- Email: [email protected]
Micro-channels machining on polycrystalline diamond by nanosecond laser
Optics and Laser Technology 108 (2018) 333–345
Contents lists available at ScienceDirect
Optics and Laser Technology journal homepage: www.elsevier.com/locate/optlastec
Full length article
Micro-channels machining on polycrystalline diamond by nanosecond laser Youqiang Xing a,b,c, Lei Liu a,⇑, Xiuqing Hao b,⇑, Ze Wu a, Peng Huang a, Xingsheng Wang d a
School of Mechanical Engineering, Southeast University, Nanjing 211189, Jiangsu Province, PR China Jiangsu Key Laboratory of Precision and Micro-Manufacturing Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, Jiangsu Province, PR China c Shaanxi Key Laboratory of Non-Traditional Machining, Xi’an Technological University, Xi’an 710021, Shaanxi Province, PR China d Department of Mechanical Engineering, Nanjing Agricultural University, Nanjing 210031, Jiangsu Province, PR China b
a r t i c l e
i n f o
Article history: Received 29 March 2018 Received in revised form 28 May 2018 Accepted 12 July 2018
Keywords: PCD Micro-channels Nanosecond laser texturing RSM
a b s t r a c t In this paper, micro-channels are fabricated on the polycrystalline diamond (PCD) surface by a nanosecond laser. The effects of laser process parameters (laser power, frequency and scanning speed) on the responses of surface quality, channel dimension, material remove rate (MRR) and surface roughness Ra are investigated by the single factor and multi-objective optimization tests. The mathematical models between the laser process parameters and responses are developed based on the response surface method (RSM) and the adequacy of the models for the responses is studied by the analysis of variance (ANOVA) method. Ultimately, the desired micro-channels are fabricated with the optimal process parameters: laser power of 14.29 W, frequency of 12.87 kHz and scanning speed of 11.36 mm/s. The results exhibit that the developed mathematical models are in well agreement with the experimental results. Ó 2018 Published by Elsevier Ltd.
1. Introduction Micro-textures have proven to be an effective method to improve the tribological behavior of the interfaces, and that are widely used in bearing, mechanical seals, cutting tools and cylinder liners [1–4]. Meanwhile, the hole and channel may be the most successful application of the texture patters on the tribo-surfaces, and the main mechanisms of the micro-textures are attributed to the entrapment of wear debris, increased load carrying capacity, storage of lubricants and reduced contact area [5–7]. Presently, the fabrication and application of surface textures mainly focus on the metals [8,9], ceramics [10,11], cemented carbide [12,13] and coatings [14,15], and few studies about the fabrication of micro-textures on polycrystalline diamond (PCD) are reported [16]. PCD is an excellent material for using as bearings and cutting tools in automotive, aerospace and machining industries with its unique properties of high hardness, good thermal conductivity, high wear resistance and low friction coefficient [17,18]. However, the high hardness and brittleness of the PCD make it difficult to fabricate the micro textures on its surface with the conventional methods. At present, micro-textures can be fabricated by various methods, including electrical discharge machining (EDM) [19,20], ⇑ Corresponding authors. E-mail addresses: [email protected] (L. Liu), [email protected] (X. Hao). https://doi.org/10.1016/j.optlastec.2018.07.024 0030-3992/Ó 2018 Published by Elsevier Ltd.
focused ion beam etching (FIB) [21], lithography technology [22], mechanical micro-machining [23,24], laser [25,26], etc. Among these methods, EDM and laser machining technology have been attracted more attentions on the machining of PCD in most studies. For examples, Wang et al. [27] investigated the micro-hole machining performance of PCD by micro-EDM technology, and the process parameters on surface quality and material remove rate of the micro-holes are studied, ultimately, the optimal machining parameters are selected. Hashikawa [28] fabricated the micro-holes on the PCD surface by laser machining first, and then finished by EDM technology for the high surface quality. Yan et al. [29] fabricated micro-textures on the PCD surface by micro-EDM with a rotary cupronickel electrode, the effect of machining parameters (discharge energy electrode and rotation rate) on the machining characteristics were clarified, and then the micro-channels with low surface roughness as well as high form accuracy were obtained. Su et al. [16] studied the influence of laser process parameters on the formation of micro-holes and micro-channels on the PCD surface by a fiber laser, and the dimensions and topography of the micro-textures were investigated. Furthermore, the optimal process parameters were obtained for the high surface quality of micro-textures. Comparing to EDM, laser machining technology exhibits more advantages for the fabrication of micro-textures such as high efficiency, no pollution, no mechanical damage and tool wear, therefore, it is a more efficient method for fabricating micro-textures on the PCD surface.
334
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Comparing to femtosecond and picosecond lasers, the nanosecond laser exhibits high machining efficiency and low costs. Thus, the nanosecond laser is widely used for the micromachining [30]. Unfortunately, the nanosecond laser micromachining produces more heat-affected zone (HAZ), recast layers and debris due to the long laser-material interaction time and high energy density [31]. Therefore, the laser process parameters need to be further optimized for the high surface quality. In this paper, a nanosecond laser is used to produce the microchannels on the PCD surface. Single factor test and multi-objective optimization are carried out to investigate the effect of laser process parameters on the surface quality, geometric dimensions and material remove rate (MRR) of the machined channels. Meanwhile, the predictive models of the geometric dimensions, MRR and surface roughness Ra are founded based on the response surface method (RSM), and the optimal process parameters are obtained for the desired channels. 2. Experimental details 2.1. Materials The commercial PCD compacts (Far-East New Materials Co., Ltd., China) are used as the test materials. The PCD layer is formed by sintered diamond grit with Co binder to WC cemented carbide substrate at high temperature and pressure. The thickness of the PCD layer is about 0.5 mm, and it is about 2.5 mm for cemented carbide substrate. The PCD layer comprises an average grain size of 5 lm, and the surface roughness Ra is 0.038 lm. Table 1 shows the properties of the cemented carbide substrate and PCD layer. Fig. 1 shows the typical microstructure and composition analysis of the PCD surface. It illustrates that the surface of the sample is dense and the porosity is virtually absent; EDX analyses of points A and B confirm that the gray regions are the PCD phases and the white regions are the Co binders.
sional stage. The wavelength of the laser is 1064 nm, pulse duration is 10 ns, maximum power is 20 W and the maximum frequency is 20 kHz. A single laser beam is focused on the sample surface with an incident angle of 90°, meanwhile, the preparation process is monitored by a digital microscope. All experiments are performed in air condition at room temperature. 2.2.2. Single factor test Single factor tests are used to investigate the effect of laser process parameters (laser power, scanning speed and frequency) on the surface morphology, channel width (horizontal distance at the top surface of the channel) and depth (vertical distance between the top and bottom surface of the channel), the material remove rate (MRR) and surface roughness (Ra) of the machined surface. The laser power ranges from 5 to 20 W, the frequency ranges from 2 to 20 kHz, and the scanning speed ranges from 1 to 30 mm/s. After laser irradiation, the samples are cleaned twice for 30 min in an ultrasonic bath by alcohol and then dried. The surface morphologies of the channels are observed by scanning electron microscope (SEM, QUANTA FEG250, USA) and three dimensional microscope (VHX-600E, Japan), and the surface compositions are detected by an energy dispersive X-ray spectroscope (EDX, X-MAX50, UK). The channel profile with a certain width (horizontal distance at the top surface of the channel) and depth (vertical distance between the top and bottom surface of the channel) are measured by a three dimensional microscope (VHX-600E, Japan). The cross-sectional area of the channel is also measured based on the section profile and then the material remove rate (MRR) is calculated by taking into account of the scanning speed. The surface roughness (Ra) of the machined surface, which represents the surface quality of the channels at a certain extent, is measured with the same area (including four channels in the area) by the white light interferometer (Wyko NT9300, USA). For reducing the measurement error, each test is repeated three times and the average values are recorded as the output.
2.2. Experimental methods 2.2.1. Experimental setup An Nd:YAG laser (DP-H20, China) is used to produce the microchannels on PCD sample surface that is fixed on a three dimen-
2.2.3. Multi-objective optimization To further investigate the effect of the process parameters and find the optimal values for getting the higher MRR and lower surface roughness Ra. An L20 array according to the CCD (central
Table 1 Properties of cemented carbide substrate and polycrystalline diamond layer. Materials
Density (gcm-3)
Elastic modulus (GPa)
Thermal conductivity (MPa)
Thermal expansion coefficient (10-6/K)
WC + 8 wt.%Co PCD
14.5–14.9 3.6–3.8
590–610 860–900
79.5 720.0
4.5–4.6 1.0–1.18
Fig. 1. SEM micrograph and EDX analysis of the PCD surface, (a) SEM micrograph of the polished surface, (b) and (c) EDX composition analyses marked in points A and B, respectively.
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
composite design) method based on the RSM is performed in the present study. Laser power, frequency and scanning speed are selected as the factors and 5 levels of each factor are used based on the pilot experiments. Table 2 shows the values of factors at different levels. 3. Results and discussion The surface morphologies, profile and EDX analyses of the heataffected zone of the micro-channels are shown in Fig. 2. As seen in this figure, the channels are formed with the laser ablation, the formation mechanism can be explained that the materials are melted with the high temperature induced by the absorbed laser energy, and then vaporized and ejected away from the machined zone with the high recoil pressure, resulting the material removal [32]. Meanwhile, the heat-affected are formed along the micro-channels induced by high energy ablation. The 3D view in Fig. 2 (a) demonstrates that the channel is discontinuous with the unsuitable laser process parameters due to the molten materials re-solidified at the bottoms. The profile of the channel seems to be a triangle with a certain width at the top surface and vertical depth between the top and bottom surface, shown in Fig. 2 (b). The 3D morphology in Fig. 2 (c) shows that the recast layers with large amounts of bulges around the edges of the channel are higher compared to the based plane surface, which are formed by the vapor blast ejec-
Table 2 Factors and levels. Factors
Symbol
Unit
Level 1
Level 2
Level 3
Level 4
Level 5
Power Frequency Scanning speed
A B C
W kHz mm/s
10 2 1
12 5 5
15 10 12
18 15 20
20 18 25
335
tion and rapid condensation of molten materials. The EDX analyses in Fig. 2 (d)-(f) shows that the oxygen content is higher surrounding the channel, indicating that the oxidation reaction occurs in heated-affected zone with the high intensity laser ablation. Therefore, to avoid more defects at the channel bottoms and the bulges in heated-affected zone, the laser process parameters are further investigated for the formation of the optimal channels. 3.1. Single factor test 3.1.1. Surface morphology Fig. 3 shows the SEM and 3D morphologies of micro-channels with different laser powers. As can be seen that the laser power has a profound effect on the formation of micro-channels. With the laser power of 5 W (Fig. 3 (a)), the irradiated zone becomes rough and a few materials are ejected, while quite shallow channels are formed because that the energy accumulated by lower laser power on the surface is not enough to produce the obvious channels. In the case of 12 W (Fig. 3 (b)), the energy accumulated on the unit area increases and more materials are melted and ejected from the ablation zone, therefore, a better quality channels with continuous bottoms and less recast layers surrounding the edges are produced. With the continuously increasing laser power of 18 W (Fig. 3 (c)), severe ablation occurs and large amounts of molten materials are re-solidified, which result in large amounts of recast layers formed on the sidewalls, bottoms and the brims zone of the channels. Fig. 4 shows the SEM and 3D morphologies of micro-channels with different frequencies. It exhibits that the laser power seems to have a milder effect on the surface quality of the channels compared to the laser power, and that has an important influence on the depth and width of the channels. The phenomenon dues to the differences of the peak power of a single pulse and the number pules deposited in the unit area for the given laser power and
Fig. 2. Surface morphologies, profile and EDX analyses of the heat-affected zone of the micro-channels.
