Spindle electrical signal-based evaluation of lubrication performance in milling with micro-textured tools under different lubrication modes

Abstract

Micro-textured tools and advanced lubrication can reduce friction and improve milling stability of difficult-to-machine superalloys. This study quantified the effects of micro-texture geometry and lubrication mode on dynamic cutting load in GH4169 milling. Circular micro-textured tools with different spacing (Ls), diameter (D), and edge distance (Le) were tested under dry milling, MoS₂, cold-air minimum quantity lubrication (CMQL), and supercritical CO₂ minimum quantity lubrication (scCO₂MQL). Three-channel spindle electrical signals were analyzed by short-time Fourier transform, and total time–frequency energy, tooth-passing band energy, and spectral entropy were extracted. CMQL showed the best performance, reducing average total time–frequency energy by 9.57%, 7.74%, and 3.65% compared with dry milling, MoS₂, and scCO₂MQL, respectively. The minimum energy under optimal CMQL reached 5842.62, with spectral entropy of 2.9937. The dominant texture factor varied with lubrication mode, confirming the applicability of spindle electrical signals for lubrication evaluation.

Keywords

cutting load micro-textured tools spindle electrical signals short-time Fourier transform solid lubrication minimum quantity lubrication

Introduction

High-temperature alloys are a class of austenitic metals with high alloying content based on iron, nickel, or cobalt, capable of long-term service under temperatures exceeding 600 °C and high-stress conditions. Due to their excellent high-temperature strength, creep resistance, thermal stability, and fatigue performance,^1,2 these materials are widely used in aerospace engines, gas turbines, nuclear power equipment, and petrochemical applications that demand exceptional thermal stability.³ However, their high strength and low thermal conductivity lead to accelerated tool wear, increased cutting forces, and degraded surface quality during machining, making high-temperature alloys typical difficult-to-machine materials.⁴ Therefore, effectively reducing tool wear, minimizing cutting forces, and improving milling performance are critical research objectives in precision machining of these alloys.⁵

In recent years, micro-textured cutting tools have attracted considerable attention in machining due to their unique surface structures.^6,7 Appropriately designed micro-textures on the tool's rake face can significantly modify the tool–chip friction state and enhance surface wettability, thereby improving tool wear resistance and cutting stability.^8,9 Previous studies have shown that micro-textures can shorten the tool–chip contact length, regulate friction distribution, and mitigate adhesive wear, achieving both friction reduction and load reduction.^10,11 Moreover, micro-textures can serve as lubricant reservoirs, extending the oil film lifetime under minimum quantity lubrication (MQL) conditions and further enhancing lubrication efficiency.^12,13 When solid lubricants such as MoS₂ are embedded into micro-textured grooves, self-lubricating tools can be realized, maintaining excellent wear resistance even in the absence of external cooling media.^14,15

Cooling and lubrication strategies play a crucial role in cutting friction and load variation during high-temperature alloy machining. Conventional dry milling often results in excessive cutting-zone temperatures and severe tool wear due to the lack of cooling medium. In conventional flood cooling, cutting fluids may not adequately penetrate the cutting zone, limiting cooling effectiveness. Minimum quantity lubrication (MQL) has emerged as an efficient and environmentally friendly alternative, combining cooling and lubrication functions. By atomizing a small amount of lubricant with compressed air and delivering it to the cutting zone, MQL significantly improves interfacial cooling and lubrication.^16–21 Further enhancements can be achieved using low-temperature media such as cold air or liquid nitrogen to reduce cutting-zone temperatures. Supercritical CO₂ MQL (scCO₂MQL) leverages the high specific heat and excellent thermal conductivity of CO₂ to achieve effective cooling and lubrication with minimal lubricant supply.^22,23

Although existing studies have shown that micro-textured cutting tools can improve the machining performance of hard-to-cut materials by reducing tool–chip friction, cutting force, tool wear, and interfacial contact length,²⁴ and that advanced cooling and lubrication strategies such as minimum quantity lubrication (MQL), cold-air-assisted MQL, and supercritical CO₂-assisted MQL can further improve cutting stability and reduce thermal–mechanical loads,²⁵ the coupled influence of micro-texture geometry and lubrication condition on dynamic load variation has not been sufficiently quantified. In particular, most previous studies have focused on cutting force, tool wear, surface roughness, or temperature, while limited attention has been paid to evaluating lubrication performance through spindle electrical signal characteristics. Therefore, the novelty of this work lies in establishing a spindle electrical signal-based evaluation method to quantify the effects of lubrication mode and micro-texture geometry on cutting load level and milling stability. In this study, milling experiments with circular micro-textured tools were conducted under dry milling, MoS₂ solid lubrication, cold-air minimum quantity lubrication (CMQL), and supercritical CO₂ minimum quantity lubrication (scCO₂MQL) conditions. Short-time Fourier transform (STFT) was used to extract total time–frequency energy, tooth-passing frequency band energy, and spectral entropy from spindle electrical signals. The effects of texture spacing (L_s), texture diameter (D), and texture edge distance (L_e) were quantitatively compared, and the lubrication mechanisms were further discussed from the perspectives of texture geometry, lubricant transport, and boundary film stability.

