VIS laser (2.33 eV) and a 266 nm UV laser (4.66 eV). The Ar⁺ laser and the Crylas FQCW266-50 (CryLaS GmbH, Berlin, Germany) diode-pumped continuous-wave solid-state laser were used as the VIS and UV excitation sources, respectively. The scattered light was detected by a JASCO NRS-5100 spectrometer (JASCO International Co., Ltd., Tokyo, Japan), operating in backscattering mode. The ~20 μm spots of film samples were investigated. The registration conditions were optimized based on the satisfactory signal-to-noise ratio. The Spectra Manager Micro Imaging Analysis v.2.3 (JASCO, Tokyo, Japan) and Curve Fitting v.1.9 (JASCO, Tokyo, Japan) processing software were used to analyze the registered spectra. The spectra were processed by interpolation and background subtraction, aligning levels at each side of the presented range. The curve fitting operation determined the position and a full width at half maximum FWHM of the peaks D and G. The sp³ bond content was calculated based on the G peak dispersion rate [52]. The sp³/sp² ratio, as the G peak dispersion is expressed by Equations (2) and (3):

$${sp}_{content}^3= -0.07+2.5 \times Disp\left(G\right) \pm 0.06$$

(2)

$$Disp\left(G\right)=\left|{\displaystyle \frac{Pos\left(G\right)@{\lambda }_1-Pos\left(G\right)@{\lambda }_2}{{\lambda }_1-{\lambda }_2}}\right|$$

(3)

where $Pos\left(G\right)@{\lambda }_1$ and $Pos\left(G\right)@{\lambda }_2$ are the G peak positions at ${\lambda }_1$ and ${\lambda }_2$ irradiation, respectively. ${\lambda }_1$ and ${\lambda }_2$ are the laser wavelength.

The ellipsometric azimuths ψ and Δ have been measured from 190 nm (6.5 eV) to 2000 nm (0.6 eV) for 65°, 70°, and 75° incidence. The Spectroscopic ellipsometry (SE) measurements were performed using the V-VASE device from J.A.Woollam Co., Ltd. (Lincoln, NE, USA). The ψ and Δ ellipsometric azimuths are defined as Equation (4) [53,54]:

$$\tilde{\rho }= \mathit{tan}\psi expi∆$$

(4)

where $\tilde{\rho }$ is the ratio of the amplitude reflection of the electromagnetic radiation for $p-$ and $s-$ polarization, and $i$ is the imaginary unit.

The WASE32 software (version 3.774) was used for further analysis. The refractive index $n$ and extinction coefficient $k$ optical constants were determined this way. Based on the $k$, the absorption coefficient $\alpha$ was calculated using Equation (5):

$$\alpha = {\displaystyle \frac{4\pi k}{\lambda }}$$

(5)

where $\lambda$ is the wavelength of the incident light. The energy band gap of the layers was determined using the Tauc method [55,56]. This is based on the assumption that the energy-dependent absorption coefficient $\alpha$ can be expressed by Equation (6):

$${(\alpha \times h\nu )}^{1/m}=B(h\nu - E_g)$$

(6)

where $h\nu$ is the photon energy, and $B$ is a constant. The $m$ factor depends on the nature of the electron transition and corresponds to 0.5, 1.5, 2, or 3 for the direct allowed, direct forbidden, indirect allowed, and indirect forbidden transition band gaps, respectively. Our experiment obtained the most linear curves for the indirect allowed transition ($m$ = 2). A linear fit procedure was used to extrapolate the linear parts of the plots. The bandwidths, E_g were read from the extrapolated lines that intersected the X-axis.