336
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 3. SEM and 3D morphologies of micro-channels with different laser powers (a) 5 W, (b) 12 W, (c) 18 W under the frequency of 10 kHz and speed of 10 mm/s.
Fig. 4. SEM and 3D morphologies of micro-channels with different frequencies (a) 5 kHz, (b) 10 kHz, (c) 20 kHz under the power of 12 W and speed of 10 mm/s.
scanning speed. According to the results, the channels are continuous and few recast layers are formed at the channel bottoms and sidewalls. Fig. 5 shows the SEM and 3D morphologies of micro-channels with different scanning speeds. As shown in Fig. 5 (a), the channel bottom is continuous with the scanning speed of 3 mm/s; while the heat-affected zone surrounding the channel edges is large and some recast layers and debris can be seen on the sidewalls of the channels. As the scanning speed increases to 10 mm/s and 30 mm/s (Fig. 5 (b) and (c)), the recast layers and heat-affected zone reduce, and the channel bottoms are continuous; in addition, the width and depth of the channels have a significant change. It can be explained that with the increasing scanning speed, the interaction time between the laser and material surface reduces, and the energy accumulated on the unit area decreases, which result in less removed materials and less re-solidified materials forming the recast layers. 3.1.2. Width and depth Fig. 6 shows the width and depth of the channels varied with different laser powers, frequencies and scanning speeds. As exhib-
ited, the width and depth increase with the increasing laser power from 5 to 18 W, and afterwards reduce at 20 W (Fig. 6 (a)). It can be explained that with the increasing laser power, the energy accumulated on unit area increases, which causes more melting, vaporization and material removal, eventually leads to increasing width and depth. While, in the case of 20 W, large amounts of molten materials re-solidified on the sidewalls and bottoms of the channels, resulting in the reduced width and depth. Fig. 6 (b) illustrates that the channel width increases with the increasing frequency from 2 to 10 kHz, and then reduces with the continuous increase from 10 to 20 kHz; the channel depth reduces with the increasing scanning speed. The phenomenon is explained that the peak power of single pulse reduces with the increasing pulse frequency, however, the number of pulses deposited on the surface increases and the interval time between two consecutive pulses reduces. Therefore, large numbers of pulses collide the material surface, resulting in the increasing channel width (2–10 kHz). However, excessively high pulse frequency causes quite low peak power of the single pulse (10–20 kHz), which leads to the reduced channel width [16,33]. In addition, the decreased peak power with the increasing frequency from 2 to 20 kHz leads to the results that
Fig. 5. SEM and 3D morphologies of micro-channels with different scanning speeds (a) 3 mm/s, (b) 10 mm/s, (c) 30 mm/s under the power of 12 W and frequency of 10 kHz.
337
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 6. Variations of the channel width and depth with different laser parameters. (a) power (frequency = 10 kHz, scanning speed = 10 mm/s), (b) frequency (power = 12 W, scanning speed is 10 mm/s), (c) scanning speed (power = 12 W, frequency = 10 kHz).
the laser energy mainly deposits on the top surface of the materials and the recoil pressure on the materials is reduced, which causes the reduced channel depth. According to Fig. 6 (c), the width increases with the increasing scanning speed from 1 to 3 mm/s and then reduces with the increasing scanning speed from 3 to 30 mm/s; the depth is found to be reduced proportionately to the increasing scanning speed from 1 to 30 mm/s. The reduced width and depth of the channels ascribe to the reduced number of pulses on the materials and reduced interaction time between the laser and materials; while large amounts of recast layers on the sidewalls result in a small channel width at the scanning speed of 1 mm/s. 3.1.3. Material removal rate Fig. 7 shows the MRR with laser power, frequency and scanning speed. It exhibits that the laser parameters have a profound influence on the MRR. Fig. 7 (a) shows that the MRR exhibits an upward trend with the power ranging from 5 to 18 W and then decreases with the power of 20 W (see Fig. 7 (a)). Fig. 7 (b) presents that the MRR is fluctuating with the frequency from 2 to 10 kHz where the maximum MRR obtained at 10 W, and then the MRR reduces. As shown in Fig. 7 (c) that the MRR increases with the increasing scanning speed from 1 to 3 mm/s and then exhibits a downward trend with the increasing scanning speed from 3 to 30 mm/s. The results exhibit that the MRR has a stronger relationship with the channel width and depth in Fig. 6, and the large channel width and depth contribute to the large MRR. 3.1.4. Surface roughness The variations of the surface roughness with different laser parameters are shown in Fig. 8. It reveals that the surface roughness
Ra fluctuates with the laser power ranging from 5 to 12 W and then increases with the increasing power from 12 to 20 W (Fig. 8 (a)) due to large amounts of recast layers deposited surrounding the channels with the excessive energy accumulation. The surface roughness Ra decreases with the increasing frequency from 2 to 20 kHz and the increasing scanning speed from 1 to 30 mm/s (Fig. 8 (b) and (c)) due to the reduced recast layers surrounding the channels. 3.2. Multi-objective optimization 3.2.1. Experimental results Based on the L20 array and the process parameters in Table 2, the width, depth, MRR and surface roughness (Ra) of the channels are selected as the output responses. The results are measured and listed in Table 3. 3.2.2. Mathematical models RSM is a useful method for modeling and predicting the responses affected by the input parameters aiming at optimizing the responses [34,35]. Therefore, the mathematical models between the laser process parameters and the responses are developed based on the RSM. RS model can be expressed as the following polynomial function [36]:
f ¼ a0 þ
n X i¼1
ai xi þ
n X n X i¼1
aij xi xj þ ::: þ e
j¼1
where f is the response, xi and xj is the independent variable of the ith and jth laser process parameters, n is the number of model parameters, a0, ai, aij are regression coefficient and e is the error. Usually, e is considered as a statistical error and it is in line with the normal distribution with the mean zero and variance r2.
338
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 7. Variations of the material remove rate with different laser parameters. (a) power (frequency = 10 kHz, scanning speed = 10 mm/s), (b) frequency (power = 12 W, scanning speed = 10 mm/s), (c) scanning speed (power = 12 W, frequency = 10 kHz).
Fig. 8. Variations of the surface roughness with different laser parameters. (a) power (frequency = 10 kHz, scanning speed = 10 mm/s), (b) frequency (power = 12 W, scanning speed = 10 mm/s), (c) scanning speed (power = 12 W, frequency is 10 kHz).
339
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345 Table 3 Experimental results. Test no.
A (W)
B (kHz)
C (mm/s)
Width (lm)
Depth (lm)
MRR (lm3/s)
Surface roughness Ra (nm)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
12 18 12 18 12 18 12 18 10 20 15 15 15 15 15 15 15 15 15 15
5 5 15 15 5 5 15 15 10 10 2 18 10 10 10 10 10 10 10 10
5 5 5 5 20 20 20 20 12 12 12 12 1 25 12 12 12 12 12 12
52.42 62.24 42.63 56.34 41.64 55.27 38.41 49.72 50.12 58.97 58.32 56.44 48.32 53.21 63.27 62.31 65.34 61.87 65.34 64.36
34.83 42.14 28.45 37.31 26.47 36.43 19.42 29.33 24.43 37.12 34.34 31.43 38.64 27.62 34.43 34.25 36.27 33.67 36.86 35.48
9.13 13.11 6.06 10.51 5.51 10.07 3.73 7.29 6.12 10.94 10.01 8.87 9.34 7.35 10.89 10.67 11.85 10.42 12.04 11.42
44.47 51.12 38.47 47.69 38.31 46.69 34.13 43.24 39.63 49.05 46.05 39.61 49.23 40.11 41.87 42.04 40.44 41.62 42.06 41.11
The mathematical models for the responses of channel width, channel depth, MRR and surface roughness (Ra) are obtained by the Design-Expert software, and that are expressed as follows:
Width ¼ 60:44 þ 2:67A 0:53B 50:16C þ 0:2AB þ 0:18AC þ 0:86BC 1:57A2 2:49B2 6:55C 2 0:78ABC 2:53A2 B þ 26:16A2 C þ 3:39AB2 þ 20:43C 3 Depth ¼ 34:91 þ 4:23A 2:3B 3:68C þ 0:19AB þ 0:48AC 0:43BC 1:59A2 0:9B2 0:6C 2 MRR ¼ 10:95 þ 1:44A 0:35B 4:06C 0:067AB 0:039AC þ 0:14BC 0:89A2 0:69B2 1:19C 2 0:18ABC 0:93A2 B þ 1:13A2 C þ 0:62AB2 þ 1:4C 3 Ra ¼ 41:44 þ 3:63A 2:08B 2:65C þ 0:41AB þ 0:24AC þ 0:22BC þ 0:75A2 þ 0:21B2 þ 1:05C 2 where A, B and C represent the laser power, frequency and scanning speed, respectively.