Experimental

Experimental platform for multi-lubrication milling

To investigate the influence of different lubrication methods on the milling performance of micro-textured tools when machining GH4169 superalloy, a three-axis vertical milling experimental platform capable of operating under multiple lubrication modes was established. GH4169 superalloy was selected as a representative difficult-to-machine nickel-based superalloy because its high strength, low thermal conductivity, and tendency toward tool–chip adhesion readily induce cutting load fluctuations and tool wear during milling. The workpiece material was GH4169 superalloy, with dimensions of 140 mm × 80 mm × 20 mm. The workpiece dimensions were kept constant to ensure consistent tool engagement length, clamping conditions, and signal acquisition duration among different tests. Climb milling was adopted as the machining strategy, and the tool path is illustrated in Figure 1.

Figure 1.

Schematic of the milling process and tool path.

Four lubrication strategies were applied: dry milling, MoS₂ solid lubrication, cold-air minimum quantity lubrication (CMQL), and supercritical CO₂ minimum quantity lubrication (scCO₂MQL). The experimental setup and measurement system are shown in Figure 2. The main milling parameters are summarized in Table 1. These parameters were selected with reference to the applicable operating range of the cemented carbide milling cutter, previous milling studies on nickel-based superalloys, and preliminary tests aimed at obtaining a stable cutting segment for spindle electrical signal analysis. Before each test, the workpiece surface was cleaned, the tool and workpiece clamping conditions were checked, and the lubrication delivery system was stabilized. For MoS₂ solid lubrication, the micro-textures were filled before machining.

Figure 2.

Milling experiment setup and associated testing equipment.

Table 1.

Milling experiment parameters.

Cutting speed V_c	Feed per tooth fz	Axial depth of cut ap	Radial depth of cut ae
38 m/min	0.15 mm/z	0.25 mm	7 mm

Source: Authors own work.

In this study, milling performance was evaluated mainly using spindle electrical signal features, including total time–frequency energy, tooth-passing frequency band energy, spectral entropy, and STFT spectral continuity, which characterize cutting load level and process stability under different micro-texture and lubrication conditions.

Micro-textured tool fabrication and parameter design

The experiments employed cemented carbide face milling cutters, on which circular micro-textures were fabricated on the rake face using laser processing technology. The fabrication procedure is illustrated in Figure 3, and the geometric parameters of the micro-textures are shown in Figure 4. Three micro-texture geometric parameters were considered: spacing (L_s), diameter (D), and edge distance (L_e). The parameter ranges were determined based on previous studies on micro-textured cutting tools, laser fabrication feasibility, and preliminary machining considerations. Specifically, the texture spacing was selected to maintain sufficient lubricant storage while avoiding excessive texture density, which may weaken the rake face and disturb chip flow. The texture diameter and edge distance were selected to ensure effective lubricant retention and accessibility to the tool–chip contact region while maintaining the strength of the cutting edge.

Figure 3.

Process flow for fabricating micro-textures on a cemented carbide disc milling cutter.

Figure 4.

Schematic diagram of the micro-textured tool geometry parameters.

A total of 13 unique micro-texture parameter combinations were designed, as summarized in Table 2. Under four lubrication conditions—dry milling, MoS₂ solid lubrication, CMQL, and scCO₂MQL—the effects of L_s, D, and L_e on milling performance were investigated through single-factor analysis. For the statistical analysis of each geometric factor, the baseline parameter combination was included in the spacing, diameter, and edge-distance series, resulting in 15 grouped datasets under each lubrication condition. For simplicity, these parameters are hereafter referred to as L_s, D, and L_e throughout this study.

Table 2.

Micro-texture parameter combinations for the milling experiments.