3. Results and Discussion

Figure 2 presents the current and voltage waveforms of the glow discharge generated in every process investigated in the experiment. The characteristics indicate differences in the nature of the discharge. The PMS is characterized by a relatively low 0.2 A average current, 14 mA/cm² average current density (Jd_mean), and a 10 W/cm² power density (Pd_mean). Due to the characteristics of graphite, i.e., low sputtering yield, high ionization energy of carbon atoms, and graphite structure, it is impossible to generate a discharge characterized by high power density without using an additional approach to assist sputtering and ionization. In the case of the GIMS process, the maximum current peak was registered as ~10 A. We have also noticed a significant improvement in the density of Jd_mean and Pd_mean, at 23 mA/cm² and 18 W/cm², respectively. Since the plasma is initiated by the injection of the working gas portion into the high-vacuum interelectrode space, it is possible to obtain high current values (higher ionization than in PMS) by reducing the kinetic energy dissipation of excited plasma species. The high current peak is ~1.5 ms long, and after this period, the plasma appears to become thermalized, resulting in a current level comparable to that observed in the PMS method. The full length of the plasma pulse was measured to be 50 ms. The HiPGIMS process demonstrates an approach to raise the Jd_mean more than GIMS. The HiPGIMS combines the advantages of working gas injections with the use of a high-power source. The effect was ~30 A, with a current peak lasting 0.3 ms. The Jd_mean increased to 74 mA/cm²_, and the Pd_mea was calculated as 6 W/cm². The relatively low Pd_mean compared to the other processes is due to the short HiPGIMS pulse duration (1ms HiPGIMS vs. 50 ms GIMS). Since the power density was averaged over the 500 ms period to calculate the Pd_mean, the Pd_mean does not adequately reflect the significance of the HiPGIMS process. To present the key difference between the power peak of GIMS and HiPGIMS, the power density was calculated based solely on the power peak. The power density at its peak was 254 W/cm² for GIMS and 1274 W/cm² for HiPGIMS, respectively.

These significant differences in plasma pulses should be reflected in the composition of OES spectra. In Figure 3, the results of OES investigations are presented. Figure 3a presents a comparison of the individual spectral characteristics for each examined process. The detailed information on the assigned spectral lines is presented in

Table 2. Spectral lines characteristics assigned to the spectra measured in the experiment.

Observed Wavelength (nm)	Species	Lower Level	Upper Level	Intensity (arb. u.)
				PMS	GIMS	HiPGIMS
403.7	C I	2s22p3s	2s22p5p	430	340	in noise
405.1	Ar II	3s23p4(1D)4s	3s23p4(1D)4p	390	301	in noise
406.7	C I	2s22p3s	2s22p5p	625	671	778
410.2	Ar II	3s23p4(3P)4p	3s23p4(3P)5s	233	515	in noise
412.8	Ar II	3s23p4(1D)4s	3s23p4(1D)4p	760	924	in noise
415.5	C I	2s2p3	2s22p7p	1597	963	in noise
418.3	Ar I	3s23p5(2P°3/2)4s	3s23p5(2P°3/2)5p	878	401	in noise
419.5	Ar I	3s23p5(2P°3/2)4s	3s23p5(2P°3/2)5p	2769	2153	in noise
422.6	C I	2s2p3	2s22p(2P°1/2)6f	329	475	in noise
425.3	Ar I	3s23p5(2P°1/2)4s	3s23p5(2P°1/2)5p	1929	1569	778
427.4	C I	2s22p3s	2s22p5p	1929	2192	856
429.6	Ar I	3s23p5(2P°3/2)4s	3s23p5(2P°3/2)5p	817	671	in noise
432.9	Ar II	3s23p4(3P)4s	3s23p4(3P)4p	1031	946	1401
434.4	C I	2s2p3	2s22p6p	1092	2602	5145
436.6	C II	2s2p(3P°)3d	2s2p(3P°)4f	664	1063	1131
437.7	C II	3s23p4(3P)4s	3s23p4(3P)4p	369	671	992
439.7	Ar II	3s23p4(3P)4p	3s23p4(3P)5s	294	906	1872
441.9	Ar II	3s23p4(3P)4d	3s23p4(3P)4p	486	867	1480
446.7	C I	2s2p3	2s22p(2P°3/2)5f	329	223	in noise
447.7	C I	2s2p3	2s22p(2P°3/2)5f	468	458	in noise
450.5	Ar I	3s23p5(2P°1/2)4s	3s23p5(2P°3/2)5p	604	401	in noise
454.1	Ar II	3s23p4(3P)4s	3s23p4(3P)4p	1149	1433	682
457.3	Ar II	3s23p4(3P)4s	3s23p4(3P)4p	525	924	896
458.5	Ar II	3s23p4(1D)4s	3s23p4(1D)4p	1188	1512	1092
460.3	Ar II	3s23p4(1D)3d	3s23p4(1D)4p	1850	2327	1676
465.6	Ar II	3s23p4(3P)4s	3s23p4(3P)4p	1266	1765	1149
471.9	Ar II	3s23p4(3P)4s	3s23p4(3P)4p	1209	1704	1305
473.2	C I	2s2p3	2s22p5p	739	1490	2338
475.7	Ar I	3s23p5(2P°3/2)4p	3s23p5(2P°3/2)8d	2338	2894	1401
480.2	C I	2s2p3	2s22p5p	1131	2349	4286
484.3	C I	2s22p3p	2s22p(2P°1/2)16d	586	906	1519
487.5	C I	2s22p3s	2s22p4p	2103	3090	3001
493.1	C I	2s22p3s	2s22p4p	369	536	817
496.1	Ar II	3s23p4(3P)4s	3s23p4(3P)4p	995	1024	935
501.1	C I	2s22p3p	2s22p(2P°1/2)10d	468	789	1052
505.4	C I	2s22p3p	2s22p(2P°3/2)7d	447	732	1092