In order to investigate the goodness of fit and the adequacy of models for the responses, the analysis of variance (ANOVA) method is performed. Tables 4–7 shows the ANOVA results of the developed models for the machined width, machined depth, MRR and surface roughness Ra, respectively. As shown in Table 4 that the F-value of the model is 39.72, which indicates that the model is extremely significant with a probability of 0.04%. Meanwhile, the value of Prob > F < 0.05 indicates model terms are significant. Accordingly, A, C, B2, C2, A2B, AB2 and C3 are significant model terms for the machined width. The values of R2 and Adj. R2 are 0.9911 and 0.9661, which are close to 1. The Adeq. Precision measures the ratio of signal to noise, and the ratio is expected to be greater than 4 [37]. Therefore, the ratio of 19.491 implies an adequate signal. As a result, the developed model is sufficient to predict the machined width. Table 5 shows that the p-value of the model is <0.01, and the F-value is 23.21, which implies that the model with a confidence level of 99% for its adequacy is extremely significant. The p-values for A, B, C and A2 are <0.01, which expresses that A, B, C and A2 are significant model terms. The values of R2 and Adj. R2 are 0.9543 and 0.9132, which are close to 1, meanwhile, the Pred. R2 is 0.7418, which is in reasonable agreement with the Adj. R2 of 0.9132. In addition, the Adeq. Precision of 17.908 is greater than 4,
Table 4 ANOVA for the machined width model. Source
SS
df
MS
F
p-value Prob. > F
Model A B C AB AC BC A2 B2 C2 ABC A2B A2C AB2 C3 Pure error Cor. total R2 Pred. R2
1253.11 39.39 1.43 19.26 0.31 0.25 5.97 5.85 60.83 412.36 4.82 19.93 13.89 37.65 21.17 11.27 1264.38 0.9911 N/A
14 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5 19
89.51 39.39 1.43 19.26 0.31 0.25 5.97 5.85 60.83 412.36 4.82 19.93 13.89 37.65 21.17 2.25
39.72 17.48 0.64 8.55 0.14 0.11 2.65 2.60 27.00 183.00 2.14 8.84 6.17 16.71 9.40
0.0004 0.0086 0.4613 0.0329 0.7267 0.7533 0.1646 0.1681 0.0035 <0.0001 0.2034 0.0310 0.0556 0.0095 0.0279
Adj. R2 Adeq. Precision
0.9661 19.491
Significant
340
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Table 5 ANOVA for the machined depth model. Source
SS
df
MS
F
p-value Prob. > F
Model A B C AB AC BC A2 B2 C2 Residual Lack of fit Pure error Cor. total R2 Pred. R2
543.61 242.48 69.26 176.43 0.28 1.85 1.45 35.59 10.06 4.38 26.02 18.22 7.81 569.63 0.9543 0.7418
9 1 1 1 1 1 1 1 1 1 10 5 5 19
60.40 242.48 69.26 176.43 0.28 1.85 1.45 35.59 10.06 4.38 2.60 3.64 1.56
23.21 93.18 26.61 67.80 0.11 0.71 0.56 13.68 3.87 1.68
<0.0001 <0.0001 0.0004 <0.0001 0.7491 0.4186 0.4724 0.0041 0.0777 0.2235
2.33
0.1869
Adj. R2 Adeq. Precision
Significant
0.9132 17.908
Table 6 ANOVA for the MRR model. Source
SS
df
MS
F
p-value Prob. > F
Model A B C AB AC BC A2 B2 C2 ABC A2B A2C AB2 C3 Pure error Cor. total R2 Pred. R2
115.80 11.55 0.62 0.13 0.036 0.012 0.15 1.88 4.72 13.63 0.26 2.70 0.026 1.28 0.099 2.17 117.97 0.9816 N/A
14 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5 19
8.27 11.55 0.62 0.13 0.036 0.012 0.15 1.88 4.72 13.63 0.26 2.70 0.026 1.28 0.099 0.43
19.08 26.65 1.43 0.29 0.082 0.028 0.36 4.34 10.88 31.44 0.61 6.22 0.060 2.94 0.23
0.0021 0.0036 0.2856 0.6131 0.7860 0.8731 0.5767 0.0917 0.0215 0.0025 0.4698 0.0549 0.8163 0.1468 0.6530
Adj. R2 Adeq. Precision
Significant
0.9302 16.460
Table 7 ANOVA for the surface roughness Ra model. Source
SS
df
MS
F
p-value Prob. > F
Model A B C AB AC BC A2 B2 C2 Residual Lack of fit Pure error Cor. total R2 Pred. R2
341.92 177.97 56.69 91.13 1.36 0.46 0.39 8.03 0.54 13.54 13.61 11.58 2.03 355.52 0.9617 0.7472
9 1 1 1 1 1 1 1 1 1 10 5 5 19
37.99 177.97 56.69 91.13 1.36 0.46 0.39 8.03 0.54 13.54 1.36 2.32 0.41
27.92 130.80 41.67 66.98 1.00 0.34 0.29 5.90 0.39 9.95
<0.0001 <0.0001 <0.0001 <0.0001 0.3408 0.5751 0.6027 0.0355 0.5438 0.0102
5.71
0.0594
Adj. R2 Adeq. Precision
which implies an adequate signal. Thence, the developed model can navigate the machined depth. From Table 6, it is noted that the F-value of the model is 19.08, which exhibits the model is significant with a probability of 0.21%. The p-values of A, B2 and C2 are <0.05, implying that A, B2 and C2 are significant model terms. The values of R2 and Adj. R2 are 0.9816 and 0.9302, which are close
Significant
0.9273 20.253
to 1; meanwhile, the Adeq. Precision of 16.460 is greater than 4, which reveals an adequate signal. Hence, the developed model can be used to predict the MRR of machined channels in the considered parameters ranges. From Table 7, it is clear that the model F-value of 27.92 with a probability of 0.01% exhibits that the model is extremely significant. The p-values of A, B, C, A2 and C2 are <0.05,
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
implying that they are significant model terms. Meanwhile, the Pred. R2 of 0.7472 demonstrates agreement with the Adj. R2 of 0.9273, and the Adeq. Precision of 20.253 implies an adequate signal. Therefore, the developed model for the surface roughness Ra is adequate. The predicted and actual values of machined width, depth, MRR and surface roughness Ra are shown in Fig. 9. It can be seen that most of the data points of predicted values and actual values are distributed closed to the diagonal line, implying the residuals in prediction of each response are negligible, and the actual values have a strong agreement with the predicted values. The results further prove the adequacy of the developed models. 3.2.3. Effect of laser process parameters on the responses The variations of channel width with different laser process parameters are shown in Fig. 10. As can be seen in Fig. 10 (a), with low laser power and high frequency, the machined width of the channels is low, and with the high laser power and low frequency, the machined width of the channels is large. This is because that higher laser power directly increases the accumulated energy on the ablated surface, and the low frequency leads to the increasing peak power of the single pulse, thereby, resulting in a large machined width. Fig. 10 (b) shows that with the increasing laser power and reduced scanning speed in the certain range, the machined width increases with the increase of energy accumulation and interaction time between the laser and materials. However, with quite higher laser power and lower scanning speed, excessive ablation results in large amounts of recast layers on the sidewalls and brims of machined channels, which reduces the machined width. Fig. 10 (c) shows that the frequency has a slight effect on the machined width, and the width decreases with the increasing scanning speed. The machined width reaches a maximum value at the lowest frequency and scanning speed.
341
Fig. 11 shows the variations of channel depth with different laser process parameters. It is observed from Fig. 11 (a) that the largest depth is produced with the high laser power and low frequency, and the smallest depth is produced with the low laser power and high frequency. It can be explained that high laser power provides more energy deposited on the materials, and the low frequency provides the large peak power of the single pulse, which leads to more materials that evaporated from the ablated surface. Conversely, the low laser power provides less energy deposited on the materials, on the other hand, the excessive high frequency provides low peak power though the number of pulses is increased. Therefore, the small depth is produced with less removed materials. Fig. 11 (b) shows that the maximum depth is induced by high laser power and low scanning speed, and the depth decreases with the reduced laser power and increasing scanning speed. It can be explained that the peak power increases with the high laser power, and low scanning speed increases the overlap of pulses and the interaction time between the laser and materials, which enables more materials to be ablated and ejected and then results in a large depth [31]. Fig. 11 (c) shows that the depth increases with the reduced frequency and scanning speed. This is because of the increasing interaction time between the laser and materials with low scanning speed and the increasing peak power induced by the low frequency. Fig. 12 shows the variations of MRR with different laser process parameters. As shown in Fig. 12 (a), at high laser power combined with low frequency, the peak power and the energy calculation on the material surface are quite high, which lead to the large width and depth, resulting in the large MRR. While at low laser power and high frequency, the small width and depth of the channels induced by the low peak power, thereby resulting in the small MRR. Fig. 12 (b) illustrates that the maximum MRR obtained with the large laser power combined with the small scanning speed; with the reduced laser power and increasing frequency, the MRR
Fig. 9. Relation between the actual and predicted values.
342
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 10. Variations of channel width with different laser process parameters.
Fig. 11. Variations of channel depth with different laser process parameters.
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
343
Fig. 12. Variations of MRR with different laser process parameters.
almost tends to be a decreased trend. Fig. 12 (c) exhibits that the scanning speed seems to have a more important effect on MRR compared to the frequency, and low scanning speed causes a large MRR. The maximum MRR is obtained at the lowest scanning speed
and frequency due to the large width and depth of the machined channels (see Figs. 10 and 11). Fig. 13 shows the variations of surface roughness Ra with different laser process parameters. Fig. 13 (a) reveals that the surface
Fig. 13. Variations of surface roughness Ra with different laser process parameters.
