1. Introduction

Nowadays, metallic thin films and coatings are an essential part of many advanced systems but also everyday devices [1]. Electronic devices, in former times solely based on solid printed circuit boards, are altered such that their functional structures can be fabricated on flexible substrates [2]. This allows for new fields of applications e.g. live science or medical products; as flexible systems can be conformably applied and integrated into all kinds of soft materials. Flexible substrates are often made from polymers. Most polymer foils are available in transparent or opaque form and with a wide range of mechanical properties. In addition, they show good dielectric properties and are relatively cheap to produce. For the fabrication of electronic devices mainly thermal and chemical robustness of the chosen polymer substrate have to be considered, because many microelectronic production processes are based on lithography and chemical etching [3]. The use of etchants, solvents or heat to cure photosensitive resins can reduce the selection of suitable substrate materials. As an alternative, the ablation by ultra-short pulse laser irradiation provides a maskless and very flexible process for structuring thin coatings on flexible polymeric substrates [4]. Due to its properties, this technology is used in industrial applications such as the production of touch screens, OLED systems, RFID tags, sensors and for EM shielding tasks [5]. Since very large areas can be processed, the structuring of conductive transistor electrodes for solar cell fabrication is another important area of application [6]. In general, highly energetic ultrashort pulses can be used to ablate material with marginal thermal input to the substrate, whereby the heat-affected volume in the vicinity of the ablation area is minimized [1,6].

The quality of micro laser patterning can be controlled via pulse overlap, fluence and scan repetitions [1]. For flexible electronic systems, usually certain areas of a conductive thin film coating are removed to yield the needed functional (sensor-) structures. Thus, complete galvanic insulation between separated structures has to be accomplished [7]. Precise movement of the laser beam along vertically stacked trajectory patterns allows areal ablation with pulsed laser systems. Despite the fact that single pulse ablation thresholds can be precisely determined with known techniques, the influence of scan parameters for line- and areal ablation is less investigated. Instead, trial-and-error parameter grids are mostly used in the field. Of course, these can also lead to a useful set of parameters whereby the ablation results are often evaluated according to rather subjective quality standards. However, the avoidance of substrate damage is crucial and is therefore regarded as the main optimization parameter in the context of this paper. For polyimide as a substrate, it can be shown that even minor substrate damage can be detected by purely optical measuring methods. The known methodology of ablation threshold determination evolved from rather bulky materials with the goal of cutting down into a substrate to a certain depth [8]. However, differences between thin film ablation and bulk material removal through laser induced evaporation of (bulk-) material cannot be ruled out [9]. Thus, methods and criteria allowing predictions for selective ablation results are still unclear or even missing.

Ideal selective ablation is achieved, when the thin film coating is removed without residues while the substrate stays untouched. This is especially challenging for metal coatings on susceptible polymer substrates. Because of this difficulty, authors in the past have considered small substrate damage (e.g. $1{\; }\mathrm{\mu}\textrm{m}$ abrasion from a $50{\; }\mathrm{\mu}\textrm{m}$ polymer foil after ablating a $150\;\textrm{nm}$ metallic thin film) inevitable to achieve proper electrical insulation [10]. As this is justified for some applications, it might limit the use of laser ablation processes when areal removal on transparent substrates should be accomplished (e.g. for solar cells). Even a slightly damaged surface may lead to a foggy, less transparent substrate here. Also, mechanical problems can arise from weakened substrates. To evade these limitations we will present a systematic methodology for laser parameter identification. As a result a selective ablation can be achieved, that can no longer be distinguished from the ideal selective case by optical methods. Our parameter selection approach is based on finding the sample/material specific ablation fluence threshold ${\phi _{\textrm{th}}}$ via multiple single spot D²-experiments. Two criteria will be introduced to predict ablation results such that the parameter search can be narrowed down beforehand.

2. Materials & methods

We chose the experimental setup, the inspection equipment, the sample fabrication method as well as the materials used such that results are application oriented and are easily reproducible at any time. To avoid influences caused by a wavy substrate foil, a spin coated polymer layer on a rigid carrier wafer was used.

