Appl. Phys. A 81, 753–758 (2005)

Applied Physics A

DOI: 10.1007/s00339-004-3024-0

Materials Science & Processing

z.l. li t.t. lin p.m. moranu

Thick polymer cover layers for laser micromachining of fine holes Institute of Materials Research & Engineering, 3 Research Link, Singapore 117602

Received: 15 June 2004/Accepted: 17 August 2004 Published online: 3 November 2004 • © Springer-Verlag 2004 ABSTRACT We have demonstrated that a thick polymer cover layer can significantly improve UV laser micromachining. The polymer cover layer acts as a wave-guide, concentrating the incoming laser beam onto the underlying substrate to be machined. With this method very low laser fluences can be used to drill fine, high aspect ratio holes in quartz, glass and other materials that are generally difficult to machine. We show that despite the improvements recorded, our experimental conditions are not optimized and the contribution of the cover layers in our experiments is less than what it could be. We suggest improvements to the cover layer material to improve the micomachining. The method has important advantages over conventional methods since a mask is not required and common laser equipment can be used to quickly and inexpensively form desired holes within the substrate. PACS 89.20.Bb

1

Introduction

Laser micromachining is an important industrial process. It consists of focusing a high-powered laser onto a substrate in order to form a crater by material ablation. With continued ablation the crater generally develops into a conical shaped hole. If the sidewalls of the hole are steep and smooth they may reflect a significant amount of the incoming light, concentrating it towards the ablation front at the tip of the hole [1, 2]. Evidence of this light concentration due to waveguiding comes from direct energy measurements [3] and the extremely high aspect ratio holes that can be formed. Further evidence comes from the fact that near the tips of these high aspect ratio holes they may become curved and continue to propagate for a limited time even though no light is able to strike the hole tip directly [4]. Our experiments confirm observations reported by Anthony and others. We have observed significantly bent hole tips in polymer substrates. These tips continued to propagate even though the bending was sufficient to shade the tip from direct laser light. Conventional UV lasers can be used to form these high aspect ratio holes in polymers. Tokarev et al. [5] used a 248 nm KrF laser to drill holes with an aspect ratio of 600:1 in poly(ethylene terephthalate) (PET). They also demonstrated high aspect ratio holes in polycarbonate (PC) and polyimide. u Fax: +65-6774-1042, E-mail: [email protected]

In this paper we introduce a method of using this waveguiding effect to micromachine small holes in materials that are usually difficult to machine. Results for crystalline quartz are presented and discussed throughout the paper; however, analogous results were found for other substrates such as glass. Conventionally these materials are machined using deep reactive ion etching (DRIE) or femtosecond laser ablation; however, such methods are slow and expensive and a vacuum is often needed. 2

Experimental

In our experiments a 266 nm wavelength fourth harmonic Nd:YAG pulsed laser installed in an ESI Microvia Drill M5200 was used. The laser beam has a Gaussian energy distribution and was focused to a 20 µm diameter spot at 1/e2 intensity. The full-width half-maximum pulse duration was 50 ∼ 60 ns. The experiments were carried out in ambient air at fluences between 1.3 J/cm2 and 10.8 J/cm2 . When drilling deep holes at high repetition rates the effects of the plume from a previous pulse trapped in the hole may become significant. To isolate such effects experiments were conducted at widely different repetition rates – 100 Hz and 12 kHz. We were careful to ensure that the laser pulse profiles for both repetition rates were identical. Since the laser pulse duration and maximum intensity changes with changing repetition rates, it is difficult to isolate and study the effects of repetition rate on laser drilling. In order to achieve an effective repetition rate of 100 Hz, we fixed the machine repetition rate at 12 kHz but scanned the laser rapidly and repeatedly in a straight line across the substrate surface. In between pulses the laser had moved to the next hole. By repeatedly scanning a line 800 µm long at speed of 360 mm/s each hole was subjected to an effective repetition rate of 100 Hz. Consequently the 12 kHz and 100 Hz experiments had exactly the same pulse profiles. The only difference found between the holes drilled at 12 kHz and those drilled at an effective repetition rate of 100 Hz was that the holes made at 100 Hz were slightly deeper after the same number of pulses. All experimental data presented in the paper refer to the 100 Hz conditions, however, the observations and conclusions are equally applicable to the 12 kHz experiments. Figure 1 shows the experimental set up. A 100 µm thick quartz substrate (Zhejiang Quartz Crystal Electronic Group Co. Ltd., type L) was covered with a 250 µm thick PC layer (GE Structured Products, 8010MC). The sides of both the PC layer and quartz substrate were polished to allow optical