344
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 14. Micro-channels machined with the optimal process parameters.
roughness Ra increases with the increasing laser power, and that reduces with the increasing frequency. The minimum value is obtained at the low laser power and high frequency. It can be explained that with the low laser power and high frequency, less materials are melted and re-solidified surrounding the channels, resulting in the relatively small surface roughness Ra. Meanwhile, the increasing laser power and reduced frequency increase the peak power and then cause more recast layers surrounding the channels, resulting in unsmooth surfaces. As shown in Fig. 13 (b), it reveals that the surface roughness Ra increases with the increasing laser power, and that decreases with the increasing scanning speed. The minimum surface roughness Ra is obtained at the lowest laser power and scanning speed in the parameter ranges. This is attributed to the less molten and re-deposited materials with lower peak power and interaction time between the laser and materials induced by low laser power and scanning speed. Fig. 13 (c) demonstrates that the surface roughness Ra decreases with the increase of frequency and scanning speed. It can be ascribed to the less recasts deposited surrounding the channels due to the fact that less materials are melted and re-solidified with the increase of frequency and scanning speed.
roughness Ra are investigated by single factor tests and multiobjective optimization methods. The following conclusions are obtained:
3.2.4. Experimental validation In order to further verify the validity of the developed models, the verification experiments are carried out according to the optimal parameters obtained by the Design-Expert software. With the purpose of achieving high fabrication efficiency and surface quality of the machined channels, the large MRR and small surface roughness are desired. The multi-objective optimization results for laser power, frequency and scanning speed are 14.29 W, 12.87 kHz and 11.36 mm/s, respectively; and the predicted values of the MRR and surface roughness Ra are 10.62 lm3/s and 39.86 nm with the predicted channel width and depth of 65.34 lm and 32.78 lm, respectively. The final channels are machined with the optimal process parameters, and the results are shown in Fig. 14. As can be seen that the edges and bottoms of the channels are smooth with few recast layers (Fig. 14 (a) and (b)), the actual width, depth and surface roughness Ra of the channels are about 62.82 lm, 31.56 lm and 41.38 nm, respectively, and the MRR is about 11.26 lm3/s. It can be seen that the predicted values are close to the actual machined values, meanwhile, the prediction errors of channel width, depth, MRR and surface roughness Ra are calculated, and that are 4.01%, 3.87%, 5.68%, and 3.67%, respectively. The results show that the error percentages are within acceptable range (<8%) and the developed models can predict the responses quite satisfactory.
Acknowledgements
4. Conclusions In the present research, the micro-channels are produced on the PCD surface by a nanosecond laser, the effects of laser process parameters on the channel width, depth, MRR and surface
(1) The high laser power, low scanning speed and low frequency are prone to the large channel width, depth, MRR and surface roughness Ra due to the high energy accumulation on the surface; while the channel width, depth and MRR may reduce with the excessive energy accumulation due to large amounts of recast layers. (2) The mathematical models for the machined width, depth, MRR and surface roughness Ra are developed based on the RSM, and they are adequate to predict each response. (3) The desired micro-channels are fabricated with the optimal process parameters: the laser power of 14.29 W, frequency of 12.87 kHz and scanning speed of 11.36 mm/s obtained by the multi-objective optimization. The machined width, depth and surface roughness Ra of the channels are about 62.82 lm, 31.56 lm and 41.38 nm, respectively, and the MRR is about 11.26 lm3/s.
This work is supported by the Natural Science Foundation of Jiangsu Province (BK20170676), the Jiangsu Key Laboratory of Precision and Micro-Manufacturing Technology and the Open Research Fund Program of Shaanxi Key Laboratory of Non-Traditional Machining (2017SXTZKFJG10). References [1] I. Etsion, G. Halperin, V. Brizmer, Y. Kligerman, Experimental investigation of laser surface textured parallel thrust bearings, Tribol. Lett. 17 (2) (2004) 295– 300. [2] Y.Q. Xing, J.X. Deng, X.S. Wang, K.F. Ehmann, J. Cao, Experimental assessment of laser textured cutting tools in dry cutting of aluminum alloys, J. Manuf. Sci. Eng.-Trans. ASME 138 (2016) 071006. [3] S.X. Bai, X.D. Peng, Y.F. Li, S.E. Sheng, A hydrodynamic laser surface-textured gas mechanical face seal, Tribol. Lett. 38 (2010) 187–194. [4] W. Grabon, W. Koszela, P. Pawlus, S. Ochwat, Improving tribo-logical behaviour of piston ring-cylinder liner frictional pair by liner surface texturing, Tribol. Int. 61 (2013) 102–108. [5] U. Pettersson, S. Jacobson, Influence of surface texture on boundary lubricated sliding contacts, Tribol. Int. 36 (2003) 857–864. [6] T.C. Hu, L.L. Hu, Q. Ding, Effective solution for the tribological problems of Ti6Al-4V: combination of laser surface texturing and solid lubricant film, Surf. Coat. Technol. 206 (2012) 5060–5066. [7] L.C. Tshabalala, S. Pityana, Surface texturing of Si3N4-SiC ceramic tool components by pulsed laser machining, Surf. Coat. Technol. 289 (2016) 52–60. [8] L. Rapoport, A. Moshkovich, V. Perfilyev, A. Gedanken, Y. Koltypin, E. Sominski, G. Halperin, I. Etsion, Wear life and adhesion of solid lubricant films on lasertextured steel surfaces, Wear 267 (2009) 1203–1207. [9] Z. Wang, Y.B. Li, F. Bai, C.W. Wang, Q.Z. Zhao, Angle-dependent lubricated tribological properties of stainless steel by femtosecond laser surface texturing, Opt. Laser Technol. 81 (2016) 60–66. [10] M. Wakuda, Y. Yamauchi, S. Kanzaki, Y. Yasuda, Effect of surface texturing on friction reduction between ceramic and steel materials under lubricated sliding contact, Wear 254 (2003) 356–363.
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345 [11] Y.Q. Xing, J.X. Deng, Z. Wu, F.F. Wu, High friction and low wear properties of laser-textured ceramic surface under dry friction, Opt. Laser Technol. 93 (2017) 24–32. [12] Z. Wu, J.X. Deng, Y.Q. Xing, H.W. Cheng, Effect of surface texturing on friction properties of WC/Co cemented carbide, Mater. Des. 41 (2012) 142–149. [13] S. Fang, L. Llanes, D. Bähre, Laser surface texturing of a WC-CoNi cemented carbide grade: Surface topography design for honing application, Tribol. Int. 122 (2018) 236–245. [14] Y.Q. Xing, J.X. Deng, P. Gao, J.T. Gao, Z. Wu, Angle-dependent tribological properties of AlCrN coatings with microtextures induced by nanosecond laser under dry friction, Appl. Phys. A 124 (2018) 294. [15] A.A. Voevodin, J.S. Zabinski, Laser surface texturing for adaptive solid lubrication, Wear 261 (2006) 1285–1292. [16] Y.S. Su, L. Li, N. He, W. Zhao, Experimental study of fiber laser surface texturing of polycrystalline diamond tools, Int. J. Refract. Met. Hard Mater. 45 (2014) 117–124. [17] C.W. Knuteson, T.N. Sexton, C.H. Cooley, Wear-in behaviour of polycrystalline diamond thrust bearings, Wear 271 (2011) 2106–2110. [18] R.B. da Silvaa, Á.R. Machadoa, E.O. Ezugwub, J. Bonneyc, W.F. Sales, Tool life and wear mechanisms in high speed machining of Ti-6Al-4V alloy with PCD tools under various coolant pressures, J. Mater. Process. Technol. 213 (2013) 1459–1464. [19] K.W. Liew, C.K. Kok, M.N.E. Efzan, Effect of EDM dimple geometry on friction reduction under boundary and mixed lubrication, Tribol. Int. 101 (2016) 1–9. [20] D.M. Kim, I. Lee, S.K. Kim, H.W. Park, Influence of a micropatterned insert on characteristics of the tool-workpiece interface in a hard turning process, J. Mater. Process. Technol. 229 (2016) 160–171. [21] W.L. Chang, J.N. Sun, X.C. Luo, J.M. Ritchie, C. Mack, Investigation of microstructured milling tool for deferring tool wear, Wear 271 (2011) 2433– 2437. [22] T. Obikawa, A. Kamio, H. Takaoka, A. Osada, Micro-texture at the coated tool face for high performance cutting, Int. J. Mach. Tools Manuf. 51 (2011) 966–972. [23] J. Xie, Y.H. Li, L.F. Yang, Study on 5-axial milling on microstructured freeform surface using the macro-ball cutter patterned with micro-cutting-edge array, CIRP Ann. Manuf. Technol. 64 (2015) 101–104.
345
[24] P. Guo, K.F. Ehmann, An analysis of the surface generation mechanics of the elliptical vibration texturing process, Int. J. Mach. Tools Manuf. 64 (2013) 85–95. [25] X.S. Wang, Y.N. Zhang, L. Wang, J.Y. Xian, M.F. Jin, M. Kang, Fabrication of micro-convex domes using long pulse laser, Appl. Phys. A 123 (2017) 51. [26] I. Etsion, State of the art in laser surface texturing, J. Tribol.-Trans. ASME 127 (2005) 248–253. [27] D. Wang, W.S. Zhao, L. Gu, X.M. Kang, A study on micro-hole machining of polycrystalline diamond by micro-electrical discharge machining, J. Mater. Process. Technol. 211 (2011) 3–11. [28] E. Hashikawa, Development of EDM/laser compound precision fine machining system, Manuf. Technol. Mach. Tool. 2 (2004) 46–50. [29] J. Yan, K. Watanabe, T. Aoyama, Micro-electrical discharge machining of polycrystalline diamond using rotary cupronickel electrode, CIRP Ann. Manuf. Technol. 63 (2014) 209–212. [30] M.R.H. Knowles, G. Rutterford, D. Karnakis, A. Ferguson, Micro-machining of metals, ceramics and polymers using nanosecond lasers, Int. J. Adv. Manuf. Technol. 33 (2007) 95–102. [31] B.N. Chichkov, C. Momma, S. Nolte, F. von Alvensleben, A. Tunnermann, Femtosecond, picosecond and nanosecond laser ablation of solids, Appl. Phys. A 63 (1996) 109–115. [32] H.D. Vora, S. Santhanakrishnan, S.P. Harimkar, S.K. Boetcher, N.B. Dahotre, Evolution of surface topography in one-dimensional laser machining of structural alumina, J. Eur. Ceram. Soc. 32 (2012) 4205–4218. [33] G. Kibria, B. Doloi, B. Bhattacharyya, Predictive model and process parameters optimization of Nd: YAG laser micro-turning of ceramics, Int. J. Adv. Manuf. Technol. 65 (2013) 213–229. [34] R. Biswas, A.S. Kuar, S. Sarkar, S. Mitra, A parametric study of pulsed Nd: YAG laser micro-drilling of gamma-titanium aluminide, Opt. Laser Technol. 42 (2010) 23–31. [35] X.S. Wang, Y.Q. Xing, M. Giovannini, Effect of overlap and overscan number in laser surface texturing of medical needles, Appl. Phys. A 120 (2015) 229–238. [36] H. Oktem, T. Erzurumlu, H. Kurtaran, Application of response surface methodology in the optimization of cutting conditions for surface roughness, J. Mater. Process. Technol. 170 (2005) 11–16. [37] D.C. Montgomery, Design and Analysis of Experiments, fifth ed., John Wiley & Sons, New York, 2000.
