1. Introduction

Femtosecond (fs) lasers have been applied in photonics, electronics, and bioengineering owing to their ultrashort pulse widths and extremely high peak intensities. Further, they can be used to fabricate microfluidics, micromechanics, microelectronics, and photonic microcomponents in glass [1–3]. However, the low processing efficiency and damage induced during conventional fs laser drilling or internal modification limit its applications [4–6]. Although ultrafast lasers with GHz/MHz pulse bursts have been developed to increase the processing efficiency, the cost of the apparatus can be a barrier to their industrial use [7,8]. Transient and selective laser (TSL) processing technology has proven to be an invaluable ultrahigh-speed laser machining technology in drilling [9], micro grooves processing [10], micro welding [11] and internal modification [12], achieving an ultrahigh processing efficiency (ablation rate 2.89 × 10⁵ µm³/ms) 4000 times larger than that of the conventional fs laser machining of glass (ablation rate 72 µm³/ms) [9]. In the TSL processing, electron excitation (electron density >10¹⁷cm⁻³) is induced by a single pulse of fs laser through nonlinear absorption processes, and a single pulse of microsecond (µs) laser can be selectively absorbed in this excited region [13]. The absorption coefficient increases in the region where an instantaneous filament is generated [10]. Following absorption, the µs laser pulse heats the excited region, produces plasma (electron density is increased to 10¹⁸cm⁻³), and removes the material with high efficiency. However, when only a µs laser (with a wavelength of 1070 nm) is applied, the glass material cannot be removed owing to its extremely small absorption coefficient. Considering the complex laser-material interaction during TSL processing, it is particularly important to uncover the material removal mechanism and ensure machining quality under such an ultra-high processing efficiency [14]. In particular, for the fabrication of large-area microstructures, a shift in the position of the sample surface or laser instability can also occur, even in the actual industry, which can result in different hole depths or the generation of different hole types (blind or through-hole) due to changed laser intensity on sample surface [14,15]. A critical method to achieve this goal is to use in situ observations based on real-time monitoring, data analysis, and process control [16], including time-resolved two-dimensional imaging observations using pump-probe or high-speed camera [17–21], to reveal the mechanism of fs laser pulse induced electron excitation (filament) [9,19], shockwave propagation [17], µs laser pulse induced plasma evolution [22] and material removal [10]. Further, a one-dimensional signal sensor can be used for rapid evaluation of machining performance.

Laser-induced plasma has already acted as a source for evaluating laser machining quality because of its close association with drilling processes. Monitoring and analysis of the dynamic characteristics of the plasma and related optical emission evolution have become effective methods for studying the plasma emission and characterization occurring on the timescale of ns as well as material removal process occurring on the timescale from 100 ns to µs [23,24]. Recent studies have focused on the online monitoring of laser ablation, involving research related to visual monitoring technology, acoustic emission, spectral diagnosis technology and thermal imaging technology. In terms of plasma dynamics and drilled depth, Bulgakova et al. [25] explored the radiative effects of the laser-induced ablative plasma on the heating (time interval was 20 ns) and ablation dynamics of graphite materials irradiated by nanosecond laser pulses. Consequently, they demonstrated that plasma radiative heating of the target considerably deepens the molten layer. Zhang et al. [26] investigated the spatial and temporal evolution of laser-induced plasma (time interval was 100 ps) with a transverse magnetic field using an ultrashort-gated intensified camera with temporal resolution of 8 ps and sensor sensitivity of single photon. The results indicated that the plasma characteristics, including its fluence distribution, dimensional size, and lifetime, were closely related to the ablation depth and machined surface integrity. The emission spectra of the plasma within 2.5 seconds were recorded and analyzed using an optical emission spectroscopic method during the laser drilling of Inconel 718 by Joonghan et al. [27]. The integration time of the spectral data was 20 ms and the spectral sensitivity was 0.035 nm. They examined the temporal histories of the plasma parameters, including the emission line intensity, electron temperature, and number density, with respect to the drilling depths. Lopez et al. [28] studied the ablation mechanism and matter-ejection of the GHz-burst laser drilling glass material sodalime within 200 ms by time-solved pump-probe shadowgraphy and thermal imaging with temporal resolution of 500 fs, revealing that the interaction process is fundamentally different for single-pulse and GHz-burst drilling. In terms of drilled defect evaluation, spark behavior within 0.5 seconds was observed under different cutting speeds through side visual monitoring with temporal resolution of 20 ms during laser cutting by Wen et al. [23], and the cutting speed was automatically optimized to obtain free defects and low bottom roughness. Xie et al. [29] adopted the acoustic emission (AE) analyses for fs-laser modification of silicon carbide with time interval of 0.6 ms and temporal resolution of 0.1 µs. Their results indicated that the AE signals were closely related to the material removal of SiC. Further, time-frequency analysis of the AE signals can be used to characterize defects in the surface form.

