Structural and Thermal Stability of TiN- and SiC-Based Multilayer Diffusion Barriers for Copper–Silicon Interfaces

Highlights

What are the main findings?

• The SiC layer adjacent to Cu blocks Cu diffusion and preserves TiN structural integrity.
• TiN decomposes above 600 °C when directly exposed to Cu, forming Cu₃Si and Ti silicide phases.
• TiSi₂ does not form spontaneously at the Si/TiN interface without a metallic Ti interlayer.

What are the implications of the main findings?

• The optimal stack sequence is Si/TiN/SiC/Cu for thermal stability and low sheet resistance.
• TiN and SiC bilayers enable Cu metallization in high-efficiency solar-cell architectures.
• Layer sequencing critically affects interfacial reactions and device reliability.

Abstract

In this study, the diffusion barrier performance of TiN and SiC layers was investigated in Si/TiN/Cu, Si/TiN/SiC/Cu, and Si/SiC/TiN/Cu multilayer structures to address copper diffusion issues at silicon interfaces in microelectronics. Samples were annealed in argon at 500–800 °C for 30 min, and diffusion behavior was analyzed using X-ray diffraction (XRD) and sheet resistance measurements. The Cu₃Si phase formed at 600 °C in the Si/TiN/Cu system, while no Cu₃Si appeared in the Si/SiC/TiN/Cu system up to 700 °C, indicating improved stability. Complete copper diffusion occurred in all systems at 800 °C. Sheet resistance measurements corroborated the XRD findings, demonstrating that multilayer structures incorporating TiN and SiC significantly enhance thermal stability and suppress copper diffusion. Comparison of Si/SiC/TiN/Cu and Si/TiN/SiC/Cu stacks annealed at 700 °C revealed that the stability of TiN depends on layer sequence, with SiC effectively blocking Cu migration into TiN when placed adjacent to Cu. Structural and morphological properties of TiN films were also examined, confirming their suitability as diffusion barriers. Additionally, the feasibility of forming a low-resistivity TiSi₂ layer through a single annealing step to create a TiSi₂/TiN system was explored, highlighting potential applications in advanced device integration.

1. Introduction

Photovoltaic (PV) technologies are rapidly evolving as the demand for higher power-conversion efficiency continues to grow. Modern architectures such as tunnel oxide passivated contact (TOPCon) [1,2], heterojunction (SHJ) [3,4], and Si/perovskite tandem devices [5,6] increasingly rely on contact systems with reduced resistive losses, robust thermal stability, and minimized recombination at metal–semiconductor interfaces. As the industry shifts toward thinner passivating layers, narrower contact openings, and low-temperature processing, the development of stable, low-resistance metal–silicon contacts has become one of the central challenges in next-generation PV manufacturing.

Copper (Cu) is considered a highly promising metallization material due to its low resistivity, low cost, and compatibility with large-scale processing [7]. Transitioning from Ag-based to Cu-based metallization is one of the key industrial trends aimed at reducing the cost per watt and enabling scalable electrode formation for high-efficiency solar cells. However, Cu exhibits extremely high diffusivity in silicon (Si) and silicon dioxide (SiO₂), even at moderate process temperatures [8,9]. This diffusion creates deep-level defects in the Si bandgap, accelerates minority-carrier recombination, and significantly degrades device performance. Elevated annealing temperatures further intensify Cu transport and promote the formation of copper-silicide phases (e.g., Cu₃Si), which sharply increase contact resistance and reduce operational stability of solar cells [10,11]. Given these detrimental effects, preventing Cu penetration into the Si substrate becomes a central requirement for any viable metallization scheme. Therefore, integrating thin and thermally stable diffusion barrier layers between Cu and Si is essential for enabling Cu metallization in advanced PV devices.

Titanium nitride (TiN) is one of the promising materials for this purpose [9,11]. Stoichiometric TiN films exhibit low electrical resistivity, high thermal stability, and excellent Cu diffusion-blocking capabilities. In the context of modern PV manufacturing, TiN is increasingly viewed as a diffusion barrier material compatible both with high-temperature annealing steps (used for silicide formation or dopant activation) and with emerging low-temperature metallization schemes relevant for SHJ and tandem structures. Beyond its role as a diffusion barrier, TiN has recently gained attention in plasmonic photocatalysis, where its tunable optical properties enable enhanced light–matter interactions for solar-driven chemical reactions [12]. Additionally, TiN is being explored as a selective absorber material in thermophotovoltaic devices, leveraging its high-temperature stability and spectral selectivity to improve energy conversion efficiency [13,14]. Moreover, TiN is increasingly explored as an electron-selective contact for Si solar cells, offering a suitable work function [15,16]. Thus, TiN exhibits a strong potential for integration into advanced PV architectures and energy conversion technologies.