Contents lists available at ScienceDirect
Optics and Laser Technology journal homepage: www.elsevier.com/locate/optlastec
Full length article
Micro-channels machining on polycrystalline diamond by nanosecond laser Youqiang Xing a,b,c, Lei Liu a,⇑, Xiuqing Hao b,⇑, Ze Wu a, Peng Huang a, Xingsheng Wang d a
School of Mechanical Engineering, Southeast University, Nanjing 211189, Jiangsu Province, PR China Jiangsu Key Laboratory of Precision and Micro-Manufacturing Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, Jiangsu Province, PR China c Shaanxi Key Laboratory of Non-Traditional Machining, Xi’an Technological University, Xi’an 710021, Shaanxi Province, PR China d Department of Mechanical Engineering, Nanjing Agricultural University, Nanjing 210031, Jiangsu Province, PR China b
a r t i c l e
i n f o
Article history: Received 29 March 2018 Received in revised form 28 May 2018 Accepted 12 July 2018
Keywords: PCD Micro-channels Nanosecond laser texturing RSM
a b s t r a c t In this paper, micro-channels are fabricated on the polycrystalline diamond (PCD) surface by a nanosecond laser. The effects of laser process parameters (laser power, frequency and scanning speed) on the responses of surface quality, channel dimension, material remove rate (MRR) and surface roughness Ra are investigated by the single factor and multi-objective optimization tests. The mathematical models between the laser process parameters and responses are developed based on the response surface method (RSM) and the adequacy of the models for the responses is studied by the analysis of variance (ANOVA) method. Ultimately, the desired micro-channels are fabricated with the optimal process parameters: laser power of 14.29 W, frequency of 12.87 kHz and scanning speed of 11.36 mm/s. The results exhibit that the developed mathematical models are in well agreement with the experimental results. Ó 2018 Published by Elsevier Ltd.
1. Introduction Micro-textures have proven to be an effective method to improve the tribological behavior of the interfaces, and that are widely used in bearing, mechanical seals, cutting tools and cylinder liners [1–4]. Meanwhile, the hole and channel may be the most successful application of the texture patters on the tribo-surfaces, and the main mechanisms of the micro-textures are attributed to the entrapment of wear debris, increased load carrying capacity, storage of lubricants and reduced contact area [5–7]. Presently, the fabrication and application of surface textures mainly focus on the metals [8,9], ceramics [10,11], cemented carbide [12,13] and coatings [14,15], and few studies about the fabrication of micro-textures on polycrystalline diamond (PCD) are reported [16]. PCD is an excellent material for using as bearings and cutting tools in automotive, aerospace and machining industries with its unique properties of high hardness, good thermal conductivity, high wear resistance and low friction coefficient [17,18]. However, the high hardness and brittleness of the PCD make it difficult to fabricate the micro textures on its surface with the conventional methods. At present, micro-textures can be fabricated by various methods, including electrical discharge machining (EDM) [19,20], ⇑ Corresponding authors. E-mail addresses: [email protected] (L. Liu), [email protected] (X. Hao). https://doi.org/10.1016/j.optlastec.2018.07.024 0030-3992/Ó 2018 Published by Elsevier Ltd.
focused ion beam etching (FIB) [21], lithography technology [22], mechanical micro-machining [23,24], laser [25,26], etc. Among these methods, EDM and laser machining technology have been attracted more attentions on the machining of PCD in most studies. For examples, Wang et al. [27] investigated the micro-hole machining performance of PCD by micro-EDM technology, and the process parameters on surface quality and material remove rate of the micro-holes are studied, ultimately, the optimal machining parameters are selected. Hashikawa [28] fabricated the micro-holes on the PCD surface by laser machining first, and then finished by EDM technology for the high surface quality. Yan et al. [29] fabricated micro-textures on the PCD surface by micro-EDM with a rotary cupronickel electrode, the effect of machining parameters (discharge energy electrode and rotation rate) on the machining characteristics were clarified, and then the micro-channels with low surface roughness as well as high form accuracy were obtained. Su et al. [16] studied the influence of laser process parameters on the formation of micro-holes and micro-channels on the PCD surface by a fiber laser, and the dimensions and topography of the micro-textures were investigated. Furthermore, the optimal process parameters were obtained for the high surface quality of micro-textures. Comparing to EDM, laser machining technology exhibits more advantages for the fabrication of micro-textures such as high efficiency, no pollution, no mechanical damage and tool wear, therefore, it is a more efficient method for fabricating micro-textures on the PCD surface.
334
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Comparing to femtosecond and picosecond lasers, the nanosecond laser exhibits high machining efficiency and low costs. Thus, the nanosecond laser is widely used for the micromachining [30]. Unfortunately, the nanosecond laser micromachining produces more heat-affected zone (HAZ), recast layers and debris due to the long laser-material interaction time and high energy density [31]. Therefore, the laser process parameters need to be further optimized for the high surface quality. In this paper, a nanosecond laser is used to produce the microchannels on the PCD surface. Single factor test and multi-objective optimization are carried out to investigate the effect of laser process parameters on the surface quality, geometric dimensions and material remove rate (MRR) of the machined channels. Meanwhile, the predictive models of the geometric dimensions, MRR and surface roughness Ra are founded based on the response surface method (RSM), and the optimal process parameters are obtained for the desired channels. 2. Experimental details 2.1. Materials The commercial PCD compacts (Far-East New Materials Co., Ltd., China) are used as the test materials. The PCD layer is formed by sintered diamond grit with Co binder to WC cemented carbide substrate at high temperature and pressure. The thickness of the PCD layer is about 0.5 mm, and it is about 2.5 mm for cemented carbide substrate. The PCD layer comprises an average grain size of 5 lm, and the surface roughness Ra is 0.038 lm. Table 1 shows the properties of the cemented carbide substrate and PCD layer. Fig. 1 shows the typical microstructure and composition analysis of the PCD surface. It illustrates that the surface of the sample is dense and the porosity is virtually absent; EDX analyses of points A and B confirm that the gray regions are the PCD phases and the white regions are the Co binders.
sional stage. The wavelength of the laser is 1064 nm, pulse duration is 10 ns, maximum power is 20 W and the maximum frequency is 20 kHz. A single laser beam is focused on the sample surface with an incident angle of 90°, meanwhile, the preparation process is monitored by a digital microscope. All experiments are performed in air condition at room temperature. 2.2.2. Single factor test Single factor tests are used to investigate the effect of laser process parameters (laser power, scanning speed and frequency) on the surface morphology, channel width (horizontal distance at the top surface of the channel) and depth (vertical distance between the top and bottom surface of the channel), the material remove rate (MRR) and surface roughness (Ra) of the machined surface. The laser power ranges from 5 to 20 W, the frequency ranges from 2 to 20 kHz, and the scanning speed ranges from 1 to 30 mm/s. After laser irradiation, the samples are cleaned twice for 30 min in an ultrasonic bath by alcohol and then dried. The surface morphologies of the channels are observed by scanning electron microscope (SEM, QUANTA FEG250, USA) and three dimensional microscope (VHX-600E, Japan), and the surface compositions are detected by an energy dispersive X-ray spectroscope (EDX, X-MAX50, UK). The channel profile with a certain width (horizontal distance at the top surface of the channel) and depth (vertical distance between the top and bottom surface of the channel) are measured by a three dimensional microscope (VHX-600E, Japan). The cross-sectional area of the channel is also measured based on the section profile and then the material remove rate (MRR) is calculated by taking into account of the scanning speed. The surface roughness (Ra) of the machined surface, which represents the surface quality of the channels at a certain extent, is measured with the same area (including four channels in the area) by the white light interferometer (Wyko NT9300, USA). For reducing the measurement error, each test is repeated three times and the average values are recorded as the output.
2.2. Experimental methods 2.2.1. Experimental setup An Nd:YAG laser (DP-H20, China) is used to produce the microchannels on PCD sample surface that is fixed on a three dimen-
2.2.3. Multi-objective optimization To further investigate the effect of the process parameters and find the optimal values for getting the higher MRR and lower surface roughness Ra. An L20 array according to the CCD (central
Table 1 Properties of cemented carbide substrate and polycrystalline diamond layer. Materials
Density (gcm-3)
Elastic modulus (GPa)
Thermal conductivity (MPa)
Thermal expansion coefficient (10-6/K)
WC + 8 wt.%Co PCD
14.5–14.9 3.6–3.8
590–610 860–900
79.5 720.0
4.5–4.6 1.0–1.18
Fig. 1. SEM micrograph and EDX analysis of the PCD surface, (a) SEM micrograph of the polished surface, (b) and (c) EDX composition analyses marked in points A and B, respectively.
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
composite design) method based on the RSM is performed in the present study. Laser power, frequency and scanning speed are selected as the factors and 5 levels of each factor are used based on the pilot experiments. Table 2 shows the values of factors at different levels. 3. Results and discussion The surface morphologies, profile and EDX analyses of the heataffected zone of the micro-channels are shown in Fig. 2. As seen in this figure, the channels are formed with the laser ablation, the formation mechanism can be explained that the materials are melted with the high temperature induced by the absorbed laser energy, and then vaporized and ejected away from the machined zone with the high recoil pressure, resulting the material removal [32]. Meanwhile, the heat-affected are formed along the micro-channels induced by high energy ablation. The 3D view in Fig. 2 (a) demonstrates that the channel is discontinuous with the unsuitable laser process parameters due to the molten materials re-solidified at the bottoms. The profile of the channel seems to be a triangle with a certain width at the top surface and vertical depth between the top and bottom surface, shown in Fig. 2 (b). The 3D morphology in Fig. 2 (c) shows that the recast layers with large amounts of bulges around the edges of the channel are higher compared to the based plane surface, which are formed by the vapor blast ejec-
Table 2 Factors and levels. Factors
Symbol
Unit
Level 1
Level 2
Level 3
Level 4
Level 5
Power Frequency Scanning speed
A B C
W kHz mm/s
10 2 1
12 5 5
15 10 12
18 15 20
20 18 25
335
tion and rapid condensation of molten materials. The EDX analyses in Fig. 2 (d)-(f) shows that the oxygen content is higher surrounding the channel, indicating that the oxidation reaction occurs in heated-affected zone with the high intensity laser ablation. Therefore, to avoid more defects at the channel bottoms and the bulges in heated-affected zone, the laser process parameters are further investigated for the formation of the optimal channels. 3.1. Single factor test 3.1.1. Surface morphology Fig. 3 shows the SEM and 3D morphologies of micro-channels with different laser powers. As can be seen that the laser power has a profound effect on the formation of micro-channels. With the laser power of 5 W (Fig. 3 (a)), the irradiated zone becomes rough and a few materials are ejected, while quite shallow channels are formed because that the energy accumulated by lower laser power on the surface is not enough to produce the obvious channels. In the case of 12 W (Fig. 3 (b)), the energy accumulated on the unit area increases and more materials are melted and ejected from the ablation zone, therefore, a better quality channels with continuous bottoms and less recast layers surrounding the edges are produced. With the continuously increasing laser power of 18 W (Fig. 3 (c)), severe ablation occurs and large amounts of molten materials are re-solidified, which result in large amounts of recast layers formed on the sidewalls, bottoms and the brims zone of the channels. Fig. 4 shows the SEM and 3D morphologies of micro-channels with different frequencies. It exhibits that the laser power seems to have a milder effect on the surface quality of the channels compared to the laser power, and that has an important influence on the depth and width of the channels. The phenomenon dues to the differences of the peak power of a single pulse and the number pules deposited in the unit area for the given laser power and
Fig. 2. Surface morphologies, profile and EDX analyses of the heat-affected zone of the micro-channels.