754

Applied Physics A – Materials Science & Processing

FIGURE 1 The schematic experimental setup of thick polymer film assisted ablation in substrate. Not to scale

examination. The laser was focused on the upper surface of the PC cover layer. A hole was ablated through the cover layer and into the underlying substrate. The laser-drilled holes were examined using a JVC video camera, which was connected to a Nikon Eclipse ME600 microscope. 3

Results and discussions

Figure 2 shows high aspect ratio holes drilled in PC. By placing a PC cover layer on a quartz substrate, as shown in Fig. 1, the wave-guiding effect of these high aspect ratio holes can be used for machining. Typical holes drilled in quartz using this method are shown in Fig. 3 and average quartz hole dimensions after 500 pulses are shown in Table 1. As can be seen from Table 1, when the cover layer was used a 10 µm deep hole, 3 µm in diameter, could be drilled in the quartz using a laser fluence of 1.3 J/cm2 . The “laser fluence” is defined here as being the average energy per unit area of the beam hitting the top of the PC surface, i.e. at the laser’s minimum 1/e2 spot size (diameter) of 20 µm. Without the cover layer no damage whatsoever could be seen on the quartz surface. Through holes could be drilled through the 100 µm thick quartz substrates when a cover layer was used with a fluence of 5.3 J/cm2 (see Fig. 3c). Again at this fluence no damage at all could be seen on the quartz surface if the cover layer was absent. Only once the fluence had been increased to about 8.1 J/cm2 was there any sign of damage on

FIGURE 3 Optical micrographs of holes drilled in quartz using a PC cover layer. (a) Shows a plan view of a row of 5 µm diameter holes. The holes were drilled using a fluence of 2.7 J/cm2 . (b) Shows a cross-sectional view of a typical hole drilled using a fluence of 2.7 J/cm2 . (c) Shows a cross-sectional image of deep holes drilled at a fluence of 5.3 J/cm2 . No cracking or serious heat-affected areas were found. In all cases 500 pulses were used to drill the holes

Fluence With cover layer (J/cm2 ) Hole depth Hole radius (µm) (rq ) (µm)

Optical micrograph of holes drilled in PC using a 266 nm laser. The laser fluence was 5.3 J/cm2 , and the number of laser pulses used is written beside each hole FIGURE 2

1.3 2.7 5.3 8.1

10 40 100 59

1.5 2.5 4.5 5.5

10.8

87

7.5

Without a cover layer Hole depth Hole radius (µm) (rq ) (µm) No hole No hole No hole Barely perceptible Through hole (100 µm thick quartz)

No hole No hole No hole Barely perceptible 7.5

TABLE 1 Comparison of holes formed in a quartz substrate with and without a PC cover layer. The “hole depth” was measured from the top surface of the quartz to the tip of the hole. The “hole radius” was measured at the quartz surface. In all cases the holes were measured after 500 laser pulses

LI et al.