2.1 Laser and inspection equipment

All laser based experiments were carried out using a pulsed Yb:KGW (Ytterbium-doped Potassium-Gadolinium Tungstate) solid-state laser system (microSTRUCT C, 3DMicromac), with a pulsed laser source ($\nu = 212\; \textrm{fs}$) emitting at a primary wavelength of $\mathrm{\lambda } = 1030\; \textrm{nm}$ in linear, horizontal polarization. Pulse frequency was 600 kHz for most ablation experiments. Beam quality is specified with $\textrm{M}^2\; \textrm{} < \; \textrm{}1,1$. The laser beam has a diameter of of $\; \textrm{}{\emptyset _{\textrm{Beam}}} = 4,7\; \textrm{mm}$ and can further be expanded using a Galilei lens setup. High speed laser beam positioning on the specimen is provided by a galvanometer-scanner (Scanlab intelliSCAN 14), while focusing is accomplished by a f-theta telecentric lens (Linos F-Theta Ronar) with an effective focal length of $100,1\; \textrm{mm}$ and a apodisation factor of $1,83$. The laser source was operated at a reduced power of 3W providing pulse energies between $0,28$ and $4,85\; \textrm{}\mathrm{\mu}\textrm{J}$. All testing was carried out under normal ambient conditions at room temperature.

Analysis of stationary spot experiments as well as inspection and evaluation of ablation quality was conducted with a confocal laser scanning microscope (CLSM, Keyence, VK-X260K). Its microscope objective provides a magnification of $150\textrm{x}$ at a numerical aperture of $0,95$ which enables morphological examination of the surface. In addition to the 3D height profile, a DIC (differential interference contrast) imaging was used to increase the contrast of smallest contours and edges on the surface. This technique creates a relief-like appearance to recognize otherwise invisible surface features.

2.2 Sample preparation

As mentioned before, we used polyimide substrates on rigid class carrier wafers for better handling. Polyimide foils were prepared by spin coating a liquid precursor (PI-2610/11, HD microsystems) onto the carrier as described by [11]. An adjacent thermal treatment was used for curing. Prior to sputter deposition, the foil surface was treated by inverse sputter etching to improve adhesion of the coating. A layer of $100\; \textrm{nm}$ of nickel-chromium (target composition: 80 wt% nickel and 20 wt% chromium) as piezoresistive metallic thin film alloy commonly used for strain gauge applications [12], was deposited on the substrate.

3. Calculation of fluence distributions

An areal ablation process with a pulsed laser source is accomplished by moving the beam in a scan pattern over the surface of the specimen. A combination of scanner speed v, pulse frequency f and spot radius ${\omega _0}$ determines the resulting pulse overlap parameter $\varphi $ (complete overlap at $\varphi = 0$ and no overlap for $\varphi > 1$):

 

Fig. 1. Gaussian fluence profile

Download Full Size | PDF

After scanning overlapping pulses along a straight line in $\textrm{X}$-direction, the resulting ablated cut width D can be thinner than the spotsize ${\emptyset _0}$ and depends on the specific material properties of the substrate.

In overlapping regions of two or more subsequent spots the fluence accumulates. For single laser lines the accumulated fluence profile can be gained by summing up the fluence distributions of individual pulses along its trajectory. The calculation of accumulated fluence profiles for areal ablation is more challenging, as the chosen ablation pattern has to be taken into account. We choose to work with a line pitch ${l_{P,y}}$ ($\textrm{Y}$-direction) of the stacked horizontal laser trajectories equal to the pulse spacing ${l_{P,x}}$ in the line scan direction $\textrm{X}$ (see Fig. 1: Gaussian fluence profile Fig. 2). This ablation pattern (${l_P} = {l_{P,x}} = {l_{P,y}}$) will subsequently be called “#-pattern”. It leads to an isotropic energy distribution.

 

Fig. 2. Pulse placement for an isotropic energy distribution. The red circles indicate the spot size of single pulses with their ${\omega _0}$ boundaries. Grey lines indicate the order in which ablation pattern is progressing (starting at the abscissa moving up, left to right).