336
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 3. SEM and 3D morphologies of micro-channels with different laser powers (a) 5 W, (b) 12 W, (c) 18 W under the frequency of 10 kHz and speed of 10 mm/s.
Fig. 4. SEM and 3D morphologies of micro-channels with different frequencies (a) 5 kHz, (b) 10 kHz, (c) 20 kHz under the power of 12 W and speed of 10 mm/s.
scanning speed. According to the results, the channels are continuous and few recast layers are formed at the channel bottoms and sidewalls. Fig. 5 shows the SEM and 3D morphologies of micro-channels with different scanning speeds. As shown in Fig. 5 (a), the channel bottom is continuous with the scanning speed of 3 mm/s; while the heat-affected zone surrounding the channel edges is large and some recast layers and debris can be seen on the sidewalls of the channels. As the scanning speed increases to 10 mm/s and 30 mm/s (Fig. 5 (b) and (c)), the recast layers and heat-affected zone reduce, and the channel bottoms are continuous; in addition, the width and depth of the channels have a significant change. It can be explained that with the increasing scanning speed, the interaction time between the laser and material surface reduces, and the energy accumulated on the unit area decreases, which result in less removed materials and less re-solidified materials forming the recast layers. 3.1.2. Width and depth Fig. 6 shows the width and depth of the channels varied with different laser powers, frequencies and scanning speeds. As exhib-
ited, the width and depth increase with the increasing laser power from 5 to 18 W, and afterwards reduce at 20 W (Fig. 6 (a)). It can be explained that with the increasing laser power, the energy accumulated on unit area increases, which causes more melting, vaporization and material removal, eventually leads to increasing width and depth. While, in the case of 20 W, large amounts of molten materials re-solidified on the sidewalls and bottoms of the channels, resulting in the reduced width and depth. Fig. 6 (b) illustrates that the channel width increases with the increasing frequency from 2 to 10 kHz, and then reduces with the continuous increase from 10 to 20 kHz; the channel depth reduces with the increasing scanning speed. The phenomenon is explained that the peak power of single pulse reduces with the increasing pulse frequency, however, the number of pulses deposited on the surface increases and the interval time between two consecutive pulses reduces. Therefore, large numbers of pulses collide the material surface, resulting in the increasing channel width (2–10 kHz). However, excessively high pulse frequency causes quite low peak power of the single pulse (10–20 kHz), which leads to the reduced channel width [16,33]. In addition, the decreased peak power with the increasing frequency from 2 to 20 kHz leads to the results that
Fig. 5. SEM and 3D morphologies of micro-channels with different scanning speeds (a) 3 mm/s, (b) 10 mm/s, (c) 30 mm/s under the power of 12 W and frequency of 10 kHz.
337
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 6. Variations of the channel width and depth with different laser parameters. (a) power (frequency = 10 kHz, scanning speed = 10 mm/s), (b) frequency (power = 12 W, scanning speed is 10 mm/s), (c) scanning speed (power = 12 W, frequency = 10 kHz).
the laser energy mainly deposits on the top surface of the materials and the recoil pressure on the materials is reduced, which causes the reduced channel depth. According to Fig. 6 (c), the width increases with the increasing scanning speed from 1 to 3 mm/s and then reduces with the increasing scanning speed from 3 to 30 mm/s; the depth is found to be reduced proportionately to the increasing scanning speed from 1 to 30 mm/s. The reduced width and depth of the channels ascribe to the reduced number of pulses on the materials and reduced interaction time between the laser and materials; while large amounts of recast layers on the sidewalls result in a small channel width at the scanning speed of 1 mm/s. 3.1.3. Material removal rate Fig. 7 shows the MRR with laser power, frequency and scanning speed. It exhibits that the laser parameters have a profound influence on the MRR. Fig. 7 (a) shows that the MRR exhibits an upward trend with the power ranging from 5 to 18 W and then decreases with the power of 20 W (see Fig. 7 (a)). Fig. 7 (b) presents that the MRR is fluctuating with the frequency from 2 to 10 kHz where the maximum MRR obtained at 10 W, and then the MRR reduces. As shown in Fig. 7 (c) that the MRR increases with the increasing scanning speed from 1 to 3 mm/s and then exhibits a downward trend with the increasing scanning speed from 3 to 30 mm/s. The results exhibit that the MRR has a stronger relationship with the channel width and depth in Fig. 6, and the large channel width and depth contribute to the large MRR. 3.1.4. Surface roughness The variations of the surface roughness with different laser parameters are shown in Fig. 8. It reveals that the surface roughness
Ra fluctuates with the laser power ranging from 5 to 12 W and then increases with the increasing power from 12 to 20 W (Fig. 8 (a)) due to large amounts of recast layers deposited surrounding the channels with the excessive energy accumulation. The surface roughness Ra decreases with the increasing frequency from 2 to 20 kHz and the increasing scanning speed from 1 to 30 mm/s (Fig. 8 (b) and (c)) due to the reduced recast layers surrounding the channels. 3.2. Multi-objective optimization 3.2.1. Experimental results Based on the L20 array and the process parameters in Table 2, the width, depth, MRR and surface roughness (Ra) of the channels are selected as the output responses. The results are measured and listed in Table 3. 3.2.2. Mathematical models RSM is a useful method for modeling and predicting the responses affected by the input parameters aiming at optimizing the responses [34,35]. Therefore, the mathematical models between the laser process parameters and the responses are developed based on the RSM. RS model can be expressed as the following polynomial function [36]:
f ¼ a0 þ
n X i¼1
ai xi þ
n X n X i¼1
aij xi xj þ ::: þ e
j¼1
where f is the response, xi and xj is the independent variable of the ith and jth laser process parameters, n is the number of model parameters, a0, ai, aij are regression coefficient and e is the error. Usually, e is considered as a statistical error and it is in line with the normal distribution with the mean zero and variance r2.
338
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 7. Variations of the material remove rate with different laser parameters. (a) power (frequency = 10 kHz, scanning speed = 10 mm/s), (b) frequency (power = 12 W, scanning speed = 10 mm/s), (c) scanning speed (power = 12 W, frequency = 10 kHz).
Fig. 8. Variations of the surface roughness with different laser parameters. (a) power (frequency = 10 kHz, scanning speed = 10 mm/s), (b) frequency (power = 12 W, scanning speed = 10 mm/s), (c) scanning speed (power = 12 W, frequency is 10 kHz).
339
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345 Table 3 Experimental results. Test no.
A (W)
B (kHz)
C (mm/s)
Width (lm)
Depth (lm)
MRR (lm3/s)
Surface roughness Ra (nm)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
12 18 12 18 12 18 12 18 10 20 15 15 15 15 15 15 15 15 15 15
5 5 15 15 5 5 15 15 10 10 2 18 10 10 10 10 10 10 10 10
5 5 5 5 20 20 20 20 12 12 12 12 1 25 12 12 12 12 12 12
52.42 62.24 42.63 56.34 41.64 55.27 38.41 49.72 50.12 58.97 58.32 56.44 48.32 53.21 63.27 62.31 65.34 61.87 65.34 64.36
34.83 42.14 28.45 37.31 26.47 36.43 19.42 29.33 24.43 37.12 34.34 31.43 38.64 27.62 34.43 34.25 36.27 33.67 36.86 35.48
9.13 13.11 6.06 10.51 5.51 10.07 3.73 7.29 6.12 10.94 10.01 8.87 9.34 7.35 10.89 10.67 11.85 10.42 12.04 11.42
44.47 51.12 38.47 47.69 38.31 46.69 34.13 43.24 39.63 49.05 46.05 39.61 49.23 40.11 41.87 42.04 40.44 41.62 42.06 41.11
The mathematical models for the responses of channel width, channel depth, MRR and surface roughness (Ra) are obtained by the Design-Expert software, and that are expressed as follows:
Width ¼ 60:44 þ 2:67A 0:53B 50:16C þ 0:2AB þ 0:18AC þ 0:86BC 1:57A2 2:49B2 6:55C 2 0:78ABC 2:53A2 B þ 26:16A2 C þ 3:39AB2 þ 20:43C 3 Depth ¼ 34:91 þ 4:23A 2:3B 3:68C þ 0:19AB þ 0:48AC 0:43BC 1:59A2 0:9B2 0:6C 2 MRR ¼ 10:95 þ 1:44A 0:35B 4:06C 0:067AB 0:039AC þ 0:14BC 0:89A2 0:69B2 1:19C 2 0:18ABC 0:93A2 B þ 1:13A2 C þ 0:62AB2 þ 1:4C 3 Ra ¼ 41:44 þ 3:63A 2:08B 2:65C þ 0:41AB þ 0:24AC þ 0:22BC þ 0:75A2 þ 0:21B2 þ 1:05C 2 where A, B and C represent the laser power, frequency and scanning speed, respectively.