Thick polymer cover layers for laser micromachining of fine holes

the quartz surface if the cover layer was not used. This is more than 6 times the fluence required to drill the 10 µm deep holes when the cover layer was present. Drilling in the quartz using a PC cover layer proceeds in three distinct phases (see Fig. 4). In the first phase the extremely fine tip of the PC hole reaches, and initiates a hole in, the underlying quartz (Fig. 4a). Hole initiation in quartz is due to high local energy density of the concentrated beam at the hole tip combined with the effects of heat and plasma from the ablated PC that assist quartz ablation. As drilling continues the hole in the quartz deepens and widens. The hole in the PC also widens. The effects of plasma and heat on quartz ablation can be demonstrated by applying a thin coating of absorptive ink on the quartz. With the presence of such a layer damage can be seen on the quartz at fluences as low as 0.13 J/cm2 . It should be noted, however, that the damage is restricted to the quartz surface only. No significant drilling is seen until the fluence is increased to beyond 8.1 J/cm2 . In the second phase, the quartz hole deepens; however, neither the PC holes nor the opening at the top of the quartz holes widen further (see Fig. 4b). Figure 5 shows a graph of the hole diameters measured at the top surface of the quartz substrate and the corresponding bottom surface of the PC cover layer during the second phase. As can be seen in the graph, the data points for 200, 300, 400 and 500 pulses lie virtually on top of each other for both the quartz and the PC. This indicates that neither the quartz nor the PC holes widen significantly after 200 pulses. However, our measurements show that the quartz holes deepen significantly between 200 and 500 pulses. In the final phase, drilling stops and the hole depth remains constant even with additional laser pulses (Fig. 4c). This occurs because reflective losses eventually reduce the energy density at the tip to such an extent that ablation ceases [6]. As explained above the thick cover layer facilitates the initiation a hole within the quartz (the first phase of drilling) by wave-guiding the incoming laser light to a concentrated spot and possibly by the presence of plasma from the PC itself. But what role does it play in the second phase of drilling? To inves-

FIGURE 4 Schematic representation of the phases of cover layer-assisted drilling (cross-sectional view) not to scale. (a) The first phase of drilling. The hole reaches and initiates a hole in the quartz substrate. During this phase the hole in the PC cover layer widens and the hole in the quartz deepens and widens as indicated by the arrows in the figure. (b) The second phase of drilling. During this phase the hole in the quartz deepens, as indicated by the arrow in the figure. However, both the hole in the PC cover layer and the top opening of the quartz hole do not change size. (c) The third phase. Here the light reaching the hole tip is no longer sufficient to cause ablation, and drilling stops

755

Hole opening diameters at the backside PC (solid symbols) and top surface of quartz (hollow symbols) plotted against the laser fluence. The circular, triangular, square and diamond symbols represent 200, 300, 400 and 500 pulses, respectively. The graph shows that, for all fluences used, there is very little hole widening in either the quartz or PC after 200 laser pulses. In contrast, however, the quartz holes do deepen significantly during this time. The inset shows a step formed at the interface between PC and quartz

FIGURE 5

tigate this, the assumptions and calculations described below provide an estimation of the upper limits of the cover layer’s focusing abilities. We consider single specular reflection of the laser light within the cover layer. For our experimental setup (i.e., where the cover layer is 250 µm thick) multiple reflections occurring within the cover layer can be ignored for the following reasons. If a vertical column of light hits a sidewall that lies at α degrees to the vertical, the light will be reflected at 2α from the vertical. When this light reaches the opposite sidewall (a mirror image of the first sidewall) it will strike that sidewall at 3α and the reflected light will be 4α from the vertical, and so on as the light continues to be reflected between the sidewalls. If the angle α is large (say five degrees or more) the reflective losses from these multiple reflections become very large and their contribution to ablation at the tip is negligible. If the angle α is small (e.g., about 1◦ as in the case of our experiments) the cover layer would need to be very thick or r0 would need to be very small to allow multiple reflections within the cover layer. For example, multiple reflections can only occur in our cover layer (250 µm thick and α = 1◦ ) if r0 < 1.1 µm. Consequently for the situation of interest i.e., second phase drilling where r0 is relatively large, multiple reflections can be ignored. It is important to note that the calculation and discussion below are only concerned with the PC cover layer’s light concentrating capacity – whether this light actually enters the quartz hole and helps its propogation is addressed later in this paper. We further assume a vertical, collimated Gaussian incident laser beam with intensity of I0 . The radius of the incoming laser beam at 1/e2 intensity is ω0 . The radius of the hole at the bottom of the PC film is r0 . The PC sidewalls are inclined α degrees from the vertical, and the PC layer is 250 µm thick. Rref is the maximum radius at which the laser light can strike the sidewall and still pass through the opening at the bottom of PC with a single reflection. The relation between α, r0 and Rref is given by:
 r0 (tan 2α + tan α) r0 3 − tan2 α Rref = = (1) tan 2α − tan α 1 + tan2 α

756

Applied Physics A – Materials Science & Processing

where ω0 ≥ Rref > r0 . The radius of the hole at the top of the PC layer, Rtop , is given by: Rtop = r0 + h tan α

