Improving Thermoelectric Properties of Bi2Te3 Thin Films By Manganese Co-Sputtering

Summary

A radiofrequency co-sputtering protocol was developed to fabricate manganese-doped Bi₂Te₃ thin films and evaluate how manganese input influences structural and thermoelectric transport properties. Moderate manganese incorporation improved film uniformity and power-factor-related transport behavior, while higher manganese input increased structural disorder and resistivity.

Abstract

Bi₂Te₃ remains a benchmark n-type thermoelectric (TE) material for low-temperature energy conversion, but its small band gap can reduce efficiency because of thermally generated parasitic carriers. Elemental doping has been explored to improve TE performance, although systematic studies on manganese (Mn)-doped Bi₂Te₃ thin films remain limited. In this study, a radiofrequency magnetron co-sputtering workflow was used to fabricate Mn-doped Bi₂Te₃ thin films by varying Mn target power while maintaining constant Bi₂Te₃ deposition conditions. Structural, microstructural, compositional, and TE transport properties were evaluated using X-ray diffraction, field-emission scanning electron microscopy, energy-dispersive X-ray spectroscopy, and temperature-dependent transport measurements. X-ray diffraction confirmed retention of the rhombohedral Bi₂Te₃ phase with a preferred (015) orientation, while peak shifts toward higher 2θ values were consistent with Mn-related lattice contraction. All films exhibited negative Seebeck coefficients, confirming n-type conduction. Increasing Mn doping enhanced the magnitude of the Seebeck coefficient but also increased electrical resistivity, demonstrating a transport tradeoff. The film deposited at 5 W Mn power achieved the highest power factor of 529.33 µW m⁻¹ K⁻² at 523 K because of its low resistivity combined with adequate thermopower. These results demonstrate that moderate Mn incorporation can improve the power-factor-related performance of Bi₂Te₃ thin films within the measured temperature range.

Introduction

Thermoelectric (TE) materials play a crucial role in the sustainable conversion of heat into electrical energy, primarily through the Seebeck and Peltier effects. The performance of TE materials is typically evaluated using the power factor (PF = S²/ρ), where S is the Seebeck coefficient and ρ is the electrical resistivity^1,2. A high S coupled with low ρ indicates efficient carrier transport, which is essential for applications such as waste-heat recovery, micro-power generation, and solid-state cooling in industrial systems and microelectronics^3,4,5.

Among various TE materials, bismuth telluride (Bi₂Te₃) is notable for its efficiency in near-room-temperature applications. However, its practical deployment is limited by brittleness, restricted mechanical flexibility, and a moderate figure of merit (zT), which contribute to challenges in large-scale implementation². Recent advances in thin-film fabrication technologies, particularly magnetron sputtering, provide promising approaches to address these limitations. Thin films enable improved control over composition and microstructure, which can enhance carrier mobility and mechanical flexibility, both of which are important for high-efficiency TE devices⁶. The ability to tailor these properties using radiofrequency (RF) magnetron sputtering has been demonstrated previously, supporting its use for producing uniform Bi₂Te₃ thin films with tunable thicknesses and doping levels⁷.

Cationic doping, particularly with manganese (Mn), has emerged as a promising strategy for improving the TE properties of Bi₂Te₃. Although research on Mn-doped Bi₂Te₃ remains less extensive than studies involving other dopants, evidence from related systems suggests that Mn incorporation can improve both electrical and structural properties⁸. Mn incorporation has been reported to influence carrier concentration and mobility, affecting the Seebeck coefficient and electrical conductivity, which together determine the PF of TE material⁹. In addition, Mn incorporation has been discussed in terms of substitutional or defect-associated incorporation mechanisms that may introduce lattice strain and point defects. These modifications can enhance phonon scattering and potentially reduce lattice thermal conductivity, both of which are important contributors to TE performance¹⁰. Research on related Mn-doped compounds such as Bi₂Si₂Te₆ suggests that Mn may also improve charge transport and introduce localized magnetic moments, which may strengthen phonon-scattering effects and reduce bipolar conduction¹¹. Consequently, the influence of Mn on microstructural evolution, defect formation, and grain growth is important for improving the mechanical and thermal stability of TE materials^10,12.

However, excessive Mn incorporation can promote secondary phases such as MnTe and complicate transport behavior¹³; therefore, Mn input must be carefully controlled¹⁴. In addition, Mn may introduce acceptor states within the Bi₂Te₃ lattice and potentially alter conduction behavior through suppression of topological surface transport, opening of a gap, reduced free-carrier density, and a crossover from metallic to insulating or localized conduction under sufficient doping^15,16. These considerations highlight the importance of controlling Mn incorporation during sputtering to establish reliable structure–property relationships¹⁷.

Recent studies have demonstrated that Mn can be incorporated into Bi₂Te₃-based thin films and can measurably influence structure and electronic transport. Bi_2−xMn_xTe₃ thin films exhibit Mn-dependent changes in structural and electrical properties¹⁸, while Mn:Bi₂Te₃ thin films have also been investigated through transport measurements, indicating that Mn incorporation can occur without fully suppressing the host Bi₂Te₃ phase¹⁹. More recently, magnetron sputtering has enabled the preparation of nanocrystalline Mn-doped Bi₂Te₃ thin films, supporting the feasibility of Mn incorporation using scalable thin-film deposition approaches²⁰. However, compared with representative sputtered n-type Bi₂Te₃ thin films optimized through texture engineering, vacancy-defect modulation, and carrier engineering, systematic optimization of Mn-doped Bi₂Te₃ thin films remains comparatively underexplored^7,21. This gap motivated the controlled Mn-input study presented here.