Test	Spacing L_s (μm)	Diameter D (μm)	Edge distance L_e (μm)	Lubrication condition
1	130	65	180	Dry/MoS₂/CMQL/scCO₂MQL
2	140	65	180	Dry/MoS₂/CMQL/scCO₂MQL
3	150	65	180	Dry/MoS₂/CMQL/scCO₂MQL
4	160	65	180	Dry/MoS₂/CMQL/scCO₂MQL
5	170	65	180	Dry/MoS₂/CMQL/scCO₂MQL
6	160	55	180	Dry/MoS₂/CMQL/scCO₂MQL
7	160	60	180	Dry/MoS₂/CMQL/scCO₂MQL
8	160	70	180	Dry/MoS₂/CMQL/scCO₂MQL
9	160	75	180	Dry/MoS₂/CMQL/scCO₂MQL
10	160	65	120	Dry/MoS₂/CMQL/scCO₂MQL
11	160	65	140	Dry/MoS₂/CMQL/scCO₂MQL
12	160	65	160	Dry/MoS₂/CMQL/scCO₂MQL
13	160	65	200	Dry/MoS₂/CMQL/scCO₂MQL

Source: Authors own work.

Results and discussion

STFT-based analysis of spindle electrical signals

Spindle electrical signals can indirectly reflect variations in cutting load and are an important means for investigating the dynamic characteristics of milling forces. Since spindle current is linearly correlated with cutting torque, analysis of spindle current signals during the steady-state phase allows evaluation of tool load variations under different lubrication conditions.²⁶

Milling is a discontinuous cutting process. For each test, ten passes were performed along the pre-designed tool path, while three-channel synchronous spindle electrical signals were simultaneously recorded. A single-pass signal exhibits three channels of comparable magnitude, near-zero mean values, and pronounced periodic fluctuations, as shown in Figure 5(a), reflecting the spindle's electrical response during the milling process. Due to the large volume of information contained in the raw signals, direct assessment of cutting performance is not feasible. In this study, short-time Fourier transform (STFT) was employed to conduct time–frequency analysis of the three-channel synchronous spindle signals, and a composite signal was constructed to extract characteristic values, facilitating comparison of the effects of micro-texture parameters under different lubrication conditions.

Figure 5.

Short-time Fourier transform (STFT) analysis of a single cutting pass: (a) raw three-channel spindle electrical signals and the synthesized signal; (b) cutting segment identification based on moving RMS; (c) FFT of the detected cutting segment; (d) 3D STFT time–frequency representation of the detected cutting segment.

After preprocessing the three-channel spindle signals and constructing the composite signal, clear stage-dependent variations were observed in the complete recording, indicating coverage of the tool entry, steady-state cutting, and tool exit phases. The total duration of a single pass was approximately 134.716 s. Moving root-mean-square (RMS) analysis was used to track the local amplitude variation of the spindle electrical signal over time and to identify the stable cutting interval by distinguishing it from the transient tool-entry and tool-exit stages. This segmentation is important because the entry and exit stages contain non-steady engagement effects that may interfere with the extraction of representative cutting-load features. Based on the moving RMS result, the main cutting segment from 28.73 s to 107.23 s was identified, with a duration of approximately 78.51 s, as shown in Figure 5(b). Fast Fourier transform (FFT) analysis of this interval revealed distinct energy peaks near the theoretical tooth-passing frequency of 20.17 Hz and its harmonics (Figure 5(c)), demonstrating that spindle electrical signals can capture the periodic load characteristics associated with single-tooth milling.

Further STFT-based time–frequency analysis of the main cutting segment revealed the evolution of low-frequency bands and broadband energy over time, as illustrated in Figure 5(d). The prominent time–frequency ridges correspond to dominant frequency components with concentrated energy, while localized peaks and broadband elevations reflect stage-dependent load fluctuations and non-stationary characteristics during the cutting process.

Time–frequency analysis of spindle electrical signals under different lubrication conditions based on STFT

To investigate the load characteristics under different lubrication conditions, STFT-based time–frequency features of spindle electrical signals were compared for identical micro-texture parameters across various lubrication strategies. Figure 6 presents the three-dimensional STFT comparison of the spindle signals during the fifth pass under four lubrication conditions, including the original time–frequency maps, Z-score enhanced maps, and Demean enhanced maps.

Figure 6.

Comparison of 3D STFT maps of spindle electrical signals during the 5th cutting pass under four lubrication conditions: (a) original 0–100 Hz STFT maps; (b) Z-score-enhanced STFT maps; (c) demean enhanced STFT maps.

From the original time–frequency maps, it can be observed that the spindle signals under all four lubrication conditions exhibit pronounced low-frequency dominance, with primary energy concentrated around 20–30 Hz and neighboring bands. This indicates that the dominant periodic response of the cutting process remains generally consistent across different lubrication scenarios. The results suggest that changes in lubrication primarily affect the energy intensity, spectral continuity, and local perturbation levels of the dominant frequency band, rather than shifting its position.