Although the above online monitoring methods have been successfully applied for the evaluation of drilled quality, in situ sensors are still limited by working distance, data acquisition speed, and various noises. There is an urgent need for the development of a real-time monitoring technology that can collect the entire processing information efficiently, including real-time material removal process and final drilled quality, and facilitate rapid and accurate reflection of the quality of the laser drilling process. The plasma emission information measured using a photodetector with a high detection rate (up to GHz) and small errors (<1%) enables an understanding of the laser process dynamics. Consequently, it can be used as an indicator of laser machining quality in practical industrial applications [30,31]. Based on the photodiode monitoring signals obtained during the laser cutting of aluminum and stainless-steel plates, Garcia et al. [32] determined a relationship between the monitoring signal, the quality of the performed cut, and the characteristics of the generated defects. Chen et al. [33] presented an optical detection system utilizing an optical confocal structure, which was experimentally confirmed to achieve >95% accuracy (drilled depth error <5%) for drilling holes with 15 µm in diameter and 100 µm in depth. The variation in brightness from the laser-induced plasma was used as an indicator to control laser percussion drilling by Ho et al. [34], who concluded that the intensity of the light emitted from the plasma plume correlates with the depth of the drilled hole. However, these studies were mainly applied in laser processing with low drilling velocity (4.2 µm/ms for [16]; 170 µm/ms for [33] and 0.08 µm/ms for [34]). To recognize different types of typical holes or drilling quality at such high drilling velocity during TSL processing (3.3 µm/µs for [9]), an in-depth study must be conducted to extract effective features from plasma signals.

In this study, the mechanism of TSL processing and the correlation between plasma signals and process performance based on high-speed optical monitoring (5 MHz) were explored to facilitate the realization of high-speed evaluation (1 ms timescale) of process performance using only plasma signals. A side-monitoring method using a high-speed camera was developed to capture and analyze the transient spatial information of the plasma (visible wavelengths) to elucidate the mechanism with temporal resolution of 200 ns and exposure time of 200 ns. Subsequently, the plasma characteristics, including the intensity distribution and evolution, were explored during the TSL drilling of the blind and through holes. Meanwhile, a coaxial monitoring method using a photodetector was developed to detect the rapid fluctuations of optical signals (200 ns timescale) during the TSL processing, which can be used for the high-speed evaluation (1 ms timescale) of hole types or drilling quality. Finally, the relationship between the plasma signal, plasma dimensions, and hole quality (depth, types, and cracks) was investigated and quantitatively evaluated under different laser parameters. The results are expected to contribute to the future development of monitoring systems that allow the high-speed evaluation and control of drilling quality in real time, particularly in laser processing with ultra-high ablation rate.