.

The intensities of the spectra were normalized according to the various exposure times and the number of exposure frames. Although the intensities of the optical emission lines do not directly indicate the species population in the plasma, the trend of line intensity evolution within each species group can provide valuable information on the plasma content as the process version changes [57]. Trends are observed for a specific group of plasma constituents. In Figure 3b), we present the chosen normalized intensities of the lines of the C I species. The population of sputtered and excited C I carbon atoms seems to expand as the generated plasma is expected to be more ionized. We have observed the same trend for carbon ions (C II-Figure 3c). The trend is opposite in the case of gaseous-originated plasma constituents, excited Ar I (Figure 3d) species, and Ar II ions (Figure 3e). It demonstrates the positive role of utilizing the pulse action of a gas dosing valve, which limits the population of Ar excited species to the amount required to generate the discharge, and achieves the carbon vapors through sputtering. An excess population of excited gas plasma species (represented by a process with a constant concentration of working gas–PMS) does not yield positive results in the number of ionized plasma constituents. Interestingly, the lowest intensity of ionized Ar II was observed for HiPGIMS, while the highest was for GIMS. It can be explained by the reversed movement of ions generated by HiPGIMS due to a strong electric field in the magnetic trap region, which is oriented towards the target [58,59]. It is a common phenomenon for high-power MS techniques, and its effect creates a potential barrier that traps ions near the target, preventing them from reaching the substrate/OES collimator zone.

Figure 4 shows the morphology of the a-C films’ surface. There are not many differences between the films observed. All films exhibit dense and smooth surface characteristics for films deposited with a significant number of ions. The calculated roughness $S_a$ was about ~0.5 nm.

To evaluate the sp³/sp² ratio, the a-C layers were examined by Raman spectroscopy. Monocrystal graphite exhibits a ~1580 cm⁻¹ Raman active band E_2g, labeled G (Graphite). The G peak affects the stretching vibration of the sp² sites. With the graphite amorphization, an additional mode at ~1350 cm⁻¹, A1g, appears, known as the D peak (Disorder). The D mode is attributed to the breathing modes of the sp² sites in rings. According to the Ferrari diagram of 3-stage carbon amorphization [60], the sp³ bond is expected when the in-plane correlation length L_s is below 2 nm [61]. Correctly determining the characteristics of the G and D peaks (positions, intensity, and FWHM) from the a-C Raman spectra is reported as the method of evaluating the sp³ bond content. Figure 5 presents the Raman spectra of a-C layers collected under VIS and UV laser irradiation, as well as the effects of G and D mode fitting. Evaluated parameters of fitted peaks are shown in

Table 3. Parameters of fitted G and D peaks of a-C layers Raman spectra.