(2)

where h is the thickness of the cover layer. If losses are ignored (assuming light absorption, A, by the sidewall based on the Fresnel formulae, then A(α) = 2 .n α [4]. Here n is the real part of the refractive index. For PC n ≈ 1.6. At α = 1◦ , only 5% of E ref was absorbed by the sidewall. For larger α angles these losses do become significant and the data presented in Fig. 6 would show reduced wave-guiding capacities at larger angles; however, for purposes of understanding the concept it is not necessary to include these losses), the total energy reflected, E ref , by the sidewall to the hole at the PC bottom surface is given by: Rx E ref = 2π

rI0 exp(−2 r0

r2 ) dr ω20

  r02 R2x π 2 = ω0 I0 exp(−2 2 ) − exp(−2 2 ) 2 ω0 ω0

(3)

where Rx is equal to whichever is the smaller of Rref and Rtop . The energy, E 0 , of the light that directly reaches the bottom of the PC hole, i.e., without being reflected off the walls of the hole, is given by:   r2 E 0 = 2π rI0 exp −2 2 dr ω0 0    r02 π 2 = ω0 I0 1 − exp −2 2 2 ω0 r0

(4)

Figure 6 plots the ratio of the total energy reaching the opening at the bottom of the PC cover layer ( E total = E ref + E 0 ) to the energy directly reaching the opening without first striking a sidewall ( E 0 ) – here the thickness of the cover layer, h , has been taken as 250 µm. It indicates that E total /E 0 increases rapidly with decreasing r0 /ω0 and angle, α. For the

FIGURE 6 Graph showing the light concentrating capacity of the PC cover layer. The ratio of total energy (E ref + E 0 ) to the direct incident energy (E 0 ) plotted against the ratio of the radius of the hole opening at the bottom of the PC layer to the 1/e2 radius of the incident laser beam. The numbers written on the graph indicate the α angles. The PC cover layer is assumed to be 250 µm thick (as is the case in our experiments). The energies are calculated assuming single reflections within the PC cover layer. Reflective losses are ignored

case where α = 1◦ the ratio E total /E 0 drops rapidly when r0 /ω0 is greater than about 0.22. This is due to the cover layer thickness – some of the light that theoretically could be reflected to the bottom of the cover layer with a single reflection, if the cover layer were infinitely thick, does not enter the hole in the cover layer because under these conditions Rtop < Rref . The graph shows that under certain circumstances the cover layer concentrates the incoming light extremely effectively. For example, the E total /E 0 ratio is above eight when r0 /ω0 < 0.15 and α ≤ 5◦ . For our experimental conditions, the sidewalls of the holes within the cover layer are very steep, and we measure α ≈ 1◦ . So for our experiments, even when the opening at the bottom of the PC is relatively large there is significant concentration – for example, when r0 /ω0 = 0.4, E total /E 0 ≈ 3. Given this, the cover layer should be very effective in concentrating the incoming light during the second phase of drilling. The above calculations and discussions only consider the light concentrating capacity of the PC cover layer. Whether this concentrated light enters the quartz hole is addressed below. Soon after the quartz hole has been initiated Fig. 5 shows that the opening at the top of the quartz holes is consistently smaller than those at the bottom of the PC. Consequently, at the interface between the PC and quartz there is a step in the hole wall. What is the effect of the cover layer when this step is accounted for? A schematic of the situation is shown in Fig. 7. If the step is included, the ratio between E total and E 0 becomes:      R2 (Rx )2 exp −2 ωmin − exp −2 2 ω2  0  0 r2 + 1 − exp(−2 ωq2 ) E total 0    ≈ (5) E0 rq2 1 − exp −2 ω2 0

FIGURE 7 Schematic showing a cross-sectional view of second phase drilling in a quartz substrate. r0 is the radius of the hole at the bottom of the PC cover layer. rq is the radius of the hole at the top of the quartz substrate. Rtop is the radius of the top opening of the cover layer hole. The light reflected into the hole from the right hand side of the hole is shown (light reflected from the left is omitted for clarity). Rmin and Rmax are the inner and outer radii respectively of the annulus of incoming light that is reflected from the cover layer into the quartz hole

LI et al.