2. Method

2.1 Experimental setup

The experimental setup is illustrated in Fig. 1(a). A fs laser (Pharos, Light Conversion) and a µs laser (Red Power, SPI Laser) were used in the experiments to perform the TSL drilling. The fs laser pulses had a wavelength of 514 nm, pulse width of 180 fs, and repetition rate of 10 kHz. The pulse energy used for processing (pump beam) was 100 µJ. Whereas, the µs laser pulses have a wavelength of 1070 nm, a pulse width of 30 µs and a pulse energy of 6 mJ. The fs and µs laser pulses were aligned coaxially using dichroic mirrors (DM1 and DM2) and focused onto a glass sample surface using a near-infrared apochromatic objective lens OL3 (Mitutoyo; M Plan Apo NIR 5×). The effective numerical apertures of this objective lens for fs laser and µs laser were 0.038 and 0.048, respectively. An apochromatic lens was designed to compensate for the chromatic aberrations ranging from visible to near-infrared radiation. During laser drilling, plasma light was passed through the glass sample and then collimated using two objective lenses (OL4 and OL5) with a focal length of 75 mm (Thorlabs; AC254-075-A). One notch filter NF1 centered at 514 nm (Thorlabs; NF514) and one short pass filter SPF (Optosigma; SHPF-950) were applied to decrease the disturbance (transmitted and scattered light) caused by the incident fs and µs laser pulses. Thereafter, the plasma optical signal was collected using a photodetector PD (Hamamatsu; R10467U-50-01), and recorded using a data acquisition device DAQ (National Instruments; USB-6336) with a sampling frequency of 5 MHz. For real-time imaging of the drilling process including shadowgraphy of the material removal process [32] and monitoring of the plasma emission [27,34], illumination light produced by a laser Lamp (Cavitar; CAVILUX HF) with a wavelength of 646 nm and laser fluence of 3.98 × 10⁻⁷ J/cm² was passed through the sample from a direction perpendicular to the optical axis of the laser pulses for processing, which were collected using an objective lens OL6 with numerical aperture of 0.4 (Mitutoyo; M Plan Apo NIR 20×). Images were obtained using a tube lens OL7 (Thorlabs; TTL200) with a focal length of 200 mm and observed using a high-speed camera at a speed of 5 million frames per second (Shimadzu; HPV-X2) with temporal resolution of 200 ns and exposure time of 200 ns. Further, a bandpass filter was applied to decrease the disturbance caused by plasma emission during the observation of holes in the cross-section, whereas it was removed during the observation of plasma evolution. Following the experiments, background subtraction was performed using a background image captured prior to performing the experiments to cancel noise in the plasma images. To achieve an intuitive and simple quantitative analysis, the plasma images were cropped into images with a pixel size of 237 × 46 using the ImageJ software, and each pixel corresponded to 1.67 µm. Subsequently, the brightness of the plasma at different sample positions (P) throughout its lifetime can be represented by its 8-bit grey value, which is the average value over a region of 1 × 46 pixels. To ensure a high signal-to-noise ratio, the threshold of the gray value was set to 10, and only regions with gray values > 10 were selected and used for further analysis. The plasma signal datasets were collected coaxially and recorded using a photo detector (PD). Thereafter, the wavelet-domain denoising method was applied to reduce noise disturbance using MATLAB software with the wednoise function, wherein a posterior median threshold rule and LEVEL 10 were chosen. Furthermore, a charge-coupled device (CCD) was used to characterize the surface of the sample under white light illumination using two lenses (OL1 and OL2).

 

Fig. 1. (a) Schematic of the experimental setup. OL: objective lens (OL1-OL7); BS: beam splitter; DM: dichroic mirror (DM1 and DM2); NF: notch filter; SPF: short-pass filter; R: reflector; PD: photo detector; DAQ: Data acquisition device; CCD: charge-coupled device. (b) Timing of the irradiation of the fs and µs laser pulses. t₁: The pre-irradiation time of the µs laser; t₂: The irradiation time of the µs laser following irradiation of the fs laser pulse. (c) Focal planes of the fs and µs lasers. The relative position of the sample surface is defined as P.