		G peak			D peak
		Position (cm−1)	FWHM (cm−1)	Intensity (arb. u.)	Position (cm−1)	FWHM (cm−1)	Intensity (arb. u.)
266 nm	PMS	1572.8	162.1	288.0	1381.9	342.9	53.7
	GIMS	1580.4	154.9	594.9	1383.1	300.4	194.7
	HiPGIMS	1598.3	100.4	220.5	1371.2	303.2	50.6
532 nm	PMS	1538.5	195.5	347.2	1373.6	273.7	176.7
	GIMS	1539.4	188.2	488.0	1366.7	296.9	263.5
	HiPGIMS	1537.3	162.2	1026.8	1366.4	320.8	481.1

. The ~0.5 ratio of D peak to G peak intensity, the 160–200 cm⁻¹ FWHM of G peak, and ~1538 cm⁻¹ of G peak position suggest that 10–50% of sp³ content is expected [60,61,62]. To quantitatively evaluate the sp³/sp² ratio, a method based on the dispersion of the G peak was used [52,63]. The ratio was calculated using Equation (2).

Figure 6 shows the calculated sp³/sp² ratio from Raman spectroscopy and the XPS method. Despite the differences in the calculated ratio by both methods, a noticeable common trend of increasing sp³ is observed. The sp³ was evaluated as 25 ± 6, 31 ± 6, and 50 ± 6% for PMS, GIMS, and HiPGIMS, respectively (Raman spectroscopy approach), and 17 ± 2, 20 ± 3, and 39 ± 6% (XPS approach). In summary, more sp³ was obtained in high-current discharges. This is because the increased current results in a higher plasma density. This, in turn, provides more energetic ions that bombard the growing layer. These energetic ions can effectively promote the formation of sp³ bonds [37,64,65].

The differences in sp³ content between the experimental methods used likely stem from obstacles to accurately evaluating oxygen influence on Raman spectra of a-C films. The XPS method seems to be more precise than Raman spectroscopy. The second explanation is the fact that the surface region of a-C layers is enriched in sp² bonds [66]. Therefore, an XPS study, which examines only the material at its surface (a few nanometers), always shows an increased sp² content compared to methods that are sensitive to material features detected at a greater depth. Raman spectroscopy is an example of such a method. Films fabricated in the experiment were <50 nm thin, so Raman spectroscopy provided the spectroscopic information from the entire depth of the film, in contrast to XPS. Figure 7 shows the C1s and O1s core levels of a-C layers deposited by MS techniques. Additionally, a small amount of Ar was detected in the composition of the PMS and GIMS layers, as represented by the Ar2p orbital. The Si was observed in the chemical composition of HiPGIMS a-C layer (represented by Si2p core level). The detailed chemical composition of the deposited layers is shown in

Table 4. Chemical compositions of a-C layers synthesized by various MS techniques.

	C (sp2)	C (sp3)	C (C–O)	C (C=O)	O-C	OH	O=C	SiO2	Ar	%C	%O	%Si	%Ar
PMS	63.72	10.68	14.19	4.34	3.66	0.31	2.74	-	0.36	92.93	6.71	0	0.36
GIMS	62.13	12.22	14.44	3.54	4.51	0.62	2.03	-	0.51	92.33	7.16	0	0.51
HiPGIMS	45.09	17.77	19.87	3.49	9.88	0.79	0.82	2.29	-	86.22	11.49	2.29	0