However, TiN performance strongly depends on structural changes that occur during annealing. At elevated temperatures, TiN can undergo phase transformations into nitrogen-deficient compounds, while simultaneous interdiffusion with silicon may lead to the formation of silicide phases. These transformations alter the microstructure and crystallographic orientation of the barrier layer, which, in turn, affects its electrical resistivity. Generally, the decomposition of stoichiometric TiN and the emergence of secondary phases increase resistivity by disrupting conduction pathways and introducing scattering centers. Conversely, the stabilization of preferred TiN orientation, such as TiN(111), can help preserve low resistivity under certain conditions. Thus, understanding the interplay between TiN phase evolution and resistivity during annealing is essential for optimizing its role in reliable, low-resistance metallization schemes.

In addition to TiN, silicon carbide (SiC) plays an important role as a single-layer diffusion barrier. The Si/SiC/Cu configuration provides a useful reference case for evaluating the intrinsic blocking capability of SiC without the presence of TiN. Owing to its high chemical stability, suitable electrical conductivity, and advantageous passivation properties [17,18], SiC can suppress copper penetration into silicon while simultaneously contributing to interface protection. Incorporating this system into the analysis makes it possible to separate the individual contributions of SiC and TiN and to clarify how each material operates both on its own and within the multilayer stack. This comparison is essential for understanding optimal layer sequencing and for establishing design guidelines for thermally stable Cu metallization schemes in advanced PV architectures.

The primary research problem addressed in this work is the systematic comparison of thermal stability and Cu diffusion-blocking performance between different TiN/SiC layer sequences. Specifically, this study investigates whether placing SiC adjacent to Cu (Si/TiN/SiC/Cu configuration) offers superior protection compared to placing TiN adjacent to Cu (Si/SiC/TiN/Cu configuration). This question is critical because Cu atoms exhibit different diffusion kinetics through TiN versus SiC, and the interfacial reactions at TiN/Cu versus SiC/Cu interfaces differ substantially. By establishing the optimal barrier configuration, this study aims to provide design guidelines for thermally robust Cu-based contacts in TOPCon, SHJ, and tandem solar cells.

Given the increasing relevance of Cu-compatible metallization and thermally stable barrier systems for TOPCon, SHJ, and tandem solar cells, this work focuses on the synthesis and detailed characterization of TiN and SiC/TiN diffusion-barrier layers and their behavior during annealing. Special attention is devoted to X-ray analytical methods to probe the evolution of TiN, SiC, and Ti–Si interfacial phases under varying thermal treatments. Through this approach, the study determines how thermal treatments influence the microstructure, phase stability, and electrical performance of the barrier layers in different configurations.

2. Materials and Methods

Czochralski monocrystalline p-type (100) Si wafers with specific electrical resistivity of 10–25 Ω·cm, a diameter of 150 mm, and a thickness of 520 µm were used as substrates. Prior to film deposition, Si substrates underwent a three-step RCA cleaning process [19]. The method includes treatment in the following solutions: NH₄OH:H₂O₂:H₂O (1:1:5) at 80 °C for 10 min, HF for 20 s, and HCl:H₂O₂:H₂O (1:1:6) at 75–80 °C for 10 min. Each step was followed by rinsing in flowing deionized water and drying for 15 min.

TiN_x layers were deposited by reactive magnetron sputtering using a high-purity 99.9% Ti target. The sputtering parameters were adopted from the previously optimized conditions reported in [20] and were set as follows: 1200 W target power, 0.8 Pa working pressure, 30/1 N₂/Ar gas ratio with a nitrogen flow of 30 sccm, and 180 °C substrate temperature. For the structural characterization, the films were deposited for 600 s. For the barrier properties evaluation, 144 s and 288 s deposition times were used, yielding TiN_x layer thicknesses of approximately 50 nm and 100 nm, respectively.

SiC films were deposited by RF magnetron sputtering using a stoichiometric SiC target at a 200 W power and 0.4 Pa working pressure in an Ar atmosphere. A deposition time of 1500 s was applied to obtain a film thickness of approximately 50 nm. The sputtering conditions used in this study were based on the optimized parameters for high-quality SiC film growth reported in [21,22].

To examine the barrier properties against Cu diffusion, Cu layers were deposited by DC magnetron sputtering using a high-purity 99.5% Cu target. The sputtering process was carried out at 1200 W power and 0.9 Pa working pressure in an Ar atmosphere, with deposition time ranging from 12 s to 36 s.

Figure 1 illustrates the schematic configurations of the three multilayer structures designed to evaluate copper diffusion behavior and barrier performance. In Figure 1a, a single TiN layer is directly deposited between silicon and copper, serving as the primary diffusion barrier in Si/TiN/Cu. This simple stack highlights the baseline performance of TiN against copper migration. In Figure 1b, an additional SiC layer is introduced between the silicon substrate and TiN in Si/SiC/TiN/Cu. This arrangement allows SiC to act as a complementary barrier, enhancing the overall thermal stability and delaying copper silicide formation. In Figure 1c, the SiC layer is positioned between TiN and copper, creating a different sequence where SiC directly interfaces with Cu in Si/TiN/SiC/Cu. This configuration is particularly important because the SiC layer can effectively block copper atoms from penetrating into TiN, thereby preserving TiN’s structural integrity at elevated temperatures. Together, these schematics emphasize how the placement of SiC relative to TiN and Cu critically influences diffusion-blocking efficiency, with the Si/SiC/TiN/Cu and Si/TiN/SiC/Cu stacks offering improved stability compared to the simpler Si/TiN/Cu system. Figure 1 provides a visual schematic for understanding the role of layer sequence in optimizing barrier performance for advanced microelectronic applications.

To investigate interfacial reactions and phase formation, the multilayer stacks were subjected to rapid thermal annealing or tube furnace annealing in an Ar atmosphere. Annealing was performed over a temperature range of 500–800 °C for 30 min.

The structural characteristics of the films were investigated by X-ray diffraction (XRD) on the CompleXray C6 system (The P.N. Lebedev Physical Institute, Moscow, Russia [23]) and by X-ray photographic method using Debye cameras (HUBER Diffraktionstechnik GmbH & Co. KG, Rimsting, Germany). Diffraction rings were recorded on photographic film using a URS-01 installation (Odessa Factory of X-ray Equipment, Odessa, Ukraine, 1980). Intensity of diffraction lines was measured using a microdensitometer MD-100 (Carl Zeiss, Jena, Germany). Film density and thickness were measured by X-ray reflectometry (XRR) method [24] using CuK_α (0.154 nm) spectral lines on the CompleXray C6 system [23]. The surface microstructure was characterized by atomic force microscopy (AFM) using a JSPM-5200 system (JEOL, Tokyo, Japan) operated in tapping mode. Sheet resistance of the films was me asured using a four-point probe system at room temperature using a Keithley 2400 SMU Source Measure Unit (Keithley Instruments, Cleveland, OH, USA).

3. Results and Discussion

3.1. Synthesis of the TiN_x Diffusion Barrier

In advanced PV contact systems, multilayer stacks such as Si/TiSi₂/TiN/Cu are considered to be highly attractive. A key challenge is understanding the formation of Ti–Si phases (TiSi₂, Ti₅Si₃, Ti₅Si₄, etc.) during rapid or prolonged thermal treatments, since these phases largely determine the contact resistivity and stability [25,26]. A particularly relevant question concerns the feasibility of depositing TiN directly onto Si, forming a Si/TiN_x interface and subsequently annealing at 700–800 °C without an initial Ti layer or intentional silicide formation [14,27]. Whether such processing conditions lead to the spontaneous formation of conductive Ti–Si phases, or whether TiN can maintain its integrity as a diffusion barrier under these temperatures, remains insufficiently understood. This uncertainty extends to more complex barrier schemes, such as SiC_x/TiN bilayers, where the combined thermal stability, interlayer reactions, and Cu diffusion-blocking behavior have not yet been systematically clarified [26,28].

Thus, to ensure the formation of an effective TiN-based diffusion barrier, the presence of a TiSi₂ ohmic interlayer at the TiN/substrate interface is essential. Therefore, this section investigates the potential of spontaneous formation of a TiSi₂ layer at the Si/TiN interface during TiN deposition and subsequent annealing. Particular attention is given to assessing whether such interfacial reactions can occur without the intentional introduction of a metallic Ti layer during the sputtering process, which would significantly simplify the fabrication route and reduce process complexity.

To determine the thickness and density of the deposited TiN films, XRR measurement was employed. The measurements were carried out before and after annealing at 800 °C in order to assess structural changes, densification effects, and possible interfacial reactions induced by high-temperature treatment. The corresponding experimental results are presented in Figure 2.

Figure 2 presents the XRR curves of TiN films on Si before and after annealing. In the as-deposited state, the film exhibits clear Kiessig fringes, with periodic oscillations in the XRR curve extending to approximately 2.5° in 2θ. These fringes indicate a smooth, uniform film with well-defined thickness, while the sharp drop in intensity at 0.639° marks the critical angle of total external reflection. The regular periodicity and high amplitude of the fringes confirm excellent interface quality between the TiN layer and the Si substrate. After annealing at 800 °C in Ar for 30 min, the critical angle shifts to 0.556°, indicating a reduction in film density. The Kiessig fringes are strongly suppressed and scarcely visible, consistent with increased surface roughness and interfacial degradation caused by thermal decomposition and agglomeration. Insets show the same data plotted on linear intensity scales. In Figure 2 the X-axis represents the 2θ angle in degrees, while the Y-axis corresponds to X-ray intensity in arbitrary units, displayed on logarithmic scales in the main plots and linear scales in the insets.

Analysis of the XRR curves shows that the as-deposited TiN films exhibit a critical angle of total external reflection of 0.639° in Table S1. According to density calculations, this value corresponds to approximately 5.131 g/cm³, suggesting the formation of a Ti₆₀N₄₀ (Ti₃N₂) composition with an excess of titanium atoms relative to stoichiometric TiN. The film thickness extracted from the periodicity of the Kiessig fringes was estimated to be about 244 nm in Table S2.

The results of measuring the XRR spectra of TiN films after annealing at 800 °C in Ar are shown in Figure 2b. A pronounced reduction in the TiN film density is evident from the XRR data after annealing. The post-annealing critical angle, 2θ_s = 0.556°, corresponds to a calculated density of 3.89 g/cm³ in Table S3, substantially lower than both metallic titanium (4.54 g/cm³) and stoichiometric TiN (5.44 g/cm³). Such a decrease indicates significant structural and compositional modifications induced by the 800 °C thermal treatment. This density reduction is consistent with the possible formation of lower-density Ti–Si phases such as TiSi (4.03 g/cm³), Ti₅Si₄ (4.26 g/cm³), and Ti₅Si₃ (4.32 g/cm³), which can arise as a result of interfacial reactions between TiN and Si during high-temperature annealing. In addition, variations in surface morphology and topography make a significant contribution to the apparent density reduction. AFM characterization in Figure S1 revealed that the initially uniform nanocrystalline surface of TiN films transforms during annealing at 700 °C and 750 °C into large surface agglomerates measuring 300–400 nm, comparable to the film thickness. These agglomerates gradually decrease in size to about 50 nm after 700–800 °C annealing. As a result, the film exhibits a pyramidal microstructure with pronounced surface deformation and distinct voids between adjacent features. These voided regions effectively decrease the average density probed by XRR, consistent with the measured value of 3.89 g/cm³ in Table S3.

The pronounced surface roughness also prevented reliable extraction of the film thickness from the XRR spectra. The pyramidal morphology suppresses Kiessig fringes and attenuates the primary reflection, rendering standard thickness fitting impossible. Such suppression of interference oscillations is a well-established consequence of increased surface roughness and interfacial heterogeneity in XRR analysis.

To investigate the evolution of the structural properties during annealing, XRD measurements were performed in Figure S2. The results indicate that TiN nanocrystals are formed immediately after deposition, as evidenced by the continuous diffraction rings observed in the photographic film. Reflections corresponding to seven crystallographic planes of TiN were identified, namely TiN (111), (200), (220), (311), (222), (331), and (420) [29,30].

To assess changes in the phase composition, the Debye diffraction patterns were analyzed by comparing the relative positions of the X-ray reflections in the as-deposited state and after annealing at 700 °C, 750 °C, and 800 °C for 30 min in Figure S2. The XRD intensity profiles were extracted from the photographic films using a microdensitometer, and the resulting spectra are presented in Figure 3.

Figure 3 and Figure S3 show X-ray diffraction patterns and intensity curves of titanium nitride films on Si synthesized by magnetron sputtering (1200 W, 604 s, N₂/Ar = 30/1, 0.8 Pa, 100 °C), presented for the as-deposited state and after annealing at 700–800 °C for 30 min in Ar. For the as-deposited film in Figure 3a, characteristic TiN reflections are observed, including TiN(111) at ~36.6°, a broad composite reflection near 42.5° from overlapping Ti₂N(200) (JCPDS 00-023-1455) and TiN(200) (JCPDS 00-038-1420) reflections, TiN(220) at ~61.8°, and weaker higher-order reflections, with the broadness of these reflections indicating a nanocrystalline structure with 4–6 nm grain size. After annealing at 700 °C (b), Ti₂N transforms to Ti₃N₂ (JCPDS 00-040-0958), evidenced by new reflections near 40°, while TiN(200) intensity decreases and TiN(111) remains strong. Additional reflections at ~38.8° and ~43.2° correspond to Ti₅Si₃(200) (JCPDS 00-029-1362) and Ti₅Si₄(118) (JCPDS 00-027-0907), marking the onset of Ti–Si interdiffusion. At 750 °C in Figure 3c, TiN(111) becomes the dominant reflection, indicating preferred orientation, while Ti₅Si₃ and Ti₅Si₄ diffraction reflections intensify, showing continued silicide formation. Finally, at 800 °C (d), TiN(111) orientation is further enhanced, now the strongest reflection intensity in the pattern, accompanied by well-developed Ti₅Si₃ and Ti₅Si₄ phases. This progressive phase evolution highlights nitrogen redistribution and Si diffusion through TiN with increasing temperature, while no TiSi₂ phase is detected at any stage.

This analysis shows that annealing leads to noticeable shifts in the diffraction line positions, while their number increases progressively with temperature, indicating the formation and transformation of crystalline phases. Immediately after TiN film deposition, an overlap between the Ti₂N(200) and TiN(200) diffraction lines is observed, resulting in a broadened composite reflection in the 41–45° region in Figure 3a. Such broadening is typical for closely spaced diffraction lines originating from structurally related nitride phases and suggests the simultaneous presence of both Ti₂N and TiN in the as-deposited state in Figure S3.

After annealing at 700 °C, the nanocrystalline Ti₂N phase undergoes a phase transformation into Ti₃N₂ in Figure 3b. This transformation is accompanied by a pronounced decrease in the TiN(200) diffraction line intensity, indicating a redistribution of nitrogen-deficient titanium nitride phases. Subsequent annealing at 750 °C and 800 °C promotes the development of the TiN phase, accompanied by the emergence of a pronounced preferred orientation along the TiN(111) plane. It should be noted that at these temperatures, mutual diffusion of Si and Ti atoms occurs within the Si–TiN system, leading to the formation of TiSi, Ti₅Si₄, and Ti₅Si₃ nanocrystals, as evidenced in Figure 3b–d [31,32,33]. However, nanocrystals of the low-resistance TiSi₂ phase (i.e., Ti₅Si₁₀) were not detected. This absence can be attributed to insufficient interdiffusion between Ti and Si, which is hindered by the high chemical stability of TiN nanocrystals. As a result, these nanocrystals undergo only partial decomposition during annealing at 700–800 °C for 30 min in Ar.

Figure 4 provides a schematic conceptual representation of the structural and compositional evolution of TiN films on Si substrates during annealing. In the as-deposited state of Figure 4a, the film consists of a continuous ~240 nm thick TiN layer containing randomly oriented TiN 4–6 nm nanocrystals together with Ti₂N nanocrystals on the Si substrate with a sharp TiN/Si interface. After annealing at 700 °C in Figure 4b, partial decomposition of TiN occurs with the formation of Ti₃N₂, accompanied by the emergence of interfacial Ti₅Si₃ nanocrystals near the Si interface and Ti₅Si₄ nanocrystals in regions of higher Si concentration. These crystalline phases are surrounded by an amorphous Ti–N–Si mixture resulting from nanocrystal breakdown. At 750 °C in Figure 4c, the volume fraction of both TiN and silicide phases increases due to crystallization of the amorphous material, with Ti₅Si₃ concentrated at the Si interface and Ti₅Si₄ distributed throughout the film thickness, reflecting progressive Si diffusion. After annealing at 800 °C in Figure 4d, the TiN layer develops a pronounced TiN(111) preferred orientation, illustrated by aligned crystallites, while surface agglomeration produces pyramidal features with voids between them. Continued growth of both silicide phases is observed, with gradual conversion of Ti₅Si₃ into Ti₅Si₄ and increased interfacial roughness. The spatial distribution of Ti₅Si₄, further from the interface compared to Ti₅Si₃, reflects the Si concentration gradient established by diffusion from the substrate. Arrows in the schematic indicate Si diffusion pathways. Overall, Figure 4 emphasizes that phase evolution occurs through a combination of solid-state reactions, diffusion-driven redistribution, and crystallization from transient amorphous regions.

The average crystallite sizes ε and the integral intensity of the nanocrystalline phase are summarized in

Table 1. Average crystallite sizes ε and integrated intensity I_int of nanocrystalline phases in TiN films as a function of annealing temperature.

T, °C	Ti5Si3 (200)		TiN (111)		TiN (200)		Ti2N (200)		TiN (220)		Ti5Si4 (118)		ΣTiN	ΣTiySix
	Iint	ε, nm	Iint	ε, nm	Iint	ε, nm	Iint	ε, nm	Iint	ε, nm	Iint	ε, nm	Iint	Iint
100	-	-	0.26	4.1	0.64	5.7	0.54	5.9	0.32	3.5	-	-	1.22	-
700	0.15	5.9	0.25	4.5	0.09	4.5	-	-	-	-	0.08	4.1	0.34	0.23
750	0.25	6.2	0.49	4.3	0.2	3.6	-	-	-	-	0.28	3	0.69	0.53
800	0.12	6.2	0.85	5.1	0.32	4.1	-	-	-	-	0.43	3.1	1.17	0.55

. In the as-deposited state, the film contains TiN nanocrystals with average sizes of approximately 4 nm in the (111) plane and 5.8 nm in the (200) plane, along with Ti₂N nanocrystals observed in the (200) plane. An approximate evaluation of the volume of nano-TiN, calculated as the sum of intensities I_TiN(111) + I_TiN(200) + I_TiN(220), and the volume of nanosilicides, calculated as I_Ti5Si3(200) + I_Ti5Si4(118), shows that after annealing at 700 °C for 30 min in Ar, the volume of the nano-TiN phase decreases by a factor of 2 in Table 1. This reduction is due to the decomposition of TiN nanocrystals and the formation of Ti₅Si₃(200) (~6 nm) and Ti₅Si₄(118) (~4 nm) nanocrystals, as well as the emergence of Ti₃N₂ nanocrystals and an amorphous mixture of Ti, N, and Si atoms resulting from nanocrystal breakdown.

Annealing at 750 °C leads to an increase in the volume of TiN, Ti₅Si₃, and Ti₅Si₄ nanocrystals due to crystallization of the amorphous atomic mixture and the formation of smaller nanocrystals, which may result in decreased average crystal size. After annealing at 800 °C for 30 min in Ar, both the volume of the nano-TiN phase and the crystallite sizes increase. The combined volume of Ti₅Si₃ and Ti₅Si₄ phases remains unchanged at 800 °C, but Ti₅Si₃ gradually transforms into Ti₅Si₄ due to the diffusion of Si atoms into Ti.

As a result, XRD analysis revealed that after annealing at 700 °C, 750 °C, and 800 °C, Ti₃N₂ nanocrystals formed within the TiN films alongside TiN nanocrystals. This observation is consistent with the X-ray reflectometry data, which showed a film density of 5.131 g/cm³, corresponding to a Ti₃N₂ composition (TiN_0.66). Notably, TiSi₂ nanocrystals were not detected, which may be attributed to the high stability of TiN as a diffusion barrier [34,35,36]. The strong Ti–N bonding prevents Ti atoms within TiN nanocrystals from diffusing into Si to form TiSi₂, whereas silicon atoms can still penetrate through the barrier layer at 700–800 °C, leading to the formation of TiSi, Ti₅Si₄, and Ti₅Si₃ nanocrystals. The absence of TiSi₂, in which 5 titanium atoms correspond to 10 silicon atoms, indicates insufficient silicon diffusion through the TiN layer and highlights both the effectiveness of the barrier and the inability to form a TiSi₂ layer at the Si/TiN interface below 800 °C. This behavior may also be influenced by the considerable thickness of the TiN films and the strong Ti–N bonds, which hinder mutual diffusion of Si and Ti [37]. Therefore, it is assumed that the deposition of an intermediate Ti layer is required to form a low-resistance TiSi₂ layer at the Si/TiN interface.

3.2. Barrier Properties of TiN and SiC Films Against the Diffusion of Cu Atoms into Si

Cu provides low electrical resistance, but its high diffusivity in Si leads to Cu₃Si formation during annealing, causing severe degradation of contact properties. Therefore, effective diffusion barriers are required to prevent Cu–Si interdiffusion. This subsection examines the thermal stability of Si/TiN/Cu and Si/SiC/TiN/Cu structures and evaluates the ability of TiN and SiC_x nanolayers to suppress diffusion at various temperatures. In all experiments, the Cu layer thickness was approximately 300 nm. The Si/TiN/Cu structure contained a 100 nm TiN layer, whereas the Si/SiC/TiN/Cu structure incorporated 50 nm SiC and 50 nm TiN layers. The resulting structures were subjected to annealing at 500 °C, 600 °C, 700 °C, and 800 °C for 30 min.

The thermal evolution of the Cu–TiN–Si multilayer system, shown schematically in Figure 1a, was investigated by annealing at 500–800 °C for 30 min in an Ar atmosphere. Corresponding XRD data are shown in Figure 5a. At 500 °C, XRD analysis revealed the presence of TiN(111), TiN(200), TiN(220), and Cu(111), Cu(200) phases, indicating that the TiN barrier maintained its structural integrity and the Cu layer preserved its initial crystallographic texture. Upon annealing at 600 °C, the intensity of the TiN(111) diffraction line increased, while the intensity of Cu(111) and Cu(200) diffraction lines decreased. Simultaneously, the Cu₃Si(300) phase emerged, suggesting the onset of Cu–Si interdiffusion. The increase in TiN(111) intensity is attributed to crystallite growth, whereas the decrease in Cu and other TiN diffraction lines reflects consumption of Cu in the formation of the initial Cu₃Si phase.

At 700 °C, the intensity of TiN(111) dramatically decreased, accompanied by reductions in TiN(200), TiN(220), Cu(111) and Cu(200) diffraction lines. This corresponds to partial TiN degradation and enhanced Cu–Si interaction, resulting in the formation of additional Cu₃Si phases (JCPDS 00-051-0916), namely Cu₃Si(300) and Cu₃Si(012). These observations indicate that TiN’s effectiveness as a diffusion barrier diminishes at this temperature, facilitating Cu–Si silicidation. At 800 °C, the TiN layer underwent complete structural degradation, and the Cu layer at the surface was fully oxidized due to residual oxygen. XRD patterns in Figure 5a showed no detectable TiN or metallic Cu phases, while Cu oxide phases dominated. The absence of TiN and Cu phases indicates total decomposition of the barrier and Cu layers, with Cu oxidation replacing Cu–Si interactions.

Given the limited thermal stability of TiN observed in the Cu–TiN–Si stacks, a complementary analysis was performed on Si–SiC–TiN–Cu structures in Figure 5b. The annealing of the Si/SiC/TiN/Cu multilayer structure at 500–800 °C for 30 min in an Ar atmosphere revealed that the thermal behavior of the system is strongly governed by the presence of the SiC interlayer in Figure 5b. The TiN layer maintains its structural integrity at 500 °C, and the Cu layer retains its initial crystallographic texture. No significant phase transformations were observed at 600 °C, suggesting that the multilayer stack remains structurally stable at this temperature. The intensity of the Cu(200) (JCPDS 00-004-0836) phase slightly decreases, and the CuO(020) (JCPDS 00-041-0254) phase begins to appear.

At 700 °C, the intensity of the TiN(111) diffraction line decreased, whereas the Cu(111) and Cu(200) diffraction lines increased in Figure 5b. This change is attributed to recrystallization and grain growth within the Cu layer, leading to structural reorganization rather than chemical interaction. Importantly, no Cu₃Si silicide phase was detected at this temperature. This behavior is explained by the characteristics of the SiC layer: Si atoms in SiC are strongly bonded in the rigid Si–C covalent lattice and therefore are not available to react with Cu. In addition, SiC acts as a highly effective diffusion barrier, significantly suppressing Cu–Si interdiffusion. As a result, the Cu₃Si phase, which is commonly observed around 600 °C in conventional Si/TiN/Cu systems, does not form in the presence of the SiC interlayer.

At 800 °C, degradation of the TiN barrier results in exposure of the Cu surface, which subsequently reacts with residual oxygen present in the argon atmosphere, forming copper oxide phases. Nevertheless, the Cu₃Si phase remains absent even at 800 °C. The lack of Cu–Si interaction at 800 °C further confirms the exceptional thermal and chemical stability of SiC. Although the formation of Cu₃Si would require the release of Si atoms from the SiC lattice, breaking the strong Si–C bonds demands significantly higher energy, making such a reaction impossible at 800 °C. Thus, even when TiN undergoes degradation, oxygen–copper reactions dominate over any potential Cu–Si interaction, preventing silicide formation.

To study the barrier properties, a three-layer Si/TiN/SiC/Cu structure, schematically shown in Figure 1c, was also synthesized. The TiN and SiC layer sequence was changed to investigate the effects of layer arrangement on barrier characteristics. Film parameters and deposition modes remained unchanged.

For the assessment of electrical characteristics, the sheet resistance was measured using the four-point probe technique. The sheet resistance values obtained for all multilayer structures at different annealing temperatures are summarized in

Table 2. Sheet resistance R_s of multilayer structures as a function of annealing temperature.

	Rs, Ohm/sq
Thin Films	RT	500 °C	600 °C	700 °C	800 °C
c-Si/TiN/Cu	0.27	79.4	268,576	251,791	3,139,036
c-Si/SiC/TiN/Cu	0.23	11.9	32.5	2.3	2,987,958
c-Si/TiN/SiC/Cu	0.23	0.74	29	4.9	3,941,129

. At room temperature, all samples exhibit low 0.23–0.27 Ω/sq sheet resistance, reflecting the high conductivity of the as-deposited Cu layer. After annealing at 500 °C, the Si/TiN/Cu structure shows a substantially increased 79.4 Ω/sq sheet resistance, whereas the Si/SiC/TiN/Cu and Si/TiN/SiC/Cu structures experience only minor changes (11.9 Ω/sq and 0.74 Ω/sq, respectively). This difference indicates that the presence of the SiC interlayer, regardless of its position in the stack, significantly improves the thermal stability of the metallization scheme at moderate temperatures.

The sheet resistance and thermal stability of the TiN/SiC multilayer diffusion barriers developed in this work were benchmarked against other copper barrier systems reported in the literature. The results clearly demonstrate the superior performance of the SiC-containing structures relative to conventional TiN-only barriers. The Si/TiN/Cu stack failed at 600 °C, with sheet resistance R_s rising sharply to more than 2.5 × 10⁵ Ω/sq at 700 °C, confirming rapid Cu–Si interdiffusion and silicide formation. In contrast, both Si/SiC/TiN/Cu and Si/TiN/SiC/Cu stacks remained stable up to 700 °C, exhibiting remarkably low sheet resistance values of 2.3 Ω/sq and 4.9 Ω/sq, respectively. This level of stability is comparable or superior to other advanced barrier systems such as TaN and Ti/TiN bilayers, which typically fail around 650 °C with sheet resistance in the 10³–10⁴ Ω/sq range [27]. Furthermore, the performance of the TiN/SiC multilayers approaches W/SiC composite barriers, which remain stable up to 750 °C but require more complex fabrication. These comparisons highlight that the dual-barrier mechanisms of TiN and SiC provide complementary protection. SiC effectively blocks Cu ingress due to strong Si–C bonding, while TiN maintains electrical conductivity and prevents Si outdiffusion. The ability of these multilayers to sustain low resistance at 700 °C underscores their potential for integration into Cu-based metallization schemes in advanced photovoltaic and microelectronic devices, offering a simpler and scalable alternative to more complex refractory metal systems. Similar limitations of TiN-only barriers have been reported in recent work on amorphous-like TiN films, which show significant resistance increases upon annealing above 600 °C [14].