Download Full Size | PDF

For illustration purposes, “uncovered areas” that are not irradiated (when ${\omega _0}$ is interpreted as setting a sharp irradiation boundary) are marked in Fig. 2: Gaussian fluence profile

Figure 2. In all our experiments the pulse overlap is chosen, such that full coverage of the ablation zone is achieved $0 < \varphi < 1\; \textrm{}$in order to avoid these uncovered regions.

3.1 Accumulated fluence profile (AFP) as an irradiation model

Eichstädt et al. have introduced an accumulated fluence profile (AFP) which is valid for laser engraved lines as well as spatially extended areas [13]. Thus, the fluence term includes x,y-coordinates shifted by multiples (${n_x},{n_y}$) of the distance between pulses (${l_P} = {l_{P,x}} = {l_{P,y}}$):

 

Fig. 3. Accumulated fluence profile (red) obtained from a line of laser spot intensities.

Download Full Size | PDF

To avoid substrate damage, local high fluence maxima should be avoided, in particular if they exceed the substrate threshold fluence. To account for the fluence variance the overshoot variable ${\theta _P}$ is established, which relates ${\mathrm{\Gamma }_{\textrm{AV}}}$ to ${\mathrm{\Gamma }_{\textrm{AV}}}$:

 

Fig. 4. Accumulated 1D fluence profiles (red) for line ablation with: a) No pulse overlap and b) Pulse overlap such that peak accumulated fluence is practically still equal to maximum spot fluence c) Pulse overlap such that valley accumulated fluence equals maximum spot fluence d) Large pulse overlap with accumulated fluence profile can be considered “flat” and overshoot indicator is small e.g. below a certain threshold.

Download Full Size | PDF

The fluence distribution in the case of 2D areal ablation patterns is illustrated in Fig. 5. Note that values for the accumulated fluence profile differ between line (1D)- and areal (2D)-ablation for the same value of $\varphi $. Assuming a standard areal ablation pattern in which laser lines (written in $\textrm{X}$-direction) are stacked above each other (in $\textrm{Y}$-direction), the accumulated fluence profile sums up by the contribution of successive pulses along the laser line as well as by pulses from vertically neighboring lines. Thus, $\mathrm{\Gamma }$ will always be higher for 2D accumulated profiles than for 1D accumulated line profiles assuming the same value of $\varphi $. A homogeneously spread accumulated fluence profile (small overshoot variable ${\theta _P}$) is required to determine well-defined process parameters allowing a controlled areal ablation.

 

Fig. 5. 2D areal accumulated fluence distributions (${\phi _0}$ = 1 J/cm², ${\omega _0}$ = 16 µm) with: a) Large gaps between individual spots such that fluence level locally drops to almost zero b) Overlapping pulses resulting in an accumulation of pulse fluences c) Pulse spacing similar to the pulse radius resulting in a practically homogeneous (flat) accumulated fluence profile far above ${\phi _0}$.

Download Full Size | PDF

3.2 Ablation pattern

Process parameters are considered well suited, if they allow complete homogenous ablation of a metallic thin film. At the same time, the underlying polyimide substrate must stay transparent indicating that surface damage is avoided. This is possible only if the accumulated fluence is homogeneously spread. To obtain a texture free homogeneous areal ablation, an isotropic and practically homogeneous 2D fluence distribution is required. The isotropic fluence distribution requirement excludes some of the ablation patterns that are in practical use. Here we use the #-pattern as shown in Fig. 2: Gaussian fluence profile Figure 2 where horizontal lines are vertically stacked at a pitch of ${l_{P,y}} = {l_P,x}$.

As seen in Fig. 6(a), the maximal pulse distance ${l_{P,\textrm{max}}}$ to completely cover an area with the "#-pattern” is given by ${\approx} 0,71\cdot D$. With increasing pulse overlap ($\varphi \to 0$) more area on the substrate is irradiated multiple times (e.g. Figure 6(b)). This makes the comparison of single pulse thresholds with ablation thresholds for line or areal ablation a challenging task as the incubation effect must be considered (see example of areal fluence distribution in Fig. 7).

 

Fig. 6. Illustration of #-pattern: a) Explaining the maximal pulse distance ${l_{P,\textrm{max}}}$ to prevent uncovered areas of the specimen b) Providing an example pattern where the borders of overlapping pulse areas and spot centers are marked. The white area is receiving fluence contributions from four single pulses.