In order to investigate the goodness of fit and the adequacy of models for the responses, the analysis of variance (ANOVA) method is performed. Tables 4–7 shows the ANOVA results of the developed models for the machined width, machined depth, MRR and surface roughness Ra, respectively. As shown in Table 4 that the F-value of the model is 39.72, which indicates that the model is extremely significant with a probability of 0.04%. Meanwhile, the value of Prob > F < 0.05 indicates model terms are significant. Accordingly, A, C, B2, C2, A2B, AB2 and C3 are significant model terms for the machined width. The values of R2 and Adj. R2 are 0.9911 and 0.9661, which are close to 1. The Adeq. Precision measures the ratio of signal to noise, and the ratio is expected to be greater than 4 [37]. Therefore, the ratio of 19.491 implies an adequate signal. As a result, the developed model is sufficient to predict the machined width. Table 5 shows that the p-value of the model is <0.01, and the F-value is 23.21, which implies that the model with a confidence level of 99% for its adequacy is extremely significant. The p-values for A, B, C and A2 are <0.01, which expresses that A, B, C and A2 are significant model terms. The values of R2 and Adj. R2 are 0.9543 and 0.9132, which are close to 1, meanwhile, the Pred. R2 is 0.7418, which is in reasonable agreement with the Adj. R2 of 0.9132. In addition, the Adeq. Precision of 17.908 is greater than 4,
Table 4 ANOVA for the machined width model. Source
SS
df
MS
F
p-value Prob. > F
Model A B C AB AC BC A2 B2 C2 ABC A2B A2C AB2 C3 Pure error Cor. total R2 Pred. R2
1253.11 39.39 1.43 19.26 0.31 0.25 5.97 5.85 60.83 412.36 4.82 19.93 13.89 37.65 21.17 11.27 1264.38 0.9911 N/A
14 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5 19
89.51 39.39 1.43 19.26 0.31 0.25 5.97 5.85 60.83 412.36 4.82 19.93 13.89 37.65 21.17 2.25
39.72 17.48 0.64 8.55 0.14 0.11 2.65 2.60 27.00 183.00 2.14 8.84 6.17 16.71 9.40
0.0004 0.0086 0.4613 0.0329 0.7267 0.7533 0.1646 0.1681 0.0035 <0.0001 0.2034 0.0310 0.0556 0.0095 0.0279
Adj. R2 Adeq. Precision
0.9661 19.491
Significant
340
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Table 5 ANOVA for the machined depth model. Source
SS
df
MS
F
p-value Prob. > F
Model A B C AB AC BC A2 B2 C2 Residual Lack of fit Pure error Cor. total R2 Pred. R2
543.61 242.48 69.26 176.43 0.28 1.85 1.45 35.59 10.06 4.38 26.02 18.22 7.81 569.63 0.9543 0.7418
9 1 1 1 1 1 1 1 1 1 10 5 5 19
60.40 242.48 69.26 176.43 0.28 1.85 1.45 35.59 10.06 4.38 2.60 3.64 1.56
23.21 93.18 26.61 67.80 0.11 0.71 0.56 13.68 3.87 1.68
<0.0001 <0.0001 0.0004 <0.0001 0.7491 0.4186 0.4724 0.0041 0.0777 0.2235
2.33
0.1869
Adj. R2 Adeq. Precision
Significant
0.9132 17.908
Table 6 ANOVA for the MRR model. Source
SS
df
MS
F
p-value Prob. > F
Model A B C AB AC BC A2 B2 C2 ABC A2B A2C AB2 C3 Pure error Cor. total R2 Pred. R2
115.80 11.55 0.62 0.13 0.036 0.012 0.15 1.88 4.72 13.63 0.26 2.70 0.026 1.28 0.099 2.17 117.97 0.9816 N/A
14 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5 19
8.27 11.55 0.62 0.13 0.036 0.012 0.15 1.88 4.72 13.63 0.26 2.70 0.026 1.28 0.099 0.43
19.08 26.65 1.43 0.29 0.082 0.028 0.36 4.34 10.88 31.44 0.61 6.22 0.060 2.94 0.23
0.0021 0.0036 0.2856 0.6131 0.7860 0.8731 0.5767 0.0917 0.0215 0.0025 0.4698 0.0549 0.8163 0.1468 0.6530
Adj. R2 Adeq. Precision
Significant
0.9302 16.460
Table 7 ANOVA for the surface roughness Ra model. Source
SS
df
MS
F
p-value Prob. > F
Model A B C AB AC BC A2 B2 C2 Residual Lack of fit Pure error Cor. total R2 Pred. R2
341.92 177.97 56.69 91.13 1.36 0.46 0.39 8.03 0.54 13.54 13.61 11.58 2.03 355.52 0.9617 0.7472
9 1 1 1 1 1 1 1 1 1 10 5 5 19
37.99 177.97 56.69 91.13 1.36 0.46 0.39 8.03 0.54 13.54 1.36 2.32 0.41
27.92 130.80 41.67 66.98 1.00 0.34 0.29 5.90 0.39 9.95
<0.0001 <0.0001 <0.0001 <0.0001 0.3408 0.5751 0.6027 0.0355 0.5438 0.0102
5.71
0.0594
Adj. R2 Adeq. Precision
which implies an adequate signal. Thence, the developed model can navigate the machined depth. From Table 6, it is noted that the F-value of the model is 19.08, which exhibits the model is significant with a probability of 0.21%. The p-values of A, B2 and C2 are <0.05, implying that A, B2 and C2 are significant model terms. The values of R2 and Adj. R2 are 0.9816 and 0.9302, which are close
Significant
0.9273 20.253
to 1; meanwhile, the Adeq. Precision of 16.460 is greater than 4, which reveals an adequate signal. Hence, the developed model can be used to predict the MRR of machined channels in the considered parameters ranges. From Table 7, it is clear that the model F-value of 27.92 with a probability of 0.01% exhibits that the model is extremely significant. The p-values of A, B, C, A2 and C2 are <0.05,
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
implying that they are significant model terms. Meanwhile, the Pred. R2 of 0.7472 demonstrates agreement with the Adj. R2 of 0.9273, and the Adeq. Precision of 20.253 implies an adequate signal. Therefore, the developed model for the surface roughness Ra is adequate. The predicted and actual values of machined width, depth, MRR and surface roughness Ra are shown in Fig. 9. It can be seen that most of the data points of predicted values and actual values are distributed closed to the diagonal line, implying the residuals in prediction of each response are negligible, and the actual values have a strong agreement with the predicted values. The results further prove the adequacy of the developed models. 3.2.3. Effect of laser process parameters on the responses The variations of channel width with different laser process parameters are shown in Fig. 10. As can be seen in Fig. 10 (a), with low laser power and high frequency, the machined width of the channels is low, and with the high laser power and low frequency, the machined width of the channels is large. This is because that higher laser power directly increases the accumulated energy on the ablated surface, and the low frequency leads to the increasing peak power of the single pulse, thereby, resulting in a large machined width. Fig. 10 (b) shows that with the increasing laser power and reduced scanning speed in the certain range, the machined width increases with the increase of energy accumulation and interaction time between the laser and materials. However, with quite higher laser power and lower scanning speed, excessive ablation results in large amounts of recast layers on the sidewalls and brims of machined channels, which reduces the machined width. Fig. 10 (c) shows that the frequency has a slight effect on the machined width, and the width decreases with the increasing scanning speed. The machined width reaches a maximum value at the lowest frequency and scanning speed.
341
Fig. 11 shows the variations of channel depth with different laser process parameters. It is observed from Fig. 11 (a) that the largest depth is produced with the high laser power and low frequency, and the smallest depth is produced with the low laser power and high frequency. It can be explained that high laser power provides more energy deposited on the materials, and the low frequency provides the large peak power of the single pulse, which leads to more materials that evaporated from the ablated surface. Conversely, the low laser power provides less energy deposited on the materials, on the other hand, the excessive high frequency provides low peak power though the number of pulses is increased. Therefore, the small depth is produced with less removed materials. Fig. 11 (b) shows that the maximum depth is induced by high laser power and low scanning speed, and the depth decreases with the reduced laser power and increasing scanning speed. It can be explained that the peak power increases with the high laser power, and low scanning speed increases the overlap of pulses and the interaction time between the laser and materials, which enables more materials to be ablated and ejected and then results in a large depth [31]. Fig. 11 (c) shows that the depth increases with the reduced frequency and scanning speed. This is because of the increasing interaction time between the laser and materials with low scanning speed and the increasing peak power induced by the low frequency. Fig. 12 shows the variations of MRR with different laser process parameters. As shown in Fig. 12 (a), at high laser power combined with low frequency, the peak power and the energy calculation on the material surface are quite high, which lead to the large width and depth, resulting in the large MRR. While at low laser power and high frequency, the small width and depth of the channels induced by the low peak power, thereby resulting in the small MRR. Fig. 12 (b) illustrates that the maximum MRR obtained with the large laser power combined with the small scanning speed; with the reduced laser power and increasing frequency, the MRR
Fig. 9. Relation between the actual and predicted values.
342
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 10. Variations of channel width with different laser process parameters.
Fig. 11. Variations of channel depth with different laser process parameters.
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
343
Fig. 12. Variations of MRR with different laser process parameters.
almost tends to be a decreased trend. Fig. 12 (c) exhibits that the scanning speed seems to have a more important effect on MRR compared to the frequency, and low scanning speed causes a large MRR. The maximum MRR is obtained at the lowest scanning speed
and frequency due to the large width and depth of the machined channels (see Figs. 10 and 11). Fig. 13 shows the variations of surface roughness Ra with different laser process parameters. Fig. 13 (a) reveals that the surface
Fig. 13. Variations of surface roughness Ra with different laser process parameters.