. The main components of the C1s core level are carbon peaks originating from sp²- and sp³-hybridized bonds, at 284.6 and 285.5 eV, respectively [67,68]. The C1s level incorporates additional C–O bonds on the surface, and carbonyl bonds (C=O) [67]. These levels were determined at binding energies of 286.3 and 288.8 eV, respectively. The O1s orbital incorporates three elementary components: O–C component at 532.8 eV, carbonyl O=C component at 531.5 eV, and O–H component at 534.6 eV [69]. The Ar2p orbital was present in the XPS spectra of PMS and GIMS layers. The measured argon 2p level consists of the Ar2p_3/2 sublevel at 241.8 eV and Ar2p_1/2 at 244.2 eV [70]. The Ar contamination of the layer structures is reported and studied in the literature. The total amount of working gas trapped in the layer structure may depend on the ion energy and flux to the substrate, which can be controlled by the shape, amplitude, and duration of the voltage and current pulses [71,72]. In the case of a-C layers deposited in our experiment, Ar was detected during the process, characterized by an increased content of Ar I and Ar II plasma species (see Figure 3d,e).

In the case of the HiPGIMS process, where the population of Ar species was evaluated as relatively low compared to other MS processes, no Ar2p core level was measured. The Si2p orbital was detected in the HiPGIMS layer with a SiO₂ component at 103.0 eV [73]. In our opinion, the presence of Si atoms is the effect of partial sputtering and recondensation of the substrate surface at the initial stage of carbon layer deposition. The sputtering process can be efficient in the HiPGIMS process since it is recognized as the most energetic in terms of the degree of ionization of the plasma.

Spectroscopic ellipsometry was used to examine the optical properties of a-C layers and to determine their thickness. The thickness was determined to be 16.4 ± 0.1, 35.4 ± 0.5, and 34.3 ± 0.4 nm for PMS, GIMS, and HiPGIMS, respectively. Calculated deposition rate was: 16 Å/min (PMS), 2.95 Å/min (GIMS-10% duty cycle), and 2.86 Å/min (HiPGIMS-0.2% duty cycle). This is an expected trend, based on the increased effects of sputtering, measured as current density and the population of carbon species, as registered by OES in GIMS and HiPGIMS techniques. The optical constants (refractive index—n and the extinction coefficient—k), absorption coefficient α, and Tauc plots ${(\alpha \mathsf{\bullet }h\nu )}^{1/2}$ are shown in Figure 8. PMS and the other processes have a difference in the shape of n. For UV, the n value takes very low values, ranging from 0.5 to 1.0. With the λ increase in the VIS range, the n significantly increases to 1.8 and remains at that level to the IR spectral range. This specific n shape is characteristic of low-density carbon materials, especially those rich with sp² clusters [74,75]. The k of a-C layer synthesized by PMS is also characteristic of sp² carbon, exhibiting a strong maximum in the visible range and low values in the UV and IR regions [74]. The n and k plots for layers fabricated using the other process are similar for dense carbon materials, contributing a significant content of the sp³ phase [74]. This conclusion is supported by the previously presented results, which indicate that the GIMS and HiPGIMS processes favor the synthesis of the sp³ phase, in contrast to the PMS.

The α plot exhibits similar features to k. The plot of the PMS process exhibits characteristic features for low-density carbon, where absorption rises in the IR region (~1 eV) and the UV region (~5.5 eV). The absorption region of a-C layers deposited by other methods is wider and characteristic of carbon materials, exhibiting an increased content in the sp³ phase [74].

The Tauc plots show differences in the electron structure of deposited a-C layers. The calculated bandgap energy (Eg) is presented as 1.00 ± 0.03, 0.75 ± 0.02, and 1.65 ± 0.03 eV for PMS, GIMS, and HiPGIMS, respectively. The highest value of E_g presents the layer that contributes the highest amount of sp³–HiPGIMS. None of the layers presents a sharp edge of absorption. It means that there is a significant influence of $\pi \to {\pi }^{\ast }$ transition characteristic for the sp² phase of carbon.