Download Full Size | PDF

 

Fig. 7. 2D fluence distribution obtained with a “#-pattern” at ${l_P} = 20\; \textrm{}\mathrm{\mu}\textrm{m} = 0,63\; \textrm{}\cdot D\; \textrm{}$. As this is smaller than ${l_{P,\textrm{max}}}$, all areas between pulses are irradiated multiple times.

Download Full Size | PDF

3.3 Further relevant ablation parameters

Liu et al. stated that any laser ablation mechanism starts with the absorption of laser energy in the target [15]. Because metals show high absorptivity, light is fully absorbed well within the first micrometer below the surface [7]. Typical coating thicknesses are between 50 and a couple of hundred nanometers. For a pure nickel coating, penetration depth was found to be approximately 50 nm using Lambert-Beer’s law [7,16]. The coating on our samples was fabricated with a sputter target made with 80 wt% nickel and 20 wt% chromium, such that the penetration depth of nickel gives a good approximation for the NiCr alloy.

Pulse radius as well as threshold fluence are important factors for laser induced ablation processes. Liu et al. have found a linear relation between the square of the optically measurable diameter of the laser modified zone D and the fluence plotted on a logarithmic scale [17]:

Jee et al. have shown that material specific threshold fluence diminishes with increasing number of incident pulses [18]. This holds true for coating and substrate, which is why large pulse overlap ablations often result in substrate damage. The law of incubation can be stated as a formula, using the so-called incubation factor $\xi$:

4. Spot ablation

To describe the selective ablation criteria in case of thin film ablation the minimum level of the accumulated fluence profile ${\mathrm{\Gamma }_{\textrm{AV}}}$ will be related to the previously determined accumulated single spot threshold fluence. As the threshold fluence is dependent on the number of incident pulses, the so called incubation factor must also be determined. Thus, the material specific threshold fluence ${\phi _{\textrm{th}}}$ and the incubation factor $\xi $ of the coating were identified at first. This was accomplished with the aid of multiple D²-experiments where the coating was irradiated with different numbers of single laser pulses ${N_P}$ at increasing energy. In order to gain the material specific incubation factor, the threshold values for different numbers of pulses was plotted such that a line fit could be conducted. For these experiments the laser energy was configured, such that a single shot ${N_P} = 1$ can fully trepan the full coating thickness at the highest fluence level used. We used confocal laser scanning microscopy to measure diameter and ablation depth at the irradiated spot. The energy per pulse was varied in the range of $0,29\; \textrm{}\mathrm{\mu}\textrm{J}$ to $4,83\; \textrm{}\mathrm{\mu}\textrm{J}$, giving fluence values between $70\; \textrm{mJ}/\textrm{cm}^2$ and $1,11\; \textrm{J}/\textrm{cm}^2$.

4.1 Ablation zones

An ablation spot on a thin film coated polymeric substrate is not clearly confined and multiple characteristic zones can be observed. Figure 8 shows three distinguishable ablation zones. Outer areas with low fluences only show surface modifications and cracks in the coating. We can define the size of this influence zone by the expansion of the cracks. Higher fluences yield a ring-shaped area of selective ablation in which the substrate is unveiled but not damaged. This will be called selective ablation zone. For the highest fluence around the spot center a destruction zone evolves, which is recognized optically by a milky and/or rough appearance of the polymeric substrate. Fluence thresholds for each of the three zones can be determined.

 

Fig. 8. LSM micrograph (x150) showing three distinguishable ablation zones for a NiCr-coated sample (100 nm) with two different imaging technics: a) 3D CLSM image b) Differential interference contrast-image (DIC)

Download Full Size | PDF

4.2 Single pulse experiment

The development of the ablation zones at increasing laser fluence is illustrated in the subsequent micrographs of Fig. 9. In Fig. 9(a) the fluence is low so that only some surface modification of the coating was observed. With increasing fluence a clear influence zone with cracks started to appear (Fig. 9(b)). Close to the selective ablation threshold fluence the coating lifted up and formed a bladder (Fig. 9(c)). Also, the coating was sometimes completely removed in this area due to the strong particle suction system inside the laser machine. Further increase in fluence allowed to generate a clear selective ablation zone as shown in Fig. 9(d). The step in height between the coating (green area) and the unveiled substrate (light blue) was in the range of 100 nm which is the thin film thickness.