In this study, Mn-doped Bi₂Te₃ thin films were prepared by RF magnetron co-sputtering using a controlled Mn-input series and were evaluated using X-ray diffraction (XRD), field-emission scanning electron microscopy (FESEM), energy-dispersive X-ray spectroscopy (EDX), and temperature-dependent Seebeck coefficient and electrical resistivity measurements. Figure 1 schematically illustrates the co-sputtering configuration used in this work. The aim was to use a consistent sputtering framework to examine how Mn input influences film microstructure and PF-related transport behavior under comparable processing conditions²².

Figure 1. Schematic of the RF magnetron co-sputtering configuration used for Mn-doped Bi₂Te₃ thin-film deposition. Schematic representation of the co-sputtering arrangement showing the Bi₂Te₃ target, Mn target, substrate holder, and substrate positioning inside the deposition chamber. Please click here to view a larger version of this figure.

Protocol

No human participants, animal subjects, or biological tissues were used in this study; therefore, institutional ethics approval was not required.

Overview
Mn-doped Bi₂Te₃ thin films were deposited on soda-lime glass (SLG) substrates by RF magnetron co-sputtering. Depositions were performed without intentional substrate heating, and the chamber/substrate temperature during sputtering was maintained at 298 K ± 1 K. The target-to-substrate distance was fixed at 55 mm for both sputter guns. The Bi₂Te₃ target was operated at a fixed RF power of 100 W, while the Mn target RF power was varied (0, 5, 10, and 15 W) to tune Mn incorporation across the doping series. Argon (Ar) was used as the sputtering gas at 4 sccm, and after plasma stabilization, the working pressure during deposition was maintained at approximately (3–5) × 10⁻³ Torr. The substrate holder was rotated continuously at 10 rpm in a single fixed direction to support film uniformity. Film thicknesses after 30 min deposition ranged from 0.722–1.195 µm, depending on Mn RF power (Table 2). This protocol assumes familiarity with standard vacuum practice and safe operation of RF sputtering and ultraviolet (UV)–ozone equipment, and minor adjustments may be required depending on the sputtering system used²³.

Substrate preparation
SLG substrates were cut into 1.5 cm × 2.5 cm pieces, handled by the edges to avoid contamination, and washed thoroughly using laboratory detergent. Substrates were cleaned using a diluted laboratory glassware detergent solution prepared in DI water and then rinsed thoroughly until no visible residue remained. DI water with resistivity of ≥18.2 MΩ·cm at 298 K was used for all rinsing steps. The substrates were then sequentially cleaned in an ultrasonic bath using methanol, acetone, and isopropyl alcohol for 10 min each at 37 kHz, followed by rinsing in DI water for 20 min in a temperature-controlled ultrasonic bath at 353 K²³. After cleaning, the substrates were held at the edges using clean tweezers, dried using gentle nitrogen gas (N₂, 99.99% purity) until no visible liquid remained, and heated on a hot plate at 393 K for 10 min to remove residual moisture. Immediately before deposition, substrates were treated in a UV–ozone cleaner for 10 min using the instrument’s default operating conditions.

Chamber preparation and target installation
Before deposition, dust was removed from the chamber interior and sputter guns using a clean rubber blower. The blower was visually inspected and purged several times away from the chamber before use to minimize accidental particle introduction. Prior to the first deposition run, any old foil was removed, and fresh heavy-duty aluminum foil was used to line the chamber surfaces and cover the substrate holder. The foil was secured tightly to prevent shifting during pumping and substrate rotation and to reduce contamination from residues of previous depositions. The chamber viewing window was protected using a clean petri dish lid (approximately 94 mm diameter) to minimize film buildup on the window while still allowing plasma observation during pre-sputtering and deposition. The Bi₂Te₃ target was installed on RF gun 1 and the Mn target on RF gun 2 (Figure 2), ensuring that both targets were centered and securely assembled, and the guns were covered with aluminum shields. No additional pre-use cleaning was performed on the targets; however, surface conditioning was achieved during the pre-sputtering step to remove surface contamination and native oxides. The substrate holder was wiped with methanol and allowed to dry completely. Electrical isolation was then checked using a multimeter in continuity (audible) mode. Grounding was accepted when an audible tone was detected between the chamber body and gun housing, whereas electrical isolation was confirmed when no tone was detected between the chamber body and each target surface (open circuit).

Figure 2. Annotated photographs of the RF magnetron co-sputtering system used for Mn-doped Bi₂Te₃ thin-film deposition. Representative photographs showing the sputtering chamber, RF guns, Bi₂Te₃ and Mn targets, substrate holder, substrates, cooling system, vacuum pump, and parameter-control units used during thin-film deposition. Please click here to view a larger version of this figure.

Mn-doped Bi₂Te₃ thin-film deposition
Pre-deposition
Test substrates were initially placed on the rotating substrate holder to evaluate deposition stability during target conditioning. The chamber door was closed, and pump-down was initiated. Gentle pressure was applied around the door at the start to ensure proper sealing, and the pressure drop was confirmed on the gauge. Chamber pressure was monitored using the system vacuum gauge integrated with the sputtering system controller. Pumping continued until chamber pressure reached ≤5 × 10⁻³ Torr. The system was considered ready when the pressure reading remained stable for ~3 min, corresponding to a pressure drift of ≤2 × 10⁻⁴ Torr during the stabilization period.