Download Full Size | PDF

Only one pair of the fs and µs laser pulses was used for the TSL drilling, and the pulse width of the µs laser was adjusted using a delay generator (Stanford Research Systems; DG645). In addition, a delay generator was used to adjust the irradiation timing of the two laser pulses to ensure temporal overlap. As shown in Fig. 1(b), t₁ is defined as the time before irradiation of the single fs laser pulse, whereas t₂ is defined as the irradiation time of the single µs laser pulse following the irradiation of the fs laser pulse, thus determining the actual processing time. The laser-beam profiles and focal positions of the two laser pulses were measured using a beam profiler (SP620U, Spiricon). Despite the use of an apochromatic objective lens, the focal positions of the two laser pulses slightly differed. The focal plane of the fs laser was located 55 µm away from that of the µs laser, as shown in Fig. 1(c). It is worth to state that the small vertical shift (55 µm) between the two laser focal plane has little effect on the experimental results in this paper because fs laser is only utilized to induce the electron excitation region inside the glass, not for processing glass [14]. The focal spot diameters measured at the focal plane of the fs and µs lasers were 9.7 and 14.6 µm, respectively. The difference in the central-axis position of the two laser pulses was approximately 0–3 µm.

2.2 Experimental design and measurement

Non-alkaline glass (70 mm × 25 mm × 0.3 mm, AGC Inc., AN100) was selected as the sample. To further investigate the formation and dynamics of the plasma, drilled holes, and correlated optical signals, and thus evaluate the drilling quality under various machining parameters, different laser parameters, including fs laser energy (E₁) in the range of 60–200 µJ, actual processing time (t₂) in the range of 20–95 µs, sample position (P) in the rang of −40 to 190 µm, and total pulse energy of µs laser (E₂) in the rang of 2.4-6 mJ, were selected for the experiments. Sample position P is defined as the relative position of sample front surface to the focus plane of µs laser. The zero position of sample position P means the sample front surface is located at the same position of the focus plane of µs laser as shown in Fig. 1(c). To change the value of P, the position of the sample surface was precisely adjusted, and a separation P of −40 µm implies that the focus plane of the µs laser was located above the sample front surface. Furthermore, the entrance diameters, shapes, and depths of the drilled holes were determined using a laser microscope (Olympus; LEXT OLS4100) after cleaning the samples for 5 min using an ultrasonic cleaning machine to eliminate any unwanted deposition.

3. Results and discussion

3.1 Mechanism and related plasma characteristics

3.1.1 Blind hole drilling

To investigate the mechanism of TSL drilling, the evolution of the plasma, blind, and through-hole formation owing to constant t₁ 10 µs and different processing times (t₂ = 20 µs and 95 µs) was observed, as shown in Figs. 2 and 3. As shown in Fig. 2(a), a plasma cloud appeared above the glass surface (in air) with E₁ of 100 µJ, E₂ of 6 mJ, constant P of 100 µm and µs laser pulse duration of 30 µs (t₁ = 10 µs and t₂ = 20 µs). The intensity of the plasma in air (can be represented by its gray value) gradually increased from 0.2 to 1.2 µs, then rapidly decreased from 1.2 to 4.2 µs, and finally remained approximately constant till 18.2 µs. This is because at the beginning of the µs laser pulse irradiation, the ionized particles were ejected on a very short timescale and moved toward the surrounding area, thereby forming a plasma with an approximate shape of a mushroom cloud for spatial dispersion in air. The maximum intensity of the plasma at 1.2 µs originated from a subsequent thermal process [35]. With increase in the irradiation time of the µs laser from 1.2 to 20 µs, the induced plasma inside glass gradually extended as the increase in the hole depth, resulting from heat accumulation [36], increased, and the shape of the plasma was similar to that of conical shape of the hole. The spatially long plasma can be induced by ionized vapour materials distributed along the hole and be observed due to the plasma emission [37,38]. The plasma vanished at 21 µs and a blind hole with depth of 99.1 µm was formed. The x and z directions were defined as shown in Fig. 2(a), and the variations in the gray value (intensity) of the plasma along z and x directions are illustrated in Figs. 2(b) and (c), respectively. The gray value of each point in Figs. 2(b) and (c) correspond to the average gray value along x-axis for each z-value and along z-axis for each x-value, respectively. Two parts of the plasma were divided along the z direction: plasma in air (part I) and plasma inside the glass (part II). As evident, the largest intensity of plasma was obtained at 1.2 µs in air with a size of 38.3 µm × 23.3 µm. The largest intensity of plasma inside glass occurred at 16.2 µs; thereafter, the intensity of plasma continued to decrease with the increasing hole depth, which may be attributed to the increase in the effective irradiated area resulting in a decrease of the applied fluence [39]. This is also proven by the dropping drilling velocity, as shown in Fig. 2(a). As shown in Fig. 2(c), the high-intensity plasma was mainly located in the region (part IV) with a diameter of approximately 20 µm, which was approximately the same as the drilled diameter of 21.4 µm. Moreover, its intensity distribution was consistent with that of a Gaussian distribution.