 

Fig. 9. Results of single pulse experiments with increasing fluence (left: 3D CLSM image / right: direct optical view) a) First material reaction b) Cracks form inside the influence zone c) The coating lifts up d) Selective ablation showing no substrate damage

Download Full Size | PDF

4.3 Multiple pulse experiments

Incubation experiments with multiple pulse repetitions (at $\varphi = 0$) were carried out with gradually increasing pulse energy where the increasing size of the influence, the ablation and the destruction zone was observed. For small fluences a cracked influence zone with a small selective ablation area in the spot center became visible on the surface (Fig. 10(a)). Note that laser fluence is equal to the one used for the single pulse experiment in Fig. 9(a). With increased fluence the diameter of the influence zone and selective thin film ablation zone in the spot center became larger (Fig. 10(b)). Solely some dust particles sometimes remained on the substrate. These could be cleaned using an ultrasonic bath. With even higher fluence, substrate damage started to appear (Fig. 10(c)). In Fig. 10(d) the destruction zone had expanded drastically such that the zone of selective areal ablation is narrowed.

 

Fig. 10. Results of multiple (${N_P} = 6$) pulse experiments with increasing fluence (left: 3D CLSM image / right: direct optical view) a) Large influence zone with cracks b) Selective ablation zone, dust particles remain but no substrate damage can be detected c) Substrate damage starts to appear d) Selective ablation zone ring narrows

Download Full Size | PDF

The measured diameters of different experiments were used to draw the D²-plot in Fig. 11. The plot shows only the data of the selective ablation zone. Selective threshold fluence for ${N_P} = 80$ is found to be ${\phi _{\textrm{th},80}} = 0,07\; \textrm{J}/\textrm{cm}^2$.

 

Fig. 11. D²-plot for a thin film selective ablation of 100 nm NiCr on polyimide. The diagram shows lines according to Liu's formula. They were obtained for each pulse repetition series by linear regression. Differences in the slopes result from measurement uncertainties and small instabilities of the ablation setup as the spot radius ${\omega _0}$ stayed constant during all experiments.

Download Full Size | PDF

The accumulated fluence ${\mathrm{\Gamma }_{\textrm{AP}}}$ that achieves selective ablation was calculated for stationary spot experiments by multiplying the selective threshold fluence with the number of irradiated pulses:

 

Fig. 12. 3D CLSM image proving that full selective ablation is achievable

Download Full Size | PDF

In section 3 we already calculated the gaussian spot size to be ${\emptyset _{0,\textrm{calc}}} = 40,1\; \textrm{}\mathrm{\mu}\textrm{m}$. In comparison, the evaluation of the D²-experiment in the selective ablation regime lead to a spot radius of only ${\omega _{0,\textrm{D}^2}} \approx 16\; \textrm{}\mathrm{\mu}\textrm{m}\; \Rightarrow {\emptyset _{0,\textrm{D}^2}} = 32\; \textrm{}\mathrm{\mu}\textrm{m}$. In Fig. 13, it can be seen that the D²-evaluation tends to yield higher values if the number of incident pulses increases. If plotted data is averaged for the three different zones, the radius of the influence zone ${\omega _{0,\textrm{D}^2}} \approx 20\; \textrm{}\mathrm{\mu}\textrm{m} \Rightarrow {\emptyset _{0,\textrm{D}^2}} = 40\; \textrm{}\mathrm{\mu}\textrm{m}$ correlates best with the calculated size.

 

Fig. 13. Plot of different zone radii evaluated from D²-experiments

Download Full Size | PDF

This means that influence zone data is most valuable to define the spot size, whereas selective ablation zone data must be used to define the threshold fluence.