Previous studies have highlighted that normalization of STFT spectra enhances the discernibility of local anomalies, improving contrast between different conditions.²⁷ In the Z-score enhanced maps, each frequency component is standardized, effectively mitigating the masking effect of strong background signals and making subtle fluctuations under different lubrication conditions more discernible.²⁸ The maps show that under CMQL and scCO₂MQL, the spectral surface is relatively smooth with weaker local undulations near the dominant frequency band. In contrast, under Dry and MoS₂ lubrication, frequent local peaks and troughs appear, indicating stronger time-varying perturbations despite the same dominant low-frequency background.

Demean enhancement removes the time-averaged background of each frequency component, providing a more direct representation of local deviations relative to the mean power level.²⁹ Under CMQL, local positive and negative deviations around the dominant frequency band are minor, and the spectral surface exhibits good overall continuity. This judgment was further supported by the lower total time–frequency energy and spectral entropy obtained under CMQL, indicating a more concentrated energy distribution and weaker non-stationary fluctuation. Dry and MoS₂ conditions show more frequent local protrusions and depressions, with alternating positive and negative deviations in the low-frequency region, reflecting stronger load fluctuations relative to the average background. The scCO₂MQL spectra exhibit intermediate characteristics, with overall stability superior to Dry and MoS₂, yet still presenting some localized deviations.

Comparison of STFT feature results for micro-texture parameters under different lubrication conditions

To further analyze the load characteristics associated with different micro-texture parameters, three evaluation metrics were selected: total time–frequency energy ( $E_{t t}$ ), tooth-passing frequency band energy ( $E_{f t}$ ), and spectral entropy ( $S E$ ). These metrics were used to quantify spindle electrical signals under various micro-texture configurations:

$E_{t t}$ represents the overall time–frequency energy of the spindle signal during the entire steady cutting period, serving as a comprehensive indicator of cutting load. Lower values indicate smaller cutting loads, calculated as :

E_{t t} = \sum_{n = 1}^{N} \sum_{k = 1}^{K} P (k, n)

(1)

where

P (k, n)

is the STFT power spectrum matrix, K and N denote the frequency and time indices, respectively.³⁰

$E_{f t}$ reflects the intensity of single-tooth periodic cutting loads in the spindle signal. Higher values indicate clearer periodicity and more stable cutting. The theoretical tooth-passing frequency is given by:

f_{t} = \frac{z n_{s}}{60}

(2)

where z is the number of teeth and

n_{s}

is the spindle speed in rpm. Considering a frequency band

Ω_{f t}

around

f_{t}

, the tooth-passing frequency band energy is defined as:

E_{f t} = \sum_{n = 1}^{N} \sum_{k \in Ω_{f t}} p (k, n)

(3)

where

Ω_{f t} = | f_{t} - Δ f, f_{t} + Δ f |

Δ f

denotes the bandwidth of the tooth-passing frequency band.³¹

$S E$ quantifies the concentration or dispersion of the time–frequency energy distribution, with smaller values indicating lower signal randomness and more stable cutting.³² First, the time-averaged spectrum is obtained :

\bar{P} (k) = \frac{1}{N} \sum_{n = 1}^{N} P (k, n)

(4)

Next, the normalized probability distribution is calculated:

p_{k} = \frac{\bar{P} (k)}{\sum_{k = 1}^{K} \bar{P} (k)}

(5)

Finally, the spectral entropy is defined as:

S E = - \sum_{k = 1}^{K} p_{k} \ln p_{k}

(6)

To reduce the influence of anomalous signals on the evaluation metrics, the results of ten milling passes were averaged for each micro-texture parameter configuration. The extracted feature values of spindle electrical signals under different micro-texture geometric conditions are presented in Figures 7–9.

Figure 7.

Single-factor effect of micro-texture spacing on spindle electrical signal features. (a) $E_{t t}$ , (b) $E_{f t}$ , (c) Spectral entropy.

Figure 8.

Single-factor effect of micro-texture diameter on spindle electrical signal features. (a) $E_{t t}$ , (b) $E_{f t}$ , (c) Spectral entropy.

Figure 9.

Single-factor effect of micro-texture edge distance on spindle electrical signal features. (a) $E_{t t}$ , (b) $E_{f t}$ , (c) Spectral entropy.

By extracting the spindle electrical signal feature values from the single-factor experiments with different micro-texture spacing (L_s), the influence of Ls on $E_{t t}$ , $E_{f t}$ , and $S E$ under various lubrication conditions is shown in Figure 7.