 

Fig. 2. (a) Evolution of plasma plume and drilled blind hole in cross-section with µs laser pulse duration of 30 µs, including t₁= 10 µs and t₂= 20 µs. (b) Variation of gray value (intensity) of plasma along z direction. (c) Variation of gray value (intensity) of plasma along x direction.

Download Full Size | PDF

 

Fig. 3. (a) Evolution of plasma plume and drilled through hole in cross-section with µs laser pulse duration of 105 µs, including t₁= 10 µs and t₂= 95 µs. (b) Variation of gray value (intensity) of plasma along z direction. (c) Variation of gray value (intensity) of plasma along x direction.

Download Full Size | PDF

3.1.2 Through hole drilling

The plasma evolution with E₁ of 100 µJ, E₂ of 6 mJ, constant P of 100 µm and µs laser pulse duration of 105 µs (t₁ = 10 µs and t₂ = 95 µs) is shown in Fig. 3(a). A conical shape of plasma was generated and extended as the irradiation time of µs laser was increased from 2 to 80 µs and the filament was also observed at 2 µs inside the glass. The plasma below the bottom surface of the sample was induced at 90 µs, implying that the formation of a through-hole at this time. During the drilling process, most of the removed materials were ejected toward the top surface of the sample by the recoil pressure generated via evaporation [40], thus producing a long plasma inside the glass. Once the through-hole was formed, part of the material was removed from the bottom of the hole and ionized by the µs laser, and plasma was formed below the sample. The increasing gray value below the sample (Part III) in Fig. 3(b) confirms the presence of plasma. After the µs laser was stopped, a plasma above the sample was observed at 100 µs, which was composed of the ionized species ejected from drilling area [41]. As shown in Fig. 3(b), the dimensions of plasma inside sample with gray value larger than 10 gradually increased with an average expansion rate of 4.0 m/s from 5 to 50 µs. In addition, the approximately same average drilling speed of 3.9 m/s was achieved during this time. Moreover, the plasma with high intensity was mainly located at the region (part V) with a diameter of approximately 44 µm as shown in Fig. 3(c), which was nearly identical to the drilled top diameter of 46.2 µm. These results indicate a close relationship among the plasma, size of the drilled hole, and drilling speed.

3.1.3 Laser energy dependence on plasma characteristic and hole depth

The laser energy dependence of plasma characteristic and hole depth during the TSL drilling was explored with varying E₁ in the range of 60–200 µJ, varying E₂ in the range of 6–2.4 mJ, constant P of 100 µm, and t₂ of 20 µs, as shown in Fig. 4. As evident, the peak value of the plasma lengths and corresponding plasma images were approximately identical with different fs laser energy E₁ and constant E₂ from Fig. 4(a) and (b), respectively. This implies that the complete drilling process was rarely affected by the fs laser energy when t₂ was 20 µs. This is because the lifetime of the excited electron region (excited free electrons in the conduction band) induced by the fs laser was on the nanosecond timescale, resulting in the µs laser absorption into the electron excitation, which only occurred during the early drilling process [9,42]. Thereafter, the excited electron region disappeared, and the material was removed because of the similar heat accumulation effect of the µs laser absorption into the plasma, leading to an almost unchanged hole depth of 100 µm, as shown in Fig. 4(c). Regarding the relationship between the plasma dimensions and hole depth with different E₂, it is apparent that a lower E₂ usually implies a smaller plasma length and hole depth, as shown in Fig. 4(d). This is because a high laser intensity created by a large E₂ can produce high-density vaporized materials and a more violent ionization process, causing a larger plasma size. Thus, more laser energy can be transferred and absorbed by the internal materials, resulting in the melting or vaporization of large volumes of materials that flow out under recoil pressure and deep holes.

 

Fig. 4. Evolution of the plasma and the variation of hole depth under different laser energy (E₁ and E₂). (a) Evolution of plasma length. (b) Evolution of plasma plume in cross-section. The variation trend of dimensions of plasma and hole depth under different (c) E₁ and (d) E₂. The variation trend of hole shapes (diameter and taper) under different (e) E₁ and (f) E₂.

Download Full Size | PDF

Similarly, the variation trends of the hole shapes (diameter and taper) with different laser energies (E₁ and E₂) are shown in in Figs. 4(e) and (f), which were almost the same as those of the hole depth. This can be attributed to the E₁ exerting a minimal effect on the drilling process with t₂ on the timescale of µs, whereas the ionization process, melting or vaporization of materials, and final hole shape (diameter and taper) can be influenced by E₂.

3.2 Relationship between plasma features, plasma signal and hole depth

3.2.1 Sample position dependence on plasma signal and hole depth

The temporal and spatial evolution of the plasma and related variations in the optical signal under different P values were explored, which can be used to evaluate the drilling performance. The results with varying P, constant t₂ of 20 µs, E₁ of 100 µJ, and E₂ of 6 mJ are shown in Fig. 5. As evident, the plasma length continued to increase as the processing progressed, and then decreased rapidly when the processing time exceeded 18.2 µs. The largest plasma length during the drilling processes at different sample positions was captured by a high-speed camera at 18.2 µs, and the variation trend of the length at this time is shown in Fig. 5(b). The results indicated that with an increase in the sample position (P), the plasma length first increased and then attained the largest value of 116.7 µm at P of 100 µm. Thereafter, the plasma length gradually decreased at larger sample positions. This can be attributed to the longer and larger intensity of the excited electron region being induced at the proper sample position [42], thus resulting in an extended absorption region of the µs laser and a deeper depth.

 

Fig. 5. Evolution of the plasma and related variation of optical signal under different focus position (P). Evolution of (a) plasma length, (b) plasma plume in cross-section, and (c) gray value (intensity) of plasma captured by high-speed camera. (d) Amplitude variation of plasma signal captured by PD. (e) Relation of amplitudes of plasma signal obtained by PD and related dimensions of plasma and hole obtained by high-speed camera. The average amplitudes of plasma signal were obtained under the drilling time ranging as 19–20 µs.

Download Full Size | PDF

The amplitude changes of the acquired plasma signals obtained by the PD at different sample positions throughout the processing time are illustrated in Fig. 5(d), which shows a variation trend similar to that of the plasma length and gray value obtained by the high-speed camera, as shown in Figs. 5(a) and (c). Moreover, significant voltage variation occurred when t₂ was in the range of 0–1 µs (marked by a black arrow), and was owing to the light emitted from the plasma generated above the glass surface as shown in Fig. 2(a). As shown in Figs. 5(b) and (d), the corresponding variations in the peak amplitudes at different sample positions demonstrate a close connection with the drilling depth. To quantitatively evaluate the relation between plasma signal and drilling depth, the average amplitudes of plasma signal with different sample positions were calculated and analyzed for drilling time in the range of 19–20 µs. Consequently, a linear relationship between the average amplitudes of the plasma signal collected by the PD and the hole depth captured by the high-speed camera was obtained, as shown in Fig. 5(e). The larger values of the plasma signal corresponded to larger plasma lengths and hole depths under different sample position. The coefficient of linearity determination (R²) with high values of 0.94 and 0.97, indicated high consistency and reliability, indicating that the plasma signal collected by the photodetector in the timescale of µs can be used for high-speed evaluation of the change in plasma size and hole depth, without the need for time-consuming camera observations.