4.4 Incubation effect

From our D²-experiments of NiCr we found the incubation factor for selective ablation to be ${\xi _{\textrm{NiCr},\; \textrm{sel}.\; \; \textrm{abl}}} = 0,53$. This is smaller than typical values for bulk material ablation of metals (see section 3.3). Therefore, we conclude that selective ablation of thin films is not only dependent on material evaporation but also majorly influenced by stress assisted interface destruction between coating and substrate. Figure 14 summarizes the threshold fluences for the three ablation zones defined in section 4.1 at increasing pulse numbers. It can be seen, that the threshold fluence of ablation and destruction zone are identical for number of pulses ≥ 40. Even though selective ablation can also be achieved in a narrow window at higher pulse numbers, we interpret this as a suitable upper boundary condition.

 

Fig. 14. Plot of threshold fluences for different ablation zones with increasing number of pulses at 600 kHz. Incubation factor $\xi $ can be derived from these curves.

Download Full Size | PDF

Thus, laser frequency, mark speed and pulse divider must be chosen in such a way, that the number of pulses hitting the very same position on the specimen stays below the upper boundary condition (${N_P} \le 40$ in this case). Otherwise, substrate damage is likely to occur.

4.5 Influence of laser frequency

The laser pulse frequency is not considered to influence the ablation threshold. Our experiments however revealed, that ablation behavior changes for higher numbers of pulses depending on laser frequency. This demands for an adjustment of the current well known methodology for ablation threshold determination developed by Liu et al. [2] which is not considering frequency influence. In Fig. 15, it can be seen, that the experimental selective threshold fluence for smaller number of pulses (1-10) was indeed frequency independent. For higher pulse numbers (≥20) however, threshold values obtained using equation [Eq. (8)] for 16 Hz and 600 kHz differed.

 

Fig. 15. Direct comparison of D²-experimental data confirms that ablation thresholds are influenced by laser frequency only at higher number of pulses ≥ 20.

Download Full Size | PDF

An incubation factor ${\xi _{\textrm{NiCr},\; \textrm{sel}.\; \textrm{abl},16\textrm{Hz}}} = 0,67$ for the selective ablation zone was obtained at 16 Hz pulse frequency. Figure 16 clarifies this phenomenon. As the threshold calculation is based on the measurable ablated zone diameter, the diameter values gained from the two upper rows of the figure were very similar. In contrast, in case of higher pulse numbers (the two lower rows) the affected zone is always larger for the higher pulse frequency. Also, the selective ablation and destruction zone have joined and cannot be perceived separately. Thus, the purely selective ablation is absent at higher fluences. This means that the transition from purely selective ablation to substrate damage becomes more sudden at higher pulse numbers.

 

Fig. 16. Comparison of multiple pulse experiments at increasing single pulse fluence. In case of ${N_P} = 8$ (two upper rows) the ablated diameter is similar and independent of laser frequency. At the bottom rows ${N_P} = 20$ the selective zone ablated at 600 kHz is always larger than at 16 Hz. Black areas indicate severe substrate damage where the crater is deeper than 200 nm.

Download Full Size | PDF

In addition, optically verifiable ablation happened at a fluence below the selective ablation threshold when pulse frequency is high (second row), even though the calculated threshold values matched for both frequencies. We assume this to be the results of different delamination effects. In case of low frequency ablation, the coating was “slowly” pulverized leading to an increased amount of residual particles and a poorly definable ablation zone border. For high frequency ablation though, the coating experienced a sudden flat lift and was removed at once. This is why the coating tends to bulge upwards at the boundary of the ablated zone at 600 kHz.

5. Selective ablation in scan patterns

From theory, there are two ways to achieve homogeneous line and areal ablation. On the one hand, the substrate can be irradiated with low fluence and a high pulse overlap resulting in a smooth accumulated fluence profile. In this case, the accumulated fluence level is much higher than the maximum spot fluence value ${\phi _0}$. On the other hand, lower pulse overlap values allow for higher single pulse fluence levels (which showed to be beneficial for coating removal) while the overall fluence variance (${\theta _P} < \; \textrm{e}.\textrm{g}.\; \textrm{}1\; \textrm{\%}$) is still acceptably low.