Under CMQL, $E_{t t}$ , $E_{f t}$ , and $S E$ all increase with increasing L_s, indicating higher tool load energy, more pronounced periodic response, and greater cutting fluctuations. Under Dry conditions, as L_s increases from 130 μm to 170 μm, $E_{t t}$ slightly decreases, reflecting reduced load fluctuations during cutting; $E_{f t}$ remains largely unchanged, indicating stable periodic cutting behavior; $S E$ decreases, suggesting lower cutting signal variability. Under MoS₂ lubrication, $E_{t t}$ initially increases and then decreases with increasing L_s, with a peak at L_s = 140 μm due to locally enhanced friction, followed by a reduction in fluctuation; $E_{f t}$ gradually decreases with L_s; $S E$ also decreases, reaching a minimum at L_s = 170 μm. Under scCO₂MQL, $E_{t t}$ slightly decreases, while both $E_{f t}$ and $S E$ decline, indicating reduced tool load energy, stable periodic behavior, and diminished cutting disturbances.

The above trend indicates that texture spacing affects the balance between lubricant storage, texture distribution density, and chip–tool contact stability. A smaller spacing can provide more frequent lubricant supply sites and enhance the continuity of the interfacial lubrication effect, whereas excessive spacing may reduce the effective coverage of the textured region. However, overly dense textures may also disturb chip flow or reduce the load-bearing area of the rake face. Therefore, the influence of L_s differs among lubrication conditions, depending on whether the dominant mechanism is boundary film stability, solid lubricant release, or lubricant infiltration.

By extracting the spindle electrical signal feature values from single-factor experiments with different micro-texture diameters (D), the influence of D on $E_{t t}$ , $E_{f t}$ , and $S E$ under various lubrication conditions is presented in Figure 8. As D increases, $E_{t t}$ , $E_{f t}$ , and $S E$ exhibit overall fluctuating trends.

Under CMQL, both $E_{t t}$ and $E_{f t}$ remain relatively low, with $S E$ also at a lower level, indicating small cutting loads, strong periodicity in single-tooth load, and a stable milling process. Under MoS₂ lubrication, a peak is observed around D = 65 μm, reflecting concentrated frictional load and larger fluctuations in periodic cutting behavior. The Dry and scCO₂MQL conditions show smaller amplitude fluctuations, with values slightly higher than CMQL, suggesting greater cutting disturbances and more dispersed signal energy.

The variation with texture diameter reflects the competition between lubricant retention capacity and local structural disturbance on the rake face. Increasing D can enlarge the lubricant storage volume and improve the possibility of lubricant participation in the tool–chip contact region. However, an excessively large texture diameter may reduce the effective load-bearing area and intensify local stress or chip-flow disturbance. This explains why the signal features do not change monotonically with D under all lubrication conditions, especially under MoS₂ lubrication, where the filling, release, and depletion of the solid lubricant can cause more pronounced fluctuations.

By extracting the spindle electrical signal feature values from single-factor experiments with different edge distances (L_e), the influence of L_e on $E_{t t}$ , $E_{f t}$ , and $S E$ under various lubrication conditions is shown in Figure 9. As L_e increases, $E_{t t}$ under Dry and MoS₂ lubrication exhibits a decreasing trend, while $E_{f t}$ and $S E$ remain relatively stable under CMQL and scCO₂MQL.

Specifically, under CMQL, $S E$ shows smooth fluctuations and the cutting process remains stable, with lower variability compared to the other three lubrication conditions. Under MoS₂ lubrication, $E_{t t}$ and $S E$ slightly decrease in the L_e range of 160–180 μm, indicating more pronounced cutting load fluctuations.

The edge distance L_e determines the relative position between the textured region and the main tool–chip contact zone. When the textures are too far from the cutting edge, the stored lubricant may not effectively enter the high-friction contact region. In contrast, when the textures are too close to the cutting edge, cutting-edge strength and chip flow stability may be affected. Therefore, an appropriate L_e is required to balance lubricant accessibility and edge integrity. The relatively smoother variation under CMQL indicates that the externally supplied low-temperature lubricant helps maintain a more stable boundary film, reducing the sensitivity of the signal features to changes in L_e.

To quantify the influence of the three micro-texture geometric parameters on cutting load, the range analysis method was applied to the spindle electrical signal feature values. The range of each parameter was used as a sensitivity indicator, and the optimal micro-texture parameter combinations under each lubrication condition were identified, as shown in Figure 10.

Figure 10.

Sensitivity analysis of micro-texture parameters under different lubrication conditions.