3.2.2 Processing time dependence on plasma signal and hole depth

The plasma phenomenon and related optical signals were more complex with larger processing time t₂ and drilling depth, particularly when a through-hole was generated. The results with t₂ ranging as 20–95 µs, constant P of 100 µm, E₁ of 100 µJ, and E₂ of 6 mJ are shown in Fig. 6. The peak value of plasma length and gray value of plasma continued to increase with larger processing time t₂ as in Figs. 6(a) and (b). As shown in Fig. 6(c), with the increase in processing time t₂ from 20 to 60 µs, the drilling depth exhibited an almost linear increase and similar high drilling velocity of 2.4 × 10³ mm/s (Phase I). Subsequently, the drilling velocity gradually decreased with larger processing time [43] and a through hole with drilling depth of 300 µm was generated when t₂ was 90 µs or larger (Phase II). Note that because of the smaller length of excited electron region (150 µm), compared to the length of Phase I (210 µm), the decrease of the drilling velocity is not relevant to the excited electron region, but to plasma shielding and energy loss of µs laser during the multiple reflections at the inner wall of hole [9]. It is evident that the peak amplitudes of the plasma signal in Phase I were almost same as shown in Fig. 6(d), and the amplitude of signal gradually decreased with the increase of t₂, which was attributed to the smaller peak plasma light emission induced owing to the weaker strength of the laser intensity on the axial direction as the drilling depth increased [44,45]. There was a slow increase in the amplitude of the plasma signal when drilling near the back surface of the sample, and a rapid increase when a through-hole was generated in Phase II. This is owing to a new plasma source occurring near the back of the sample, as shown in Figs. 6(c) and 3(b) and explained in Section 3.1. Thus, the plasma emission was enhanced, leading to an increased signal amplitude.

 

Fig. 6. Evolution of the plasma and related variation of optical signal under different processing time (t₂). Evolution of (a) plasma length, (b) the gray value (intensity) of plasma, and (c) plasma plume in cross-section obtained by high-speed camera. (Phase I: high drilling velocity with drilling time ranging as 20– 60 µs; Phase II: low drilling velocity with drilling time greater than 60 µs). (d) Amplitude variation of plasma signal obtained by PD. Relation of amplitudes of plasma signal collected by PD and related dimensions of plasma and hole captured by high-speed camera (e) in Phase I and (f) in Phase II. The average amplitudes of plasma signal were calculated under the drilling time ranging as 15–60 µs in Phase I, and as 60–95 µs in Phase II.

Download Full Size | PDF

As shown in Fig. 6(d), the corresponding variation of the peak amplitudes and slope of the signal curve ranging as 20–95 µs at different sample positions demonstrated the close connection with drilling depth. The relationship can be divided into two parts: Phase I and Phase II based on the changing drilling velocity and depth, as shown in Fig. 6(c). For a quantitative evaluation of the relation between plasma signal, plasma length, and drilling depth in Phase I, the average amplitudes of plasma signal with different processing time were calculated and analyzed for drilling time in the range of 15–60 µs. The linear relationship between the average amplitudes of the plasma signal collected by the PD and the hole depth captured by the high-speed camera can be obtained from Fig. 6(e). Larger values of the plasma signal corresponded to larger plasma lengths and hole depths. The coefficient of linearity determination (R²) with high values of 0.94 and 0.99 indicated the high consistency and reliability. Regarding the relation in Phase II, the average amplitudes of plasma signal were calculated when the drilling time was in the range of 60–95 µs. The fitting equation between the average amplitudes of the plasma signal and hole depth can be obtained from Fig. 6(f), and the high value of R² also proves the reliability. The above results indicated that the increase in the optical signal variation induced by plasma is associated with an increase in drilling depth under different processing time. Thus, a quantitative high-speed evaluation of the drilled depth can be achieved based on the collected optical signals and fitting equations. Because of its ability to significantly change the drilling depth under different t₂, the drilling process can be easily controlled in real time through online monitoring of plasma signals.