Thick polymer cover layers for laser micromachining of fine holes

where Rx is equal to whichever is the smaller of Rtop and Rmax (as defined in Fig. 7). Here E total and E 0 refer to the light entering the quartz hole. From Fig. 5 it can be seen that our step is about 2.5 µm over a wide range of laser fluences (the difference in diameters between the PC and quartz holes is mainly due to the materials’ differences in ablation threshold. Since the heat-affected zone is relatively small, only a small fraction of the difference can be attributed to the difference in materials’ coefficients of thermal expansion (i.e., due to differential shrinkage during cooling)). So assuming our experimental conditions of α = 1◦ , ω0 = 10 µm, a step size 2.5 µm and a cover layer thickness of 250 µm, the maximum radius at which the laser light can strike the sidewall once and be reflected into the quartz hole, Rmax becomes: Rmax ≈ 5 + 3rq , the minimum radius at which single reflection will pass through the quartz hole is Rmin ≈ 5 + rq and now Rtop ≈ 2.5 + rq + 250 tan α where all radii are measured in micrometers. The equation for the ratio between E total and E 0 becomes:      2 (5+r )2 exp −2 ω2q − exp −2 (Rωx2)  0   0 rq2 + 1 − exp −2 ω2 E total 0    ≈ (6) E0 rq2 1 − exp −2 ω2 0

Now E 0 is the energy of the light entering the quartz hole directly and E total is the total light entering the quartz hole (i.e., the sum of that entering directly and that entering due to single reflection off the PC sidewall). rq and ω0 are in micrometers. Figure 8 shows a comparison of the concentrating effects (measured in terms of the ratio between E total and E 0 ) under our experimental conditions (i.e., with the 2.5 µm step, (5)), and where no step is assumed (i.e., where rq = r0 – the answer is obtained by simply substituting r0 with rq in (3) and (4)). It

Graph plotting the ratio E total /E 0 against the ratio rq /ω0 . The energies E total and E 0 refer to the light entering the quartz hole. Here we have assumed our experimental conditions, i.e., α = 1◦ and ω0 = 10 µm. The “Without step” case assumes that there is no step in the hole between the PC cover layer and the quartz, i.e., r0 = rq . For the “With step” case, the step between the PC and quartz is assumed to be 2.5 µm (i.e., r0 − rq = 2.5 µm), as is the case in our experiments. The PC cover layer is assumed to be 250 µm thick (as is the case in our experiments). The energies are calculated assuming single reflections within the PC cover layer. Reflective losses are ignored FIGURE 8

757 r

can be seen from the graph that if ωq0 < 0.15 then the step actually increases the amount of light reflected from the cover layer into the quartz hole. When the radius of the quartz hole is bigger than this, the step significantly reduces the amount of light reflected into the hole. The reason why the step has an enhancing effect when the quartz hole is small but an inhibiting effect once the quartz hole becomes bigger can be seen from Fig. 7. When a step is present, it blocks light immediately adjacent to the quartz hole from being reflected into the hole. Consequently reflected light enters the quartz hole from an annulus. The inner and outer diameters of this annulus are determined by α and the step size (and of course by the size of the quartz hole). Since the beam has a Gaussian energy distribution, the light intensity hitting the annulus is lower than that of the light striking immediately adjacent to the quartz hole. But since the total surface area of the annulus is far greater than the area of the blocked light, under certain circumstances the sum of all the light reflected from the annulus into the quartz hole is greater than what would occur if no step were present. For our experimental conditions this somewhat counterintuitive situation occurs only the quartz hole is very small, once the hole has grown bigger the step inhibits the drilling since it shades potentially useful light from entering the hole (see Fig. 7). So the step may actually be useful in initiating the quartz hole, however, once the quartz hole has grown in size it hinders useful light from being reflected into the hole. This can be seen in Fig. 8 where the total energy entering the quartz hole for the “With Step” case is higher than the “Without Step” case for small holes. This situation reverses once the quartz hole radius increases. So during the second phase of drilling, where the radius of the quartz hole relatively large, our cover layer plays a relatively minor role in enhancing the drilling. From our measurements (not shown here), we find the quartz hole expands to its full radius very soon after it’s initation. During most of its developement it remains with the same hole opening radius ( Rq ) but deepens significantly (between 200 to 500 pulses, Fig. 5). However, Table 1 shows that very deep holes can be formed even with a relatively large radius. For example, when drilling at a fluence of 2.7 J/cm2 a quartz hole with a radius of 2.5 µm and a depth of 40 µm can be formed. Similarly when drilling at a fluence of 5.3 J/cm2 a quartz hole with a radius of 4.5 µm and a depth of 100 µm can be formed. Without the cover layer there is no noticeable damage to quartz at either of these fluences. So although the cover layer is doing little to help drilling during much of the evolution of the quartz hole (i.e., during second phase drilling) it still allows very deep holes to be formed. It appears, therefore, that the main role of our cover layers is in the initiation of a steep-walled conical hole in the quartz. Once such a hole has been initiated it is able to propagate itself due to its own wave-guiding ability. Although under our experimental conditions the cover layer plays little role in the second phase drilling, our calculations indicate that the cover layer could be tailored to provide continued assistance during the second phase. The desirable thickness, step size and α angle would depend on what substrate were to be drilled and what sized holes were required – for small holes a step may be desirable, for larger holes a smooth wall (i.e., little or no step) may allow the formation of deeper holes. We suggest that the step size may be tuned