344
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345
Fig. 14. Micro-channels machined with the optimal process parameters.
roughness Ra increases with the increasing laser power, and that reduces with the increasing frequency. The minimum value is obtained at the low laser power and high frequency. It can be explained that with the low laser power and high frequency, less materials are melted and re-solidified surrounding the channels, resulting in the relatively small surface roughness Ra. Meanwhile, the increasing laser power and reduced frequency increase the peak power and then cause more recast layers surrounding the channels, resulting in unsmooth surfaces. As shown in Fig. 13 (b), it reveals that the surface roughness Ra increases with the increasing laser power, and that decreases with the increasing scanning speed. The minimum surface roughness Ra is obtained at the lowest laser power and scanning speed in the parameter ranges. This is attributed to the less molten and re-deposited materials with lower peak power and interaction time between the laser and materials induced by low laser power and scanning speed. Fig. 13 (c) demonstrates that the surface roughness Ra decreases with the increase of frequency and scanning speed. It can be ascribed to the less recasts deposited surrounding the channels due to the fact that less materials are melted and re-solidified with the increase of frequency and scanning speed.
roughness Ra are investigated by single factor tests and multiobjective optimization methods. The following conclusions are obtained:
3.2.4. Experimental validation In order to further verify the validity of the developed models, the verification experiments are carried out according to the optimal parameters obtained by the Design-Expert software. With the purpose of achieving high fabrication efficiency and surface quality of the machined channels, the large MRR and small surface roughness are desired. The multi-objective optimization results for laser power, frequency and scanning speed are 14.29 W, 12.87 kHz and 11.36 mm/s, respectively; and the predicted values of the MRR and surface roughness Ra are 10.62 lm3/s and 39.86 nm with the predicted channel width and depth of 65.34 lm and 32.78 lm, respectively. The final channels are machined with the optimal process parameters, and the results are shown in Fig. 14. As can be seen that the edges and bottoms of the channels are smooth with few recast layers (Fig. 14 (a) and (b)), the actual width, depth and surface roughness Ra of the channels are about 62.82 lm, 31.56 lm and 41.38 nm, respectively, and the MRR is about 11.26 lm3/s. It can be seen that the predicted values are close to the actual machined values, meanwhile, the prediction errors of channel width, depth, MRR and surface roughness Ra are calculated, and that are 4.01%, 3.87%, 5.68%, and 3.67%, respectively. The results show that the error percentages are within acceptable range (<8%) and the developed models can predict the responses quite satisfactory.
Acknowledgements
4. Conclusions In the present research, the micro-channels are produced on the PCD surface by a nanosecond laser, the effects of laser process parameters on the channel width, depth, MRR and surface
(1) The high laser power, low scanning speed and low frequency are prone to the large channel width, depth, MRR and surface roughness Ra due to the high energy accumulation on the surface; while the channel width, depth and MRR may reduce with the excessive energy accumulation due to large amounts of recast layers. (2) The mathematical models for the machined width, depth, MRR and surface roughness Ra are developed based on the RSM, and they are adequate to predict each response. (3) The desired micro-channels are fabricated with the optimal process parameters: the laser power of 14.29 W, frequency of 12.87 kHz and scanning speed of 11.36 mm/s obtained by the multi-objective optimization. The machined width, depth and surface roughness Ra of the channels are about 62.82 lm, 31.56 lm and 41.38 nm, respectively, and the MRR is about 11.26 lm3/s.
This work is supported by the Natural Science Foundation of Jiangsu Province (BK20170676), the Jiangsu Key Laboratory of Precision and Micro-Manufacturing Technology and the Open Research Fund Program of Shaanxi Key Laboratory of Non-Traditional Machining (2017SXTZKFJG10). References [1] I. Etsion, G. Halperin, V. Brizmer, Y. Kligerman, Experimental investigation of laser surface textured parallel thrust bearings, Tribol. Lett. 17 (2) (2004) 295– 300. [2] Y.Q. Xing, J.X. Deng, X.S. Wang, K.F. Ehmann, J. Cao, Experimental assessment of laser textured cutting tools in dry cutting of aluminum alloys, J. Manuf. Sci. Eng.-Trans. ASME 138 (2016) 071006. [3] S.X. Bai, X.D. Peng, Y.F. Li, S.E. Sheng, A hydrodynamic laser surface-textured gas mechanical face seal, Tribol. Lett. 38 (2010) 187–194. [4] W. Grabon, W. Koszela, P. Pawlus, S. Ochwat, Improving tribo-logical behaviour of piston ring-cylinder liner frictional pair by liner surface texturing, Tribol. Int. 61 (2013) 102–108. [5] U. Pettersson, S. Jacobson, Influence of surface texture on boundary lubricated sliding contacts, Tribol. Int. 36 (2003) 857–864. [6] T.C. Hu, L.L. Hu, Q. Ding, Effective solution for the tribological problems of Ti6Al-4V: combination of laser surface texturing and solid lubricant film, Surf. Coat. Technol. 206 (2012) 5060–5066. [7] L.C. Tshabalala, S. Pityana, Surface texturing of Si3N4-SiC ceramic tool components by pulsed laser machining, Surf. Coat. Technol. 289 (2016) 52–60. [8] L. Rapoport, A. Moshkovich, V. Perfilyev, A. Gedanken, Y. Koltypin, E. Sominski, G. Halperin, I. Etsion, Wear life and adhesion of solid lubricant films on lasertextured steel surfaces, Wear 267 (2009) 1203–1207. [9] Z. Wang, Y.B. Li, F. Bai, C.W. Wang, Q.Z. Zhao, Angle-dependent lubricated tribological properties of stainless steel by femtosecond laser surface texturing, Opt. Laser Technol. 81 (2016) 60–66. [10] M. Wakuda, Y. Yamauchi, S. Kanzaki, Y. Yasuda, Effect of surface texturing on friction reduction between ceramic and steel materials under lubricated sliding contact, Wear 254 (2003) 356–363.
Y. Xing et al. / Optics and Laser Technology 108 (2018) 333–345 [11] Y.Q. Xing, J.X. Deng, Z. Wu, F.F. Wu, High friction and low wear properties of laser-textured ceramic surface under dry friction, Opt. Laser Technol. 93 (2017) 24–32. [12] Z. Wu, J.X. Deng, Y.Q. Xing, H.W. Cheng, Effect of surface texturing on friction properties of WC/Co cemented carbide, Mater. Des. 41 (2012) 142–149. [13] S. Fang, L. Llanes, D. Bähre, Laser surface texturing of a WC-CoNi cemented carbide grade: Surface topography design for honing application, Tribol. Int. 122 (2018) 236–245. [14] Y.Q. Xing, J.X. Deng, P. Gao, J.T. Gao, Z. Wu, Angle-dependent tribological properties of AlCrN coatings with microtextures induced by nanosecond laser under dry friction, Appl. Phys. A 124 (2018) 294. [15] A.A. Voevodin, J.S. Zabinski, Laser surface texturing for adaptive solid lubrication, Wear 261 (2006) 1285–1292. [16] Y.S. Su, L. Li, N. He, W. Zhao, Experimental study of fiber laser surface texturing of polycrystalline diamond tools, Int. J. Refract. Met. Hard Mater. 45 (2014) 117–124. [17] C.W. Knuteson, T.N. Sexton, C.H. Cooley, Wear-in behaviour of polycrystalline diamond thrust bearings, Wear 271 (2011) 2106–2110. [18] R.B. da Silvaa, Á.R. Machadoa, E.O. Ezugwub, J. Bonneyc, W.F. Sales, Tool life and wear mechanisms in high speed machining of Ti-6Al-4V alloy with PCD tools under various coolant pressures, J. Mater. Process. Technol. 213 (2013) 1459–1464. [19] K.W. Liew, C.K. Kok, M.N.E. Efzan, Effect of EDM dimple geometry on friction reduction under boundary and mixed lubrication, Tribol. Int. 101 (2016) 1–9. [20] D.M. Kim, I. Lee, S.K. Kim, H.W. Park, Influence of a micropatterned insert on characteristics of the tool-workpiece interface in a hard turning process, J. Mater. Process. Technol. 229 (2016) 160–171. [21] W.L. Chang, J.N. Sun, X.C. Luo, J.M. Ritchie, C. Mack, Investigation of microstructured milling tool for deferring tool wear, Wear 271 (2011) 2433– 2437. [22] T. Obikawa, A. Kamio, H. Takaoka, A. Osada, Micro-texture at the coated tool face for high performance cutting, Int. J. Mach. Tools Manuf. 51 (2011) 966–972. [23] J. Xie, Y.H. Li, L.F. Yang, Study on 5-axial milling on microstructured freeform surface using the macro-ball cutter patterned with micro-cutting-edge array, CIRP Ann. Manuf. Technol. 64 (2015) 101–104.
345
[24] P. Guo, K.F. Ehmann, An analysis of the surface generation mechanics of the elliptical vibration texturing process, Int. J. Mach. Tools Manuf. 64 (2013) 85–95. [25] X.S. Wang, Y.N. Zhang, L. Wang, J.Y. Xian, M.F. Jin, M. Kang, Fabrication of micro-convex domes using long pulse laser, Appl. Phys. A 123 (2017) 51. [26] I. Etsion, State of the art in laser surface texturing, J. Tribol.-Trans. ASME 127 (2005) 248–253. [27] D. Wang, W.S. Zhao, L. Gu, X.M. Kang, A study on micro-hole machining of polycrystalline diamond by micro-electrical discharge machining, J. Mater. Process. Technol. 211 (2011) 3–11. [28] E. Hashikawa, Development of EDM/laser compound precision fine machining system, Manuf. Technol. Mach. Tool. 2 (2004) 46–50. [29] J. Yan, K. Watanabe, T. Aoyama, Micro-electrical discharge machining of polycrystalline diamond using rotary cupronickel electrode, CIRP Ann. Manuf. Technol. 63 (2014) 209–212. [30] M.R.H. Knowles, G. Rutterford, D. Karnakis, A. Ferguson, Micro-machining of metals, ceramics and polymers using nanosecond lasers, Int. J. Adv. Manuf. Technol. 33 (2007) 95–102. [31] B.N. Chichkov, C. Momma, S. Nolte, F. von Alvensleben, A. Tunnermann, Femtosecond, picosecond and nanosecond laser ablation of solids, Appl. Phys. A 63 (1996) 109–115. [32] H.D. Vora, S. Santhanakrishnan, S.P. Harimkar, S.K. Boetcher, N.B. Dahotre, Evolution of surface topography in one-dimensional laser machining of structural alumina, J. Eur. Ceram. Soc. 32 (2012) 4205–4218. [33] G. Kibria, B. Doloi, B. Bhattacharyya, Predictive model and process parameters optimization of Nd: YAG laser micro-turning of ceramics, Int. J. Adv. Manuf. Technol. 65 (2013) 213–229. [34] R. Biswas, A.S. Kuar, S. Sarkar, S. Mitra, A parametric study of pulsed Nd: YAG laser micro-drilling of gamma-titanium aluminide, Opt. Laser Technol. 42 (2010) 23–31. [35] X.S. Wang, Y.Q. Xing, M. Giovannini, Effect of overlap and overscan number in laser surface texturing of medical needles, Appl. Phys. A 120 (2015) 229–238. [36] H. Oktem, T. Erzurumlu, H. Kurtaran, Application of response surface methodology in the optimization of cutting conditions for surface roughness, J. Mater. Process. Technol. 170 (2005) 11–16. [37] D.C. Montgomery, Design and Analysis of Experiments, fifth ed., John Wiley & Sons, New York, 2000.