In the context of the analyzed magnetron sputtering techniques, a clear relationship exists between the measured plasma parameters and the obtained sp³/sp² hybridization ratio. The PMS technique was characterized by the lowest average current density (14 mA/cm²) and average power density (10 W/cm²), resulting in the generation of a relatively low ionized plasma. Neither the subplantation mechanism nor homogeneous nucleation of metastable phases such as sp³-hybridized carbon is favored by such a plasma environment. As a result, the carbon film with the lowest sp³ was deposited. In the case of GIMS, an increase in the average current density to 23 mA/cm² and average power density to 18 W/cm² was observed, with a peak power of 254 W/cm², resulting in the creation of a more energetic plasma environment. An increase in ions and a decrease in the working gas’s unionized species were observed. The only difference from PMS was the pulsating mechanism of gas dosing, which created favorable conditions for preserving kinetic energy by the plasma species. This resulted in a higher sp³ content in the deposited film. A sp³ increase in relation to PMS is not significant: from 25 ± 6 to 31 ± 6% (according to Raman spectroscopy) and from 17 ± 2 to 20 ± 3% (according to XPS). Despite a 1.5 ms strong current peak of the GIMS discharge, a significant part of it (48.5 ms) consists of a low-current part, resembling the PMS process. This results in the plasma composition becoming potentially complex over time, and the material of the deposited film can also be complex, exhibiting characteristics of both carbon phases formed in a low-ionization environment (sp²) and those formed in a high-ionization environment (sp³). The HiPGIMS showed the highest average current density (74 mA/cm²) and a significantly higher peak power density (1274 W/cm²), which was associated with the highest degree of plasma ionization. The most intense spectral lines of ions of sputtered atoms. These more energetic plasma conditions in HiPGIMS resulted in the highest sp³ content in the films, as determined by Raman (50 ± 6%) and XPS (39 ± 6%) measurements. Similar to GIMS, the current characteristic consists of high- and low-current parts, resulting in the formation of both sp³ and sp² carbon phases. In the case of HiPGIMS, the peak power density is significantly higher, and the low-current part is similar in length to the high-current part; therefore, the significant content of sp³ was measured in this film. In summary, increased current and power density, and consequently, higher plasma density and more energetic ion bombardment, effectively promote the formation of sp³ bonds, which are crucial for controlling the properties of amorphous carbon layers.

4. Conclusions

This study successfully investigated the synthesis and characterization of a-C layers deposited using three distinct MS techniques: PMS, GIMS, and HiPGIMS. The primary objective was to elucidate the influence of these deposition methods on the critical sp³/sp² hybridization ratio, which dictates the material’s properties.

Plasma diagnostics, particularly OES, revealed significant differences in discharge characteristics, with HiPGIMS exhibiting the highest current density and plasma ionization. Structural and compositional analyses, employing Raman Spectroscopy and XPS, consistently demonstrated a clear trend: the sp³ content progressively increased from PMS to GIMS and reached its maximum in HiPGIMS layers, achieving values of up to 50 ± 6% (Raman) and 39 ± 6% (XPS). This enhanced sp³ bonding is directly attributed to the higher plasma density and more energetic ion bombardment inherent in the HiPGIMS process. Further corroboration came from ellipsometric spectroscopy, which showed that HiPGIMS-deposited layers possessed the widest bandgap, which indicates higher sp³ content. It appears that there is still some margin for increasing the sp³ content, which would provide greater control over the current pulse, particularly the elimination of the low-current component that is most likely responsible for the deposition of the sp² fraction. Alternatively, specific effects could be achieved by using pulsed biasing of the substrate, adapted to the dynamically changing temporal structure of the generated plasma. The proper application of both the first and second methods requires further research, especially in deepening the characterization of the temporal-spatial nature of plasma pulses in GIMS and HiPGIMS methods.

In summary, this research highlights the superior capability of advanced MS techniques, particularly HiPGIMS, in precisely controlling the sp³/sp² ratio within amorphous carbon layers. The ability to tune this fundamental structural parameter offers a robust pathway for tailoring the mechanical, electrical, and optical properties of a-C films, thereby expanding their potential for diverse technological applications.