The cooling system was switched on and set to 288 K, ensuring active coolant circulation through both sputter guns before plasma ignition. Cooling was provided using recirculated DI water. High-purity argon gas (Ar, ≥99.999%) was introduced at 4 sccm and allowed to stabilize for ~2–5 min, or until the flow remained steady at ≥3.9 sccm. Substrate rotation was set to 10 rpm in a single fixed direction and maintained continuously throughout pre-sputtering and deposition. The RF power supply and matching-network controller were then powered on.

For both RF guns, matching-network settings were adjusted using the Min/Max function until Load = 50 W and Tune = 50 W, after which controller mode was switched from Manual to Auto matching. Reflected power was minimized prior to plasma ignition, and ≤5 W reflected power was used as the acceptance threshold. For pre-sputter conditioning, RF power was set to 50 W on the Bi₂Te₃ gun and 10 W on the Mn gun before plasma ignition. Stable plasma was defined as a steady glow maintained for ~1–2 min without visible flicker or sudden pressure spikes. In addition to visual inspection, plasma stability was confirmed when reflected power remained ≤5 W and operating pressure fluctuations remained within ±5% for ~1–2 min. Once stable plasma was achieved at both guns, the targets were pre-sputtered for 15 min to remove surface contamination and native oxides. If plasma failed to ignite, sputtering was stopped and a short Ar reset was performed by switching the Ar flow off for 10 s and then restoring the flow before reattempting ignition. Plasma ignition was attempted up to three times per gun. If ignition remained unsuccessful after three attempts, the process was terminated and the system was returned to a safe standby state for inspection of gas flow, cooling circulation, and matching conditions. After pre-sputtering, the test substrates were removed and the cleaned substrates were loaded for the deposition run.

Thin-film deposition
Before loading substrates for the deposition run, dust was removed from the chamber interior and sample-holder region using a clean rubber blower, and the viewing window remained protected with a clean petri dish lid for plasma observation. The targets were inspected to ensure that they were properly seated, centered, and in good condition. Targets were considered suitable when the surface appeared intact and uniformly colored, with no visible cracks, deep pits, delamination, arcing marks (“burn” spots), or loose particles. In addition, each target was required to be firmly clamped and properly centered within the sputter gun. Cleaned substrates were placed symmetrically near the outer region of the rotating holder (approximately 5–10 mm from the holder center), with the longer side aligned parallel to the holder edge (Figure 2).

The chamber was then pumped down to a base pressure of 5 × 10⁻⁵ Torr or lower. For the first deposition run, evacuation was performed for at least 5 h or until the base pressure stabilized at the target value. Base-pressure stability was defined as ≤5% pressure drift over ~10 min. After reaching base pressure, the cooling system was stabilized at 288 K, Ar flow was set to 4 sccm and stabilized (≥3.9 sccm), and substrate rotation was maintained at 10 rpm. Matching-network settings for both guns were confirmed (Load = 50 W and Tune = 50 W), and Auto matching was enabled.

Deposition was carried out with Bi₂Te₃ RF power fixed at 100 W and Mn RF power initially set to 5 W for the first Mn-doped condition. The target-to-substrate distance was fixed at 55 mm for both the Bi₂Te₃ and Mn sputter guns. For easier plasma ignition, the Bi₂Te₃ gun could be initiated at 70 W and then increased by 5 W every 10 s until reaching 100 W. The deposition timer was started only after the Bi₂Te₃ power readback remained at 100 W for ~30 s and plasma stability was confirmed. Films were deposited for 30 min. Key co-sputtering parameters are summarized in Table 1. The resulting film thicknesses were 0.886 µm (0 W), 0.722 µm (5 W), 1.079 µm (10 W), and 1.195 µm (15 W), as determined from cross-sectional analysis (Table 2). The deposition steps were repeated for Mn RF powers of 10 and 15 W using fresh substrates while keeping Bi₂Te₃ power, Ar flow, rotation speed, base pressure, temperature, and deposition time unchanged.

Parameter	Bi2Te3 Target	Mn Target
RF power (W)	100	0, 5, 10, 15
Working gas	Ar	Ar
Gas flow rate (sccm)	4	4
Base pressure (Torr)	5 × 10⁻⁵	5 × 10⁻⁵
Target-to-substrate distance (mm)	55	55
Pre-sputtering time (min)	15	15
Substrate rotation speed (rpm)	10	10
Deposition temperature (K)	298 ± 1	298 ± 1
Deposition time (min)	30	30

Table 1: RF magnetron co-sputtering parameters used for the fabrication of Mn-doped Bi₂Te₃ thin films.

Post-deposition handling
After the deposition time was complete, the system was allowed to complete its shutdown sequence. Shutdown was initiated through the system controller: RF output was switched off first, followed by reduction of the Ar flow to 0 sccm, while the cooling system remained active until the chamber was vented and the sputter guns returned to near-ambient temperature. The chamber was vented only after the turbomolecular pump (TMP) was no longer running to avoid equipment damage. The TMP status indicator was checked to confirm that the pump had stopped (speed = 0) before venting. The chamber was vented slowly using dry nitrogen through the vent valve to allow gradual pressure equalization and minimize particle disturbance.