For CMQL, the influence ranking of the parameters was L_s > D > L_e, and the optimal combination was determined as L_s = 130 μm, D = 65 μm, and L_e = 180 μm, with a total energy of $E_{t t}$ = 5843, indicating the best lubrication performance. Under Dry milling, the parameter sensitivity followed the order D >L_e > L_s, with the optimal combination of L_s = 160 μm, D = 60 μm, and L_e = 180 μm, resulting in a higher total energy ( $E_{t t}$ = 7110), reflecting larger cutting loads and stronger friction. For MoS₂ lubrication, the influence ranking was L_e > L_s > D, with the optimal combination L_s = 160 μm, D = 65 μm, and L_e = 180 μm, suggesting that the edge distance had a dominant effect on MoS₂ lubrication, leading to noticeable fluctuations in the periodic cutting load. Under scCO₂MQL, the parameter sensitivity followed the order L_s>  L_e > D, with the optimal combination of L_s= 160 μm, D = 65 μm, and L_e = 180 μm, which produced moderate cutting loads. These results demonstrate that the sensitivity of cutting performance to micro-texture parameters varies with lubrication strategy, providing guidance for selecting tool geometries that optimize lubrication efficiency and milling stability.

Analysis of lubrication mechanisms for micro-textured tools under different lubrication conditions

To further elucidate the differences in spindle electrical signal characteristics under various lubrication conditions, the lubrication behavior on the surfaces of micro-textured tools was analyzed to reveal the underlying mechanisms. Figures 11 and 12 schematically illustrate the lubrication evolution of micro-textured tools filled with solid lubricants and the formation and infiltration mechanisms of oil films under CMQL and scCO₂MQL, respectively. These figures reveal the potential action pathways of different lubrication strategies, including solid lubricant storage and depletion, boundary film formation, capillary-driven transport of lubricants within the micro-textures, and the behavior of the oil film under low-temperature conditions. The overall lubrication performance of micro-textured tools is determined jointly by the geometric parameters of the textures, the type of lubricant, temperature, lubricant transport capability, and the continuity of the boundary film.

Figure 11.

Evolution mechanism of solid lubricant storage, loss, and boundary lubrication on the surface of a micro-textured tool.

Figure 12.

Lubricant infiltration and boundary film formation mechanisms of a micro-textured tool under CMQL and scCO₂MQL conditions.

For micro-textured tools filled with MoS₂, the textures initially act as reservoirs that store and gradually release the solid lubricant, reducing direct contact and friction between the tool and chip during the early stage of cutting. This behavior is mainly attributed to the lamellar crystal structure and low shear strength of MoS₂. During cutting, the filled micro-textures retain solid lubricant within the grooves and release it gradually under the combined action of chip contact pressure, interfacial friction, and local temperature rise. The released MoS₂ can form a transfer film on the tool–chip interface, reducing direct metal-to-metal contact and lowering interfacial shear resistance. However, as cutting proceeds, the stored lubricant is progressively consumed or removed by chip flow, which weakens the continuity of the lubricating film and may lead to increased load fluctuation. As the cutting progresses, the surface solid lubricant is gradually depleted, and the tool surface transitions from the initially filled state to partial boundary lubrication and eventually to a low-lubricant state, as illustrated in Figure 11. The lubrication effect of MoS₂-filled micro-textures exhibits distinct stage-dependent behavior. Once the solid lubricant diminishes and no stable liquid boundary film is present, friction fluctuations can increase again, consistent with the observed local fluctuations in the electrical signals under MoS₂ lubrication.

In contrast, scCO₂MQL demonstrates strong lubricant transport capability. Supercritical CO₂ facilitates the dissolution, diffusion, and redistribution of the lubricant within the micro-textures and at the tool–chip interface, enhancing oil penetration between the micro-textures. The primary advantage of CMQL lies in its low-temperature environment, which helps suppress interface temperature rise, delays lubricant film rupture, and maintains adsorption of polar groups on the metal surface, thereby promoting the formation of a continuous and stable boundary lubrication film. In general, scCO₂MQL favors enhanced infiltration and flow of the lubricant, whereas CMQL emphasizes the stability and continuity of the boundary film. The difference between CMQL and scCO₂MQL also indicates that the lubrication effect of micro-textured tools is governed by the coupling between texture geometry and lubricant transport. Texture spacing and diameter affect lubricant storage capacity and replenishment frequency, while edge distance determines whether the stored or externally supplied lubricant can effectively reach the main tool–chip contact zone. Under CMQL, the low-temperature lubricant supply helps maintain the continuity of the boundary film, whereas under scCO₂MQL, the enhanced fluid penetration promotes lubricant transport within and between the micro-textures. Therefore, the final lubrication performance depends on the balance among lubricant storage, transport, film formation, and film stability.

The spindle electrical signal analysis corroborates these observations. Under CMQL, lower signal energy and improved cutting stability were observed. The stable boundary film formed under CMQL effectively reduces friction fluctuations and load variations at the tool–chip interface, resulting in superior overall performance as reflected in the electrical signal characteristics.