Suitable material ablation process conditions are typically found for a fixed scan strategy. The ablation threshold is then experimentally obtained using a parameter grid. In the following, we show that selective ablation for line- and/or areal ablation can be achieved based on the consideration of stationary spot threshold values only. This will be shown for scan patterns with practically isotropic and homogeneous fluence distribution. In the following, we will derive the according rules and show how they apply for the exemplary case of thin film NiCr ablation. We do not discuss multiple repetitions of the same ablation pattern as they do not improve the surface quality obtained after ablation. They are only necessary in cases of thicker layers.

5.1 Line ablation

Single laser lines structured into a metallic thin film coating can be used to electrically insulate certain regions of a coating. This can yield sensor structures or help to separate conductive areas. Because very little material must be removed, line ablation, which we call contour cut [19] is a time and cost effective technique. For line ablation laser pulses are placed along a line – each pulse overlapping its predecessor as defined by the parameter $\varphi $.

In line ablation, the average number of pulses ${\tilde{N}_P}\; $that hit a single position on the specimen can be defined as the product of pulse frequency f and the spot diameter ${\emptyset _0}$ divided by the mark speed v and pulse divider value $PD$. The pulse divider periodically guides a certain amount of pulses into a beam dump. The pulse frequency which is effective on the substrate can thereby be changed without changing the pulse generation frequency (600 kHz for our machine) and the single pulse fluence. $PD$=1 means that all pulses pass through the device such that they leave the source at their generation frequency:

 

Fig. 17. a) Line ablation grid. Inside each box frame, there are two ablated lines: Upper line is marked at 600 kHz, lower line is marked by placing individual pulses in a row yielding a slow frequency of 16 Hz (pulse fluence ${\phi _0}$ and the number of pulses on a single spot ${\tilde{N}_P}$ are the same for both lines). White numbers indicate cut width of lines [µm] ablated at 600 kHz. Side pictures: i) A fluence of 0,22 J/cm² and a mark speed of 2500 mm/s allowed for a selective line ablation using the standard 600 kHz provided by the laser source. In contrast, the line ablated at 16 Hz is intermittent ii) The upper line shows severe substrate damage while the lower line is still in the selective region. Edgeline quality of lower line is low iii) At these parameters the upper line is selectively ablated whereas no ablation is observed in the case of slow pulse frequency. b) Comparison of cut width D calculated for mark speed v at increasing fluence ${\phi _0}$ using equations [Eq. (8),(9)] and [Eq. (11)] with cut width measured optically (see Visualization 1 for high-resolution image of the entire figure).

Download Full Size | PDF

Three different areas are separated (indicated by no, yellow and red shading). The area on the left (no color) corresponds to conditions not leading to any ablation. The middle region (yellow) is the selective ablation area. White numbers inside the fields represent the measured cut-width of the upper line in µm. The red zone on the right marks those fields, where substrate damage is detectable by optical means. Only the lines written at 600 kHz determine to which area (no color, yellow or red) the corresponding parameter set of the box is assigned to. From the grid the single pulse fluence values ${\phi _0}$ at which selective ablation starts to occur with respect to the mark speed (e.g. ${\phi _0} = 0,16\; \textrm{J}/\textrm{c}{\textrm{m}^2}$ and mark speed of $1500\; \textrm{mm}/\textrm{s}$) can be read off. The border between either no or selective ablation is given by the following criterion (red line in the plot).

Selective ablation criterion for line-ablation:

The graph in Fig. 18 contains the threshold fluence values for selective ablation and substrate destruction determined in the previous stationary spot D²-experiments in dependence of the number of incident pulses. They define the window for selective ablation with ${\tilde{N}_P}$ as the decisive parameter in line pattern ablation.

 

Fig. 18. Logarithmic plot of fluence threshold values as obtained by stationary spot experiments (multiple pulses) which define the selective ablation window (yellow) for selective line ablation.

Download Full Size | PDF

The blue diamonds mark the fluence level ${\phi _0}$ in the plot where visible ablation started to occur in dependence of averaged number of overlapping pulses ${\tilde{N}_P}$. All these data points lie inside the proposed selective ablation zone. The knowledge of the pulse fluence ${\phi _0}$, the spot size and the threshold fluence corrected by the incubation effect allow the calculation of the resulting cut width D (see Fig. 1). The plot in Fig. 18 proves that experimental data and theory align.