The chamber was opened only after the pressure returned to approximately atmospheric pressure (≈760 Torr) and the chamber door could be released without resistance. A mask and gloves were worn during sample removal to reduce contamination and exposure to fine particles. Samples were transferred into a clean petri dish and stored in a vacuum storage container. Characterization was typically performed within 24–72 h after deposition. After sample removal, the chamber was rough-pumped for ~10–15 min until reaching the typical roughing-pressure range required before TMP operation (approximately 10^–1 Torr) to reduce ambient air and moisture exposure before switching off the remaining system controls.

Routine post-growth characterization and calculations
Crystal structure was examined by θ–2θ XRD using Cu Kα radiation (λ = 1.5406 Å) to confirm Bi2Te3 phase formation, preferred orientation, and to screen for potential secondary phases by comparison with standard JCPDS/ICDD references^24,25. Scans were collected over 2θ = 5–80° with a step size of 0.05° and a counting time of 0.5 s per step in continuous-scan mode, using a Cu tube operated at 40 kV and 40 mA. For semi-quantitative comparison, patterns were background-corrected using a visually verified automated baseline routine, followed by automatic peak search using a fixed threshold, after which the identified peaks were used for PDF matching²⁶. The diffraction patterns were matched to Bi₂Te₃ (PDF 00-015-0863) and Mn-related reference cards, including α-Mn (PDF 00-032-0637) and β-Mn (PDF 00-033-0887). The reported match percentages were used only as comparative indicators of relative crystalline contribution because thin-film preferred orientation can bias intensity-based phase estimates.

The crystallite size was estimated from the dominant Bi₂Te₃ (015) reflection using the Scherrer method^27,28. The (015) peak position (2θ) was recorded from the XRD pattern, and the full width at half maximum (FWHM, β) was obtained by Gaussian peak fitting with the baseline fixed to the local background over a restricted (015) fitting window, where β was recorded in degrees. Fits were accepted when residuals were minimized without systematic deviation. The Bragg angle was calculated as θ = (2θ)/2, and β was converted to radians using:

Crystallite size was then calculated using the Scherrer equation:

where K = 0.89, λ = 1.5406 Å, β_rad is the FWHM in radians, and θ is the Bragg angle of the (015) peak²⁹. The resulting D values were converted from Å to nm by dividing by 10. The microstrain (ε) was estimated from the peak broadening using:

where β is the FWHM in radians and θ is the Bragg angle³⁰. The dislocation density (δ), which represents the defect density within the crystal, was calculated from the crystallite size using:

where D is the crystallite size. In this work, D was expressed in nm, and therefore δ was reported in nm^{−2 31}. Representative structural calculations, including β conversion, crystallite size, microstrain, and dislocation density, are summarized in Supplementary Table 1.

Surface microstructure was assessed by FESEM operated at 3.0 kV using consistent imaging conditions (100.0 k× magnification, field of view = 2.79 µm, scale bar = 100 nm) across all Mn conditions for direct comparison³². Samples were mounted on aluminum stubs using conductive carbon tape. No conductive coating was applied; charging was minimized by using a low accelerating voltage (3.0 kV) and stable imaging conditions. Film composition was evaluated using EDX acquired with the detector integrated with the electron microscope³³. EDX point spectra were acquired at 15.0 kV with a working distance of ~8.0–8.2 mm in point-analysis mode. Acquisition was performed using the instrument’s default settings, and quantitative analysis with background subtraction was used to report wt.% and at.% values.

In-plane TE transport was measured using a Seebeck/resistivity measurement system. Thin-film samples (~2.0 cm × 1.5 cm) were mounted using a platinum four-contact adapter (Figure 3), and the Seebeck coefficient (S) was measured relative to a Pt standard under an applied temperature gradient of approximately 25 K³⁴. Measurements were performed stepwise from 323–523 K under a He gas atmosphere (~1 atm). Temperature was increased between setpoints at an average rate of ~5–6 K·min⁻¹, and at each setpoint the sample was allowed to equilibrate until the S and ρ readings stabilized (typically ~2–3 min). Three consecutive readings were then recorded per temperature point at 1.5 min intervals. The PF value was calculated from the measured S (V/K) and electrical resistivity, ρ (Ω·m), using the following equation³⁵:

Figure 3. Thin-film mounting configuration for in-plane thermoelectric (TE) measurements using a platinum four-contact adapter. Representative photographs showing the mounting arrangement of Mn-doped Bi₂Te₃ thin films during Seebeck coefficient and electrical resistivity measurements, including the platinum adapter, thermocouple probes, and electrical contact pads. Please click here to view a larger version of this figure.

Results

Representative outcomes are presented for the Mn power series (0, 5, 10, 15 W), using the undoped film (0 W) as the control condition for structural and transport comparisons.

Crystal structure (XRD)
XRD patterns confirm that all samples retained the rhombohedral Bi₂Te₃ phase (JCPDS No. 15-0863) across the Mn power series, as shown in Figure 4. The dominant reflections, including a strong (015) preferred orientation, indicate that the Bi₂Te₃ crystal structure was preserved after Mn co-sputtering. A key indicator of Mn incorporation is the slight shift of the dominant diffraction peak toward higher 2θ values with increasing Mn input. This shift is shown in the overlaid XRD patterns in Figure 5 and summarized numerically by the (015) peak positions in Table 2. The shift toward higher 2θ values is consistent with lattice contraction and compressive strain associated with Mn-related lattice modification^36,37. Peak broadening also evolved with Mn input (Table 2; Supplementary Table 1), where the crystallite size generally decreased from 14.23 nm (0 W) to 8.35 nm (15 W), consistent with reduced coherent domain size and increased ε value at higher Mn power³⁸. Similar doping-related peak shifts, FWHM broadening, and crystallite-size reduction associated with increased ε have been reported in Mn-doped NiO and related doped oxide systems^20,39.

Parameter	0 W	5 W	10 W	15 W
Peak position, 2θ (°)	27.88	28.026	28.045	28.282
FWHM, β (°)	0.56889	0.727	0.705	0.971
Crystallite size, D (nm)	14.23	11.17	11.51	8.35
Microstrain, ε	0.01	0.01271	0.01232	0.01682
Dislocation density, δ (nm-2)	0.00494	0.00801	0.00755	0.01434
Estimated Bi₂Te₃ contribution by PDF matching (%)	100	94.1	87.5	72
Estimated Mn-related PDF match (%)	0	5.9 β-Mn	12.5 α-Mn	11.1 α-Mn + 16.9 β-Mn
Film thickness (µm)	0.886	0.722	1.079	1.195

Table 2: Structural parameters of Mn-doped Bi₂Te₃ thin films derived from XRD analysis of the dominant (015) reflection. Includes 2θ position, full width at half maximum (FWHM), crystallite size (D), microstrain (ε), dislocation density (δ), semi-quantitative Bi₂Te₃ contribution estimated from PDF matching, and film thickness.

Figure 4. X-ray diffraction patterns of Mn-doped Bi₂Te₃ thin films deposited at different Mn target powers. XRD patterns of films deposited at Mn target powers of 0, 5, 10, and 15 W, together with the standard Bi₂Te₃ reference pattern (ICDD PDF 00-015-0863). The dominant (015) reflection and additional indexed peaks are indicated. Please click here to view a larger version of this figure.

Figure 5. Shift of the dominant Bi₂Te₃ (015) diffraction peak with increasing Mn target power. Overlaid XRD patterns highlighting the Mn-dependent shift of the dominant (015) reflection toward higher 2θ values for films deposited at Mn target powers of 0, 5, 10, and 15 W. Please click here to view a larger version of this figure.

In the present films, the estimated ε increased from 0.01000 (0 W) to 0.01682 (15 W), while the δ value increased from 0.00494 to 0.01434 nm⁻² (Table 2; Supplementary Table 1), supporting the interpretation that higher Mn input introduced greater lattice distortion or defect density. These structural changes may enhance phonon scattering, which can be beneficial for reducing lattice thermal conductivity, but may also increase carrier scattering and electrical resistivity, providing a plausible explanation for the non-monotonic TE response observed with increasing Mn input⁴⁰.

Semi-quantitative PDF matching analysis (Table 2) indicated that Bi₂Te₃ remained the dominant matched phase across the 0–15 W Mn power range. The estimated Bi₂Te₃ contribution decreased from 100.0% (0 W) to 94.1% (5 W), 87.5% (10 W), and 72.0% (15 W), while the remaining matched contribution was associated with Mn-related reference phases, including α-Mn (PDF 00-032-0637) and β-Mn (PDF 00-033-0887). No additional prominent crystalline peaks were observed in Figure 4 within the XRD detection limits. The reported PDF match values are presented only as comparative indicators because strong preferred orientation in thin films can bias intensity-based phase estimation²⁶ and therefore should not be interpreted as absolute phase fractions. The systematic changes observed in the XRD patterns provide structural context for the FESEM microstructural trends discussed in Figure 6.

Surface morphology and thickness (FESEM)
Surface FESEM images demonstrate clear changes in film microstructure with increasing Mn input, as shown in Figure 6. The control film (0 W) exhibits relatively larger polygonal grains with more distinct grain boundaries. At 5 W Mn input, the film surface appears more compact with finer grain features, suggesting modified nucleation and growth behavior during co-sputtering^41,42. At 10 and 15 W, the films appear more densely packed with reduced intergranular gaps compared with the 0 and 5 W samples, indicating improved surface coverage and a more compact microstructure⁴³.

Film thickness values extracted from cross-sectional measurements are summarized in Table 2. The film thickness exhibited a non-monotonic dependence on Mn target power. The undoped control film (0 W) showed a thickness of 0.886 µm, which decreased to 0.722 µm at 5 W Mn input, followed by an increase to 1.079 µm at 10 W and a maximum thickness of 1.195 µm at 15 W. This non-linear trend is plausible because changes in the co-sputtering power ratio can influence incorporation efficiency and thin-film growth kinetics, resulting in non-monotonic thickness behavior in chalcogenide thin films²⁰.

Figure 6. FESEM micrographs of Mn-doped Bi₂Te₃ thin films deposited at different Mn target powers. Surface FESEM images of Mn-doped Bi₂Te₃ thin films deposited by RF magnetron co-sputtering at Mn target powers of (a) 0 W, (b) 5 W, (c) 10 W, and (d) 15 W. All micrographs were acquired under identical imaging conditions (100.0 k× magnification; scale bar = 100 nm) for direct morphological comparison. Please click here to view a larger version of this figure.

Composition (EDX)
Elemental composition analysis by EDX confirmed that Mn incorporation increased with increasing Mn target power, as summarized in Table 3 and Supplementary Table 2. The Mn concentration gradually increased across the deposition series, reaching the highest measured value of 0.59 at.% at 15 W. The increasing Mn trend is generally consistent with the XRD-based structural observations, although the absolute values differ because EDX and PDF matching evaluate different material characteristics. The compositional trend shown in Table 3 and Supplementary Table 2 also supports the structural and microstructural observations discussed previously. Across the 5–15 W Mn power range, the measured Mn content remained relatively modest (0.01–0.59 at.%), which is consistent with Mn incorporation occurring without obvious disruption of the rhombohedral Bi₂Te₃ phase observed in the XRD patterns in Figure 4⁴⁴.

Mn RF power (W)	Bi (at.%)	Te (at.%)	Mn (at.%)
0	42.31	57.69	0.00
5	53.40	46.59	0.01
10	52.58	47.30	0.12
15	42.64	56.78	0.59

Table 3: Atomic composition of Mn-doped Bi₂Te₃ thin films determined by EDX analysis. Atomic percentages of Bi, Te, and Mn as a function of Mn target power.

Thermoelectric transport (S, ρ, and PF)
Temperature-dependent TE transport measurements demonstrate that Mn incorporation influences the S and ρ values differently, resulting in a distinct maximum in the PF. All films exhibited negative S values, confirming n-type conduction behavior, and the magnitude of S generally increased with temperature for all samples, as shown in Figure 7. The highest Seebeck coefficient magnitude was observed for the 15 W film, reaching −162.08 µV·K⁻¹ at 523 K (Table 4; Supplementary Table 3). This enhancement is consistent with Mn-induced modifications in carrier-scattering behavior and transport mechanisms discussed in the preceding structural analyses^45,46.

Figure 7. Temperature dependence of the Seebeck coefficient of Mn-doped Bi₂Te₃ thin films. Seebeck coefficient (S) as a function of temperature for films deposited at Mn target powers of 0, 5, 10, and 15 W. All measured S values were negative, indicating n-type conduction behavior. The largest Seebeck coefficient magnitude was observed for the 15 W film at 523 K. Please click here to view a larger version of this figure.

Temperature (K)	Seebeck coefficient (µV/K) S ± standard deviation (SD)
	0 W (µV·K⁻¹)	5 W Mn (µV·K⁻¹)	10 W Mn (µV·K⁻¹)	15 W Mn (µV·K⁻¹)
323	−126.02 ± 1.74	−87.80 ± 0.68	−87.68 ± 1.80	−90.92 ± 1.70
373	−131.97 ± 3.51	−93.43 ± 1.95	−98.15 ± 5.31	−106.42 ± 5.79
423	−133.71 ± 5.77	−107.76 ± 7.52	−110.94 ± 8.55	−132.91 ± 15.79
473	−131.25 ± 6.01	−127.15 ± 16.23	−132.34 ± 22.77	−142.69 ± 18.75
523	−132.31 ± 7.99	−139.98 ± 15.49	−134.95 ± 12.80	−162.08 ± 38.06

Table 4: Temperature-dependent Seebeck coefficient values of Mn-doped Bi₂Te₃ thin films. Seebeck coefficient values are reported as mean ± standard deviation (SD) for films deposited at different Mn target powers. Negative Seebeck coefficient values indicate n-type conduction behavior.

All films exhibited semiconducting transport behavior (Figure 8), where ρ decreased with increasing temperature^47,48. Notably, the 5 W film exhibited the lowest ρ, reaching 37.02 µΩ·m at 523 K, whereas higher Mn input resulted in increased resistivity. The 15 W film showed the highest ρ value of 87.40 µΩ·m at 523 K. These results suggest that low Mn incorporation (5 W) reduced ρ most effectively, whereas further Mn addition increased resistivity, indicating that excessive Mn-related disorder offset the conductivity improvement observed at lower Mn input⁴⁹. The structural evidence from XRD analysis, including peak shifting and broadening (Figure 5 and Table 2), together with the increased Mn incorporation measured by EDX (Table 3), suggests that higher Mn input introduced additional disorder and scattering centers that may reduce carrier mobility^18,50. In addition, the reduced crystallite size and increased ε and δ values at higher Mn power (Table 2) are consistent with an increased density of grain boundaries and lattice defects that can intensify carrier scattering^51,52. These same structural features may also enhance phonon scattering, which could be favorable for thermal transport reduction, indicating a trade-off that influences PF at higher Mn input^53,54.

Figure 8. Temperature dependence of the electrical resistivity of Mn-doped Bi₂Te₃ thin films. Electrical resistivity (ρ) as a function of temperature for films deposited at Mn target powers of 0, 5, 10, and 15 W. The 5 W film exhibited the lowest electrical resistivity over the measured temperature range. Please click here to view a larger version of this figure.

The combined transport response is reflected in the PF behavior shown in Figure 9. The maximum PF was achieved for the 5 W film, reaching approximately 529.33 µW·m⁻¹·K⁻² at 523 K, indicating that PF enhancement in this series was primarily associated with achieving sufficiently low ρ while maintaining adequate S³³. Although the 15 W film exhibited the highest magnitude of S value, the concurrent increase in ρ reduced the PF to 300.59 µW·m⁻¹·K⁻², highlighting the trade-off between thermopower and electrical transport that governs TE optimization⁵⁵.

Figure 9. Temperature dependence of the power factor (PF) of Mn-doped Bi₂Te₃ thin films. PF as a function of temperature for films deposited at Mn target powers of 0, 5, 10, and 15 W. PF values were calculated using , where S is the Seebeck coefficient and ρ is the electrical resistivity. The 5 W film exhibited the highest PF within the measured temperature range. Please click here to view a larger version of this figure.

Together, these results demonstrate that Mn co-sputtering preserved the Bi₂Te₃ phase while producing Mn-input-dependent changes in structure and TE transport behavior. The highest Seebeck coefficient magnitude was observed at 15 W Mn input, whereas the maximum PF occurred at 5 W, indicating that PF enhancement in this series was more strongly associated with resistivity reduction than with Seebeck coefficient enhancement^7,56.

Data Availability:
Raw data supporting the findings of this study are provided as Supplementary Tables 1–3, including: (1) XRD-derived structural calculations and crystallographic parameters, (2) normalized EDX compositional analysis and atomic percentage calculations, and (3) temperature-dependent thermoelectric transport data used for Figures 4–9 and Tables 1–4.

Supplementary Table 1. Structural parameters of Mn-doped Bi₂Te₃ thin films derived from X-ray diffraction (XRD) analysis of the dominant (015) reflection. The table includes the full width at half maximum (FWHM, β), diffraction peak position (2θ), Bragg angle (θ), crystallite size (D), microstrain (ε), and dislocation density (δ) calculated using the Scherrer method and related structural equations. Crystallite size was calculated using D = Kλ/[βcos θ] , where K = 0.89 and λ = 1.5406 Å. Microstrain and dislocation density were calculated using and ε = β/4tan θ and δ = 1/D² , respectively. Please click here to download this file.

Supplementary Table 2. Normalized EDX composition analysis of Mn-doped Bi₂Te₃ thin films deposited at different Mn RF target powers. (A) Normalized Bi–Te–Mn weight percentage composition calculated after excluding background contributions and normalizing the elemental composition according to:

(B) Atomic percentage composition calculated from the normalized elemental weight percentages after excluding carbon (C) and oxygen (O) signals using:

where w_i is the elemental weight percentage and M_i is the atomic mass of element . The measured Mn content increased progressively with increasing Mn RF power, confirming Mn incorporation across the deposition series. Please click here to download this file.

Supplementary Table 3. Temperature-dependent TE transport properties of Mn-doped Bi₂Te₃ thin films deposited at different Mn RF target powers. The table includes the measured Seebeck coefficient (S), electrical resistivity (ρ), and calculated PF as a function of temperature. Temperature conversion from Celsius to Kelvin was performed using:

T(K) = T(º C) + 273.15

The PF value was calculated using:

where S is the Seebeck coefficient and is the electrical resistivity. All measured Seebeck coefficients were negative, confirming n-type conduction behavior across the Mn-doped Bi₂Te₃ thin-film series. Please click here to download this file.

Discussion

This study demonstrated an RF co-sputtering workflow for tailoring the TE behavior of Bi₂Te₃ thin films by controlling Mn RF input while maintaining a constant Bi₂Te₃ target power. XRD confirmed the presence of the rhombohedral Bi₂Te₃ phase (JCPDS No. 15-0863) with a preferred (015) orientation across all samples. The systematic shift of the (015) reflection toward higher 2θ values is consistent with Mn-related lattice contraction and possible Mn incorporation within the XRD detection limits, while the accompanying peak broadening, increased ε value, and increased δ value shown in Table 2 support increased lattice distortion and defect density at higher Mn input^33,57. FESEM observations further revealed a qualitative evolution toward more compact and interconnected grain networks at low Mn input, providing microstructural context for the observed transport-property trends²⁰.

From a broader perspective, the proposed method provides a laboratory-scale dopant-screening framework that links Mn input to structure–transport relationships under comparable processing conditions. The transport results demonstrate a clear trade-off between the S and ρ values⁵⁸. In this Mn series, the highest Seebeck coefficient magnitude was observed at higher Mn input (15 W), whereas the highest PF occurred at lower Mn input (5 W), where reduced ρ compensated for the more moderate S value⁵⁹. This behavior illustrates why the highest S and highest PF do not necessarily occur under the same conditions and highlights the importance of interpreting PF trends in relation to both carrier transport and microstructure⁵⁹. Although Bi₂Te₃ is commonly associated with near-room-temperature applications, the 5 W film continued to show increasing TE performance with temperature, and the PF advantage became more pronounced at 523 K because of the stronger reduction in ρ (Figure 7, Figure 8, Figure 9; Table 4).

However, several study limitations should be acknowledged. First, Mn target power represents an indirect control parameter because Mn incorporation depends on system-specific plasma characteristics and target behavior; therefore, compositional calibration remains necessary for meaningful cross-laboratory comparison⁶⁰. Second, the detailed Mn incorporation mechanism, including lattice site occupancy, valence state, and depth distribution, could not be established because Hall measurements, X-ray photoelectron spectroscopy (XPS)- and transmission electron microscopy (TEM)-based compositional profiling were not available in the present study. In addition to RF co-sputtering, alternative approaches such as pulsed laser deposition, molecular beam epitaxy, or post-deposition diffusion doping may also be used to investigate Mn-induced transport modification in Bi₂Te₃ thin films. Finally, in-plane thermal conductivity could not be measured because of instrumentation limitations; therefore, TE evaluation in the present work is limited to PF analysis, and the dimensionless figure of merit (zT) could not be determined⁶⁰. Future carrier-transport analysis together with XPS- and TEM-based chemical profiling would strengthen the mechanistic interpretation and help explain why ρ trends dominate PF behavior under specific Mn-input conditions⁶¹.

This protocol provides a laboratory-scale screening route for TE thin films intended for applications such as micro-TE generators, on-chip cooling, and low-grade heat harvesting⁶². The identified Mn-input window that preserved phase formation while enabling transport optimization may provide a practical starting point for future device integration^23,63. Benchmarking against previously reported n-type Bi₂Te₃ thin films indicates that there remains substantial room for further performance improvement. For example, a defect-engineered sputtered n-type Bi₂Te₃ thin film previously achieved a PF of 6.66 mW·m⁻¹·K⁻² at room temperature⁷, while review articles and high-performance thin-film studies have reported PF values approaching ~5 mW·m⁻¹·K⁻² when defect populations and carrier transport are carefully controlled²¹. Because these studies differ in film thickness, substrate type, crystallographic orientation, operating temperature, and defect-engineering strategy, the comparison is intended as contextual rather than direct. Nevertheless, the comparison suggests that further improvement in the present Mn-doped series will likely require enhanced texture control, carrier engineering, and optimization of defect populations. In addition, post-deposition annealing may help distinguish kinetic effects from compositional effects and further optimize transport behavior^64,65.

As future work, the optimized n-type Mn-doped Bi₂Te₃ thin films may be integrated into complete TE generator modules, followed by device-level current–voltage–power characterization under applied temperature gradients to evaluate practical power output. Future studies may also compare non-annealed and annealed TEG modules, including annealing at 473 K, to examine how annealing influences electrical contact quality, interfacial stability, and overall TE performance during device operation⁶⁶.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Acetone (≥99.5%)	Merck	100012	Substrate cleaning solvent
Aluminum foil	Pospaperroll	86871915	Chamber lining and substrate-holder covering
Argon gas (Ar)	Gaslink Industrial Gases Sdn. Bhd.	N/A (supplied by distributor)	Sputtering working gas; 4 sccm flow; grade 5.0 (99.999%)
Bismuth telluride sputtering target, 50.8 mm diameter, 4.25 mm thickness, 99.999%	Changsha Xinkang Advanced Materials Co., Ltd.	xk-Bi2Te3	Bi2Te3 target; fixed RF power = 100 W
Deionized water	In-house	N/A	Substrate rinsing and ultrasonic cleaning
Energy-dispersive X-ray spectroscopy detector	Integrated with FESEM system	N/A	Thin-film compositional analysis
Field-emission scanning electron microscope	Carl Zeiss	SUPRA 55VP	Surface microstructure imaging
Glass substrates, soda-lime glass, 1.1 mm thickness	Mevid	B105-2002	Substrate for thin-film deposition
Hot plate	IKA	C-MAG HS 7	Substrate drying at 393 K
Isopropyl alcohol (≥99.8%)	Merck	109634	Substrate cleaning solvent
Laboratory detergent	Merck	107553	Initial substrate cleaning
Coolant circulation unit	Fisher Scientific	250LCU	Cooling circulation for sputter guns
Methanol (≥99.9%)	Merck	106009	Substrate cleaning solvent
Manganese sputtering target, 50.8 mm diameter, 5.08 mm thickness, 99.999%	Hunan Boyu Technology Co., Ltd.	N/A	Mn target; RF power varied from 0 to 15 W
Multimeter	RS PRO	EN61010-1 CATIII	Electrical isolation verification
Nitrogen gas (N2)	Gaslink Industrial Gases Sdn. Bhd.	N/A (supplied by distributor)	Substrate drying after cleaning; grade 4.0 (99.99%)
Petri dish lid	Merck	41121812	Chamber viewing-window protection; approximately 94 mm diameter × 16 mm height
RF magnetron sputtering system	Custom-built	N/A	Co-sputtering of Bi2Te3 and Mn thin films
Seebeck/resistivity measurement system with platinum adapter	Linseis	LSR-3 (LSR L31)	Seebeck coefficient and electrical resistivity measurements
Tweezers	Fisher Scientific	10-316B	Handling cleaned substrates
Ultrasonic bath	Elma	ElmaSonic S30H	Solvent cleaning for methanol, acetone, isopropyl alcohol, and DI water
UV–ozone cleaner	Novascan Technologies	PSDP-UV4	Substrate surface treatment
Vacuum storage container	AS ONE Corporation	VM	Sample storage after deposition
Stylus profilometer	Bruker	DektakXT	Thin-film thickness measurement
X-ray diffractometer	Bruker	D8 ADVANCE	θ–2θ XRD scans using Cu Kα radiation
DIFFRAC.EVA	Bruker	DIFFRAC.EVA version 8	Peak fitting, diffraction analysis, and graph plotting
Origin	OriginLab	OriginPro (64-bit)	Peak fitting and graph plotting