3.3 Relationship between plasma signal and cracks

During the ultrahigh-speed laser processing of glass, a crack-free fabrication process is required; therefore, a rapid crack identification method must be developed. The formation of cracks and related optical signal were obtained with t₂ of 40 µs, P of 100 µm, E₁ of 100 µJ, and E₂ of 6 mJ, as shown in Fig. 7. As evident, when the µs laser absorption into the excited electron region occurred inside the glass (red region in Fig. 7(b)), cracks were formed and expanded toward the top surface of sample in the range of 0.5–1.5 µs. Further, part of the material was separated from 2.5 to 5 µs, thus creating the black area (marked in yellow circle). The formation of cracks is attributed to the internal local pressure caused by the heating accumulation and expansion of the material [16]. The the tilted propagation of cracks maybe a random phenomenon, but it needs further research [46,47]. It is worth to mention that the precursor of the crack appearance after 1.5 µs is the local material removal inside the glass as shown in the sub figure of Fig. 7(b) at 1 µs. As evident from the plasma (white region) at 1.5 µs in Fig. 7(b), the propagation of cracks greatly extends the absorption region of the µs laser. This results in the high amplitude peak of plasma signal as marked in the red region, which is considerably more than that produced during the normal TSL processing (brown curve in Fig. 7(a)). This high plasma signal peak accompanying the appearance of cracks in the early stage of TSL processing is a useful tool for identifying whether cracks have appeared. As the drilling process continues, certain materials can be ejected from the absorption region range of 2.5–5 µs, and the absorption area continues to reduce, inducing dropping amplitude of plasma signal. When the drilling time was longer than 10 µs, no ionization process or material removal was observed, leading to a constant value of the plasma signal. This is because the existence of cracks (black region) blocked the subsequent propagation of the µs laser beam, and no drilling process occurred again.

 

Fig. 7. (a) Variation curve of optical signal during drilling normal hole and cracks with t₂ of 40 µs, P of 100 µm, E₁ of 6 mJ, and E₂ of 100 µJ. (b) Images of cracks formation.

Download Full Size | PDF

4. Conclusion

The mechanism of TSL drilling of glass was investigated by observing the spatiotemporal dynamics of the induced plasma. In the early stage, the fs laser excited region selectively absorbed the µs laser energy, leading to the generation of plasma. Subsequently, the heat accumulation effect by the µs laser absorption into the plasma contributed to plasma extension and material removal, which were significantly affected by µs laser parameters, including laser energy, focal position, and irradiation time. Furthermore, the drilling performance, including the hole depth, type, and quality, was quantitatively evaluated using real-time optical plasma signals. The variation trends of the amplitudes of the optical plasma signals were consistent with those of the intensity and dimensions of the plasma, as well as the hole depth under different laser parameters. The increase in the optical signal variation induced by plasma is associated with an increase in drilling depth under different sample position and processing time. This indicated that the plasma length and hole depth can be directly evaluated and controlled through monitoring of plasma signals. The peak amplitude of the plasma signal, which varies with different material removal processes, can be used to identify hole types (blind or through holes) and cracks. The revealed mechanism contributes to the further improvement of the performance of TSL processing. Moreover, the developed optical monitoring technology is useful for the high-speed evaluation and real-time control of machining quality during ultra-high-speed laser processing.