5.2 Areal ablation using #-pattern

In order to predict areal ablation results the incubation effect plays an important role because the 2D-irradiation pattern causes exponentially more pulse overlaps than for line ablation. Depending on the ablation pattern (here we focus on the isotropic #-pattern as described in section 3.2), the number of laser pulse hits on a single position must be determined. This value can be calculated knowing the pulse spacings used in the experiment. In order to do so, for each laser pulse of the ablation pattern, a cylinder on the sample surface with a height of one is modelled. By adding up the heights of all distributed cylinders on the surface, the number of laser impacts at any position $Z({x,y} )$ can be determined as:

 

Fig. 19. Map and histogramm of the specimen. In this case, a #-pattern with ${\omega _0}$=16 µm and a pulse spacing of ${l_P}$=20 µm (results from a mark speed of 3000 mm/s, a laser frequency of 600 kHz and a pulse divider of 4) was evaluated. Those regions that were hit by most pulses are marked yellow, whereas those areas struck only once remain blue. The histogram on the right shows how large the surface fraction is (in arbitrary units) that is hit by a certain number of laser pulses

Download Full Size | PDF

From a histogram as shown in Fig. 19 the average number of pulse hits ${\tilde{N}_P}$ on the surface can be calculated (weighing them with their corresponding area). Depending on the pulse overlap, the mean number of pulses per spot always differs assuming a constant spot diameter.

We created an areal ablation parameter grid to test ablation thresholds determined by stationary spot experiments with practical data gained from areal ablation. Figure 20 shows this parameter grid using the #-pattern each field having a size of 1000 × 1000 µm. Fluence and mark speed were adjusted in the same manner as for line ablation (look back to Fig. 17). It can be seen, that areal ablation with a single pulse fluence of ${\phi _0} = 0,11\; \textrm{J}/\textrm{cm}^2$ caused interface destruction between coating and underlying substrate (field 3/3). This ablation mechanism was denoted “flaking”. The polymer substrate appeared very clear and bright after flaking, which indicates a damage free surface. Even though selective areal ablation in this regime seemed a promising method for areal ablation, it has to be taken into account that for geometric shapes more complicated than squares we observed flakes to adhere much stronger. In our experiments, flakes generated by areal ablation contours with straight edges broke away easily whereas for any pointy features (in larger contours), such as triangles, tips or thin lines, residual flakes usually remained on the surface. The narrower the gap between the untouched coating sidewalls, the more the ablated flakes in the middle adhered to the surface.

 

Fig. 20. Areal ablation grid with pulse divider 1 at 600 kHz pulse frequency (see Visualization 2 for high-resolution image). Fluence is increased left to right. Values on the left indicate the mark speed dependent mean number of pulses ${\tilde{N}_P}$ for each row. Blue area marks all those fields where the coating has lifted in flakes from the substrate. All fields inside the selective ablation zone show some unwanted greyish substrate color. This is because of residues and dust. Nevertheless, they are electrically isolating. The red area defines those fields where severe substrate damage has occurred. The five micrographs on the bottom enlarge some of the fields inside the grid.

Download Full Size | PDF

Right of the flaking region the ablation zone appears. Laser parameters for field 9/5 have yielded a complete coating removal, even though unveiled substrate is not as bright as for field 3/3 and 4/3 (flaking ablation). Laser parameters of field 11/5 caused a distinct substrate damage noticeable by the color change. Generally speaking, substrate damage came along with surface roughness scattering the through light. Thus, the darker the substrate appeared the more severe was the damage. Ablation by flaking could provide ideally preserved substrate surfaces but appeared to be inappropriate for many applications as flake residues could cause electrical short circuits. In the ablation area to the right of the flaking regime, however, the substrate appeared comparatively dark. The explanation can be found in Fig. 14, where differences in the threshold fluences for selective ablation and damage almost vanish when the mean number of incident pulses is equal or greater than 40.

A simple way to reduce the number of incident pulses per area is the use of a pulse divider. With this, the pulse overlap can be decreased without affecting the fluence. In addition, laser position accuracy is not influenced, as this would be the case using very fast mark speeds alternatively (also technical limitations for fastest galvