Scanning strategy-dependent etching rate in the formation of through-via holes by femtosecond laser-assisted etching

In this study, we used femtosecond laser-assisted etching (FLAE) to drill through glass vias (TGVs) in 0.3 mm thick non-alkali glass substrates. In FLAE, the focus of the femtosecond laser pulses is scanned to modify the material along a preprogrammed pattern, and the modified region is preferentially removed by chemical etching. We found that the scanning strategy affected the etching rate along the laser-modified lines. Among four types of scanning strategies tested, the strategy 〈du〉—that is, scanning in a downward direction followed by an upward direction—obtained the highest etching rate. In this case, the etching rate along the laser-modified line was approximately 10 times larger than that of the unmodified region.


Introduction
Glass is used in various scientific and industrial applications owing to its attractive chemical, optical, and electric properties.One of the promising applications of glass is as an interposer, which is used for three-dimensional (3D) integration of semiconductor devices.For interposer applications, it is necessary to drill a large number of through holes (thorough-glass vias; TGVs) in a glass substrate with a diameter of approximately 10-100 μm. 1,2)Drilling methods for TGVs include deep reactive ion etching, 3) powder blasting, 4) and electrochemical discharge drilling. 5)][8][9] For example, transient and selective laser processing, which use femtosecond and microsecond laser pulses, have been reported as a high-speed drilling method. 10,11)n this study, we applied femtosecond laser-assisted etching (FLAE), also called femtosecond laser irradiation and chemical etching (FLICE) or in-volume selective laserinduced etching (ISLE), to drill TGVs.FLAE is a two-step process.The first step involves the local modification of material properties by focused femtosecond laser pulses along a preprogrammed pattern by scanning the focus position, and the second step involves chemical etching.If etching durability decreases due to laser modification, the modified region can be selectively removed.Owing to the 3D nature of femtosecond laser modification, FLAE enables 3D subtractive processing on sub-micrometer scales for transparent materials.FLAE was first reported for photosensitive glass. 12)Subsequently it was reported for silica glass, 13) which is an important and widely used material in scientific or technological applications.Since then, FLAE has been investigated intensively.It has been shown to be applicable to several materials, including BK7, 14) soda-lime glass, 15) sapphire, 16) crystalline quartz, 17) calcium fluoride, 18) yttrium aluminum garnet, 19) and lithium niobate. 20)][23][24] The potential advantage of FLAE over other methods is its flexibility in the design of holes.For example, holes with bendings and branches can be drilled. 25)][28][29][30] In this case, laser modification can be performed quickly because no laser scanning is needed, but flexibility is lost.
In this paper, we report the application of FLAE for drilling TGVs on non-alkali glass substrates.In the course of our research, we found that the scanning strategy had a substantial effect on the etching rate along the laser-modified lines.Hereinafter, the term "scanning strategy" refers to the direction, number, and order of scans along a line.Usually, to modify the material properties along a line, the laser focus is scanned only once.For this reason, the scanning strategy has rarely been considered.We show that the scanning strategy plays an unignorable role in the etching rate in FLAE.

Experimental
The samples used were 0.3 mm thick non-alkali glass substrates.The substrates were cut into pieces of approximately 10 × 10 mm 2 before being used in the experiments.
Figure 1(a) shows the optical setup.The light source was a Yb:KGW femtosecond laser (Pharos, Light Conversion) operating at 1030 nm with a pulse duration of 290 fs.The laser beam was guided into an inverted optical microscope (IX-70, Olympus), and focused using an objective lens (LCPLN20XIR, Olympus; numerical aperture of 0.45).During laser irradiation, the sample was scanned using a 3D motorized stage [combination of ALS-6012-G1M (x-and y-axes) and ALV-600B-H1M (z-axis); Chuo Precision Industrial].To drill TGVs, the sample was scanned vertically (i.e.perpendicular to the substrate surface; parallel to the zaxis) during laser irradiation.Four scanning strategies shown in Fig. 1(b) were compared.In strategies 〈d〉 and 〈u〉, the focal point was scanned only once along a vertical line, whereas in strategies 〈du〉 and 〈ud〉, the focal point was scanned twice along the same line in opposite directions.In all the strategies, the laser focus was scanned to cross the top and bottom surfaces.The pulse energy, laser repetition rate, and scanning speed were 0.8 μJ, 200 kHz, and 100 μm s −1 , respectively.
After laser modification, the samples were etched in aqueous KOH solution for 24 h.The concentration and temperature of the etching solution were 5 mol l −1 and 80 °C, respectively.
During etching, the samples were held in a homemade support block so that the both surfaces of the modified lines were always in contact with the etchant, as shown in Fig. 2(a).The etching rate of unmodified region was measured to be 0.57 μm h −1 .
To measure the etched distance, samples after etching were cleaved near the laser-irradiated lines (0.3 mm away from the lines), as shown in Fig. 2(b), and was observed through the cleaved surface using an optical microscope.An example of the observed image is shown in Fig. 2(c).As shown in the figure, the black regions, representing the etched regions, had an estuary-like shape: thicker near the surface and thinner toward the interior.This is because etching started from the surface, and the unmodified regions near the surface were also etched.The etching rate was evaluated based on the etched distances measured on such images and the etching period.For each strategy, 10 holes were drilled to examine the variation in the etching rate.

Results and discussion
The etching results of the four scanning strategies are shown in Fig. 3. First, we compare the etching rates from the top surface.The average etching rate from the top surface in each strategy was in the range of 4-8 μm h −1 .Compared with the etching rate from the bottom surface, which will be described later, the difference among the strategies was smaller, and the variation within each strategy was also smaller.Among the strategies, 〈d〉 had a relatively low etching rate, with an average value of about 4.2 μm h −1 .
Next, the etching rates from the bottom surface is examined.The etching rates from the bottom surface showed a greater difference between the strategies than those from the top surface.In strategies 〈u〉 and 〈ud〉, the etching rates from the bottom surface were considerably lower than those from the top surface.In strategy 〈u〉, the etching rate from the bottom surface was zero in all 10 trials.In strategy 〈ud〉, the etching rate was zero in 9 of 10 trials.Consequently, the average etching rate was low, even though the etching rate in one trial (5.4 μm h −1 ) was as high as the average etching rate from the top surface.This suggests that the modified line around the bottom surface was difficult to etch, and that if this region had been etched the etching toward the interior would have progressed as fast as that from the top.What strategies 〈u〉 and 〈ud〉 had in common was that the direction of the first scan was upward.We speculate that the direction of the first scan determined the etching characteristics of the modified line around the bottom surface, although the cause is not understood.
In strategies 〈d〉 and 〈du〉, in which the direction of the first scan was downward, the variation in the etching rate within each strategy was small.In all trials, the etching rate was nonzero.In strategy 〈d〉, however, the etching rates from the top and bottom surfaces were obviously different.The average etching rate from the top surface was about 4.2 μm h −1 , while that from the bottom surface was about    1.5 μm h −1 .This indicates that the etching characteristics of the modified lines were vertically asymmetric.Compared to strategy 〈d〉, strategy 〈du〉 had higher etching rates in both directions, and the average etching rate from the top surface and that from the bottom surface were almost the same.Strategy 〈du〉 had the highest etching rate: 6.1 μm h −1 from the top surface and 6.4 μm h −1 from the bottom surface.These values were approximately 10 times higher than the etching rate of the unmodified region.As shown in Fig. 3(b), the region etched from the top surface and that etched from the bottom surface were almost in contact.Although this has yet to be confirmed, we can presume that through holes can be drilled using this method.
In strategy 〈du〉, the laser beam was scanned twice along a line; the first scan was downward and the second was upward.The second scan was probably the cause of the higher etching rate than that of strategy 〈d〉.We considered two possible reasons for this.The first reason is simply the higher number of scans, regardless of direction.The second possible explanation is that the upward direction of the second scan resulted in a higher etching rate.
To investigate this, we conducted an experiment with scans only in the downward direction.We measured etching rates with 5, 10, and 15 downward scans (denoted in Fig. 4 as 〈d5〉, 〈d10〉, and 〈d15〉, respectively).Figure 4 shows the results along with the result for strategy 〈du〉.As seen, more scans resulted in higher etching rates, but it took 15 scans to reach almost the same etching rate as in strategy 〈du〉.From this result, we conclude that the high etching rate achieved by strategy 〈du〉 was caused by scanning in the upward direction after scanning in downward direction.
It has been reported that in transverse-scanning (scanning perpendicular to the optical axis), the etching rate may depend on the scanning direction. 31)In this study, we observed such dependence in scanning parallel to the optical axis.Furthermore, we found that the scanning strategy significantly affected the etching rate.These findings have important practical implications.
We also observed that in strategy 〈d〉, the etching rate exhibited directionality.It has been presumed that the mechanism of selective etching involves densification, internal stress, and changes in ring sizes in silica glass. 13,32,33)owever, these mechanisms cannot adequately explain the directionality of the etching rate along laser-modified lines parallel to the optical axis.This implies the existence of a different mechanism of selective etching that can have directional etching rates.

Conclusion
In this study, we explored TGV drilling in non-alkali glass substrates by FLAE.We found that the laser scanning strategy played an important role in the etching rate along the laser-modified line.When the first scan was upward, the etching rate from the bottom surface was very low.Conversely, a combination of a first scan in the downward direction and a second scan in the upward direction (i.e.strategy 〈du〉) resulted in a high and isotropic etching rate.With this strategy, the etching rate along the laser-modified line was approximately 6 μm h −1 , which was approximately 10 times higher than that of the unmodified region.

Fig. 1 .
Fig. 1.(a) Schematic representation of the optical setup.(b) Schematic representation of the scanning strategies.The arrows indicate the movement of the focal point with respect to the sample.In strategies 〈du〉 and 〈ud〉, the focal point was scanned twice along the same line in opposite directions, although U-shaped lines were drawn in the figure.HWP: half-wave plate, PBS: polarizing beam splitter.

Fig. 2 .
Fig. 2. (a) Scheme of etching.(b) Sample cleaving after etching.The arrow indicates the direction of observation after cleaving.(c) An example of the image obtained after etching.The black estuary-like regions are the etched regions (empty regions) that appeared along the laser-modified lines.

Fig. 3 .
Fig. 3. Comparison of etching results from the top and bottom surfaces for the four scanning strategies.(a) Images obtained after etching.(b) Graph of etching rates.The error bars indicates the spreads between the minimum and maximum values (not standard deviations).

Fig. 4 . 3 ©
Fig. 4. Dependence of the etching rate on the number of downward scans.(a) Graph of etching rates.The etching rate of strategy 〈du〉 is also shown.(b) Images obtained after etching.Note that the images of 〈d〉 and 〈du〉 were shown in Fig. 3.