Compared with previous studies that mainly evaluated micro-textured tools and cooling and lubrication strategies using cutting force, tool wear, surface roughness, or temperature, the present work provides an electrical-signal-based evaluation of lubrication performance during milling. The STFT feature results obtained from repeated milling passes show that CMQL produced the lowest average total time–frequency energy and spectral entropy among the four lubrication strategies, which supports the observation of smoother spectral surfaces and weaker load fluctuation. These experimental signal features are consistent with the proposed lubrication mechanism, indicating that the stability and continuity of the boundary lubrication film play an important role in reducing cutting load variation. Therefore, spindle electrical signal analysis can be used as a complementary method for evaluating the effectiveness of micro-textured tools and lubrication strategies in difficult-to-machine alloy milling.

Conclusions

STFT analysis of spindle electrical signals can effectively extract the low-frequency dominant features and non-stationary perturbations during milling. $E_{t t}$ reflects the overall cutting load level during the steady cutting stage, $E_{f t}$ represents the intensity of single-tooth periodic cutting action, and spectral entropy ( $S E$ ) characterizes the concentration of time–frequency energy distribution and process stability. Together, these three metrics constitute a comprehensive electrical signal feature system for evaluating the lubrication performance of micro-textured tools.

The sensitivity of micro-texture geometric parameters varies under different lubrication conditions. Under CMQL, the optimal combination was L_s = 130 μm, D = 65 μm, and L_e = 180 μm, with the parameter sensitivity ranking of L_s > D > L_e, indicating that micro-texture spacing has the most significant influence on lubrication performance. Under Dry, MoS₂, and scCO₂MQL, the sensitivity rankings were D > L_e >L_s, L_e > L_s > D, and L_s >L_e > D, respectively, further demonstrating the lubrication-dependent effects of micro-texture parameters.

Comparative analysis of the original STFT maps and enhanced three-dimensional time–frequency maps indicates that different lubrication strategies do not alter the dominant low-frequency structure of the spindle signals, but they significantly affect the strength of local time-varying perturbations and spectral continuity. Under CMQL, the spectral surfaces are the smoothest with minimal local deviations, reflecting the most stable cutting process. In contrast, Dry and MoS₂ conditions exhibit rougher spectra and stronger load fluctuations, while scCO₂MQL demonstrates intermediate behavior.

Regarding the lubrication mechanisms, scCO₂MQL provides strong potential for lubricant infiltration and transport, whereas CMQL is more effective in maintaining boundary film stability and continuity. Combined with the electrical signal results, it can be concluded that the persistence and stability of the boundary lubrication film are key factors influencing spindle load and cutting stability, which explains why CMQL ultimately demonstrates superior overall milling performance. These results indicate that spindle electrical signal analysis can provide a practical basis for evaluating lubrication states, selecting micro-texture parameters, and monitoring cutting stability in the milling of difficult-to-machine superalloys. The proposed evaluation method may also support the development of non-intrusive process monitoring strategies for micro-textured tool machining.

Footnotes

Acknowledgements

The authors gratefully acknowledge the support provided by Harbin University of Science and Technology for conducting this research.

ORCID iDs

Yuhua Zhang

Zhenhua Gao

Mingyang Yuan

Ethical considerations

This study does not involve human participants, human data, or animals, and therefore no ethical approval was required.

Author contributions

Y.Z. conceived the study and designed the experiments. Z.G. and K.L. conducted the milling tests and collected the spindle signal data. M.Y. performed the STFT and statistical analyses. Y.Z. and Z.G. wrote the manuscript, and all authors reviewed and approved the final version.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Open Project Program of the Key Laboratory for Cross-Scale Micro and Nano Manufacturing, Ministry of Education, Changchun University of Science and Technology (CMNM-KF202407).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

References

. Study on the mechanism of low-temperature machining of high-temperature alloys and tool life extension strategies. Master’s Thesis, Shandong University of Technology, China, 2023.

Pedroso

AFV

Sebbe

NPV

Costa

RDFS

, et al. INCONEL® alloy machining and tool wear finite element analysis assessment: an extended review. J Manuf Mater Process 2024; 8: 37.

Yao

Cao

, et al. Effect of milling process parameters on surface integrity of GH4169D superalloy. Cemented Carbide 2024; 41: 477–486.

Sivaiah

Ajay Kumar

Singh

, et al. Effect of novel hybrid texture tool on turning process performance in MQL machining of inconel 718 superalloy. Mater Manuf Process 2020; 35: 61–71.

Tang

Liu

, et al. Preparation and cutting performance of nano-scaled Al2O3-coated micro-textured cutting tool prepared by atomic layer deposition. High Temp Mater Process 2021; 40: 77–86.

Patel

Liu

Shah

, et al. Effect of micro-textured tool parameters on forces, stresses, wear rate, and variable friction in titanium alloy machining. J Manuf Sci Eng 2020; 142: 021007.