To compare the structural transformations in the Si/SiC/TiN/Cu and Si/TiN/SiC/Cu multilayer stacks, a comparative XRD analysis of both films annealed at 700 °C was performed. As seen in Figure 6, the intensity of the TiN(111) diffraction line in the Si–SiC–TiN–Cu system annealed at 700 °C for 30 min in an argon atmosphere is noticeably lower (I_TiN(111) = 0.42, FWHM = 0.58°) compared to the Si–TiN–SiC–Cu system (I_TiN(111) = 0.71, FWHM = 0.41°). This reduction in intensity and an increase in FWHM indicate significant structural degradation. This reduction is attributed to the partial decomposition of the TiN layer caused by the active diffusion of Cu atoms into the TiN layer. Such diffusion leads to structural disordering within the TiN phase, thereby weakening its XRD signal.

The X-ray diffractograms in Figure 6 suggest distinct differences in the structural stability of TiN depending on the multilayer configuration. The TiN(111) reflection at ~36.6° exhibits markedly different characteristics. In the Si/SiC/TiN/Cu stack, its intensity is reduced and FWHM broadens, suggesting Cu-induced structural degradation of the TiN layer. In contrast, the Si/TiN/SiC/Cu stack maintains a higher intensity and a sharper profile in terms of the FWHM, demonstrating preserved crystallinity and improved barrier performance. The TiN(200) reflection at ~42.6° follows a similar trend, further confirming the stabilizing effect of the SiC interlayer when positioned adjacent to Cu. Reflections corresponding to Cu(111) at ~43.3° and Cu(200) at ~50.4° are observed in both samples with comparable intensity, confirming similar Cu layer quality across the two configurations. Importantly, no Cu₃Si phases are detected at 700 °C in either stack, validating the effectiveness of SiC in suppressing Cu–Si interdiffusion. Minor CuO reflections appear in both cases, attributed to surface oxidation during annealing.

Overall, this comparison demonstrates that placing SiC between TiN and Cu in the Si/TiN/SiC/Cu configuration provides superior protection to the TiN barrier layer by preventing Cu penetration, thereby maintaining structural integrity and ensuring lower electrical resistance.

These results highlight the critical role of SiC as a robust high-temperature diffusion barrier and demonstrate its clear advantages over conventional TiN in Cu-based metallization schemes. Building on this understanding, future research should focus on optimizing the thickness, stoichiometry, and deposition parameters of SiC and TiN_x/SiC_x composite layers to further enhance barrier performance. Additionally, systematic studies employing rapid thermal annealing, long-term thermal cycling, and in situ structural diagnostics may provide deeper insight into the diffusion mechanisms and failure modes of multilayer stacks. Advanced characterization techniques such as cross-sectional transmission electron microscopy with energy-dispersive X-ray spectroscopy mapping, X-ray photoelectron spectroscopy depth profiling, and time-resolved XRD during annealing would provide valuable complementary information about the spatial distribution of elements, chemical states, and real-time phase evolution. The integration of such engineered barrier layers into full TOPCon and SHJ solar-cell architectures, combined with scalable Cu plating, represents a promising pathway toward low-cost, thermally stable, and industry-compatible metallization solutions for next-generation photovoltaic technologies.

4. Conclusions

The study results showed that, upon comparing the Si/TiN/Cu, Si/TiN/SiC/Cu, and Si/Si/SiC/TiN/Cu structures, the Cu₃Si phase forms at 600 °C only in the Si/TiN/Cu system, while the Si/SiC/TiN/Cu sample remained stable up to 700 °C. At 800 °C, copper in all structures underwent complete oxidation due to the small amount of oxygen present in the argon atmosphere. This behavior was also confirmed by sheet resistance measurements. Overall, multilayer barriers based on TiN and SiC demonstrate high effectiveness in enhancing the thermal stability of the copper–silicon interface.

The structural stability of the Si/SiC/TiN/Cu and Si/TiN/SiC/Cu multilayer stacks annealed at 700 °C was compared, revealing that the stability of the TiN layer is directly dependent on the layer sequence. In the Si–SiC–TiN–Cu system, diffusion of Cu atoms into the TiN layer disrupts its structure, resulting in reduced XRD intensity and increased reflection broadening. In contrast, in the Si–TiN–SiC–Cu configuration, the intermediate SiC layer acts as an effective diffusion barrier, preserving the TiN structure and ensuring relatively higher XRD line intensity and lower sheet resistance (4.9 Ω/sq vs. 2.3 Ω/sq). This dual-barrier mechanism, where SiC blocks Cu diffusion, and TiN prevents Si diffusion, provides optimal thermal stability for Cu-based metallization in photovoltaic applications.

Furthermore, since a TiSi₂ layer does not form when TiN is deposited and annealed directly on a silicon substrate below 800 °C, a Ti interlayer is necessary to obtain the TiSi₂/TiN system required for low-resistance ohmic contacts.