758

Applied Physics A – Materials Science & Processing

using a cover layer made from a functionally graded material. The ideal cover layer should have the following features: it must be easy to initiate a wave-guiding hole within the cover layer, but through the thickness of the cover layer the ablation threshold should be increased until finally at its bottom surface it would have an ablation threshold necessary to form the required size step (the step size is a function of the difference in ablation threshold between the substrate and the cover layer). A graded cover layer would also allow a hole in the cover layer to have curved walls (i.e., where α changes through the depth of the cover layer) – this could significantly enhance the effects of the cover layer. One could envisage that such a graded cover layer could be achieved using, for example, a polymer film with a through-thickness gradient of SiO2 nanoparticles. A limitation of our PC cover layers is that they degrade at higher fluences. As can be seen from Table 1, the cover layer’s affectivity decreased at fluences above 5.3 J/cm2 (i.e., the holes drilled at 8.1 J/cm2 and 10.8 J/cm2 are shallower than at 5.3 J/cm2 ). If high fluences are required the cover layer material will need to be changed to more resistant polymers and eventually to a metal or ceramic material.

cused to achieve localized machining. However, if a lens is placed close to the work piece to achieve a tightly focused beam, it quickly becomes coated with ablation debris and is destroyed. We have demonstrated that a thick polymer cover layer can be used for UV laser micromachining high aspect ratio holes in materials such as quartz and glass. Holes as small as 3 µm in diameter (10 µm deep) have been drilled. No mask is needed since the wave-guiding occurs wherever the laser drills into the cover layer, and the laser optics can be far enough removed from the work piece that no coating occurs on the lenses. Under our experimental conditions – where the cover layer and substrate material have significantly different ablation thresholds – a step is formed in the hole as it passes from the cover layer into the substrate material. This step may actually assist in the initiation of the quartz hole, however, once the radius of the quartz hole is larger this step inhibits the ability of the cover layer to reflect light into the quartz hole. We argue that the drilling process could be optimized by functionally grading the cover layer through its thickness. REFERENCES

4

Conclusions

Laser machining microscale high aspect ratio holes in materials such as quartz and glass is extremely difficult using conventional techniques. Problems include the difficulty of using a mask to define the area to be machined since a laser beam that is intense enough to drill into quartz will quickly destroy most masks. If a mask is not used, a high intensity laser beam must be very tightly fo-

1 S. Y. Bang. S. Roy, M.F. Modest: Int. J. Heat Mass Transfer 36 (14), 3529 (1993) 2 P. Solana, G. Negro: J. Phys. D: Appl. Phys. 30, 3216 (1997) 3 T.V. Kononenko, S.M. Klimentov, S.V. Garnov, V.I. Konov, D. Breitling, C. Fohl, A. Ruf, J. Radtke, F. Dausinger: Proc. SPIE 4426, 108 (2002) 4 T.R. Anthony: J. Appl. Phys. 51, 1170 (1980). 5 V.N. Tokarev, J. Lopez, S. Lazare, F. Weisbuch: Appl. Phys. A 76, 385 (2003) 6 A. Ruf, P. Berger, F. Dausinger, H. Hugel: J. Phys. D: Appl. Phys. 34, 2918 (2001)