Borgaonkar

Syed

. Eco-friendly sustainable machining with MoS2-based solid lubricant. Proc Inst Mech Eng Part J J Eng Tribol 2023; 237: 1783–1795.

Peng

Study on the friction and wear characteristics and wettability of bionic micro-textured cutting tools. Thesis, Yanshan University, China, 2024.

Fan

Deng

Gao

, et al. Cutting performance of micro-textured PCBN tool. Nanotechnol Precis Eng 2021; 4: 023004.

10.

Tong

Wang

. Analysis of wear vibration behavior of micro-textured coated cemented carbide considering high-order scale. Coatings 2024; 14: 791.

11.

Liu

, et al. Research on the effect of micro-pit parameters on tool wear in turning GH4169. Coatings 2025; 15: 543.

12.

Guo

Wang

Zhao

, et al. Recent advances in the fabrication technology and machining performance of micro-textured cutting tools. J Mech Eng 2021; 57: 172–200.

13.

Zhang

Miao

, et al. Performances of concave and convex micro-texture tools in turning of Ti6Al4V with lubrication. Int J Adv Manuf Technol 2020; 109: 1071–1092.

14.

Zhou

Luo

Long

, et al. Cutting performance of MoS2-based solid lubricant micro-textured self-lubricating cutting tools. Mech Sci Technol Aerosp Eng 2017; 36: 1307–1312.

15.

Settineri

Faga

Lerga

. Properties and performances of innovative coated tools for turning inconel. Int J Mach Tools Manuf 2008; 48: 815–823.

16.

Gajrani

Suvin

Kailas

, et al. Machining of hard materials using textured tool with minimum quantity nano-green cutting fluid. CIRP J Manuf Sci Technol 2021; 35: 410–421.

17.

Jiang

Wang

, et al. Combining micro-textures and nanofluid electroosmotic flow for improving wear resistance of Si3N4 ceramic tools in machining of nickel-based superalloys. J Mater Res Technol 2025; 39: 1110–1132.

18.

Sheng

Liu

Yan

, et al. Research progress on the application of minimum quantity lubrication in machining. Manuf Technol Mach Tool 2025: 86–97.

19.

Kang

Kim

Shin

, et al. Effect of the minimum quantity lubrication in high-speed end-milling of AISI D2 cold-worked die steel (62 HRC) by coated carbide tools. Surf Coat Technol 2008; 202: 5621–5624.

20.

Chou

. Experimental evaluation of minimum quantity lubrication in near micro-milling. J Mater Process Technol 2010; 210: 2163–2170.

21.

Liao

Lin

Chen

. Feasibility study of the minimum quantity lubrication in high-speed end milling of NAK80 hardened steel by coated carbide tool. Int J Mach Tools Manuf 2007; 47: 1667–1676.

22.

Zhang

Study on the milling performance of TC11 under supercritical CO2 minimum quantity lubrication and edge geometry. Thesis, Harbin University of Science and Technology, China, 2022.

23.

Gui

Zhang

, et al. Hybrid cooling-lubrication in milling inconel 718: scCO2-assisted MQL and OoW performance. Mater Manuf Process 2025; 40: 1654–1668.

24.

Samant

Sahoo

Kumar

, et al. Performance improvement through textured cutting tool during machining hard-to-cut materials: a review. AIP Conf Proc 2024; 3007: 100044.

25.

Khatai

Sahoo

Kumar

, et al. Recent research progress on various cooling and lubrication techniques used in sustainable hard machining: a comprehensive review. Proc Inst Mech Eng Part E J Process Mech Eng 2024; 238: 3009–3053.

26.

, et al. Monitoring of milling force based on machine tool spindle current signal. Combined Mach Tools Autom Mach Technol 2022: 156–159.

27.

Manwar

Varghese

Bagri

, et al. Online tool condition monitoring in micromilling using LSTM. J Intell Manuf 2025; 36: 935–955.

28.

Chen

Hou

, et al. Feature extraction using dominant frequency bands and time-frequency image analysis for chatter detection in milling. Precis Eng 2019; 56: 235–245.

29.

Huang

Zhu

Lei

, et al. Tool wear monitoring with vibration signals based on short-time Fourier transform and deep convolutional neural network in milling. Math Probl Eng 2021; 2021: 9976939.

30.

Cohen

. Time-frequency analysis. Prentice Hall PTR, 1995.

31.

Navarro-Devia

Chen

Dao

, et al. Chatter detection in milling processes—a review on signal processing and condition classification. Int J Adv Manuf Technol 2023; 125: 3943–3980.

32.

Shannon

CE.

A mathematical theory of communication. Bell Syst Tech J 1948; 27: 379–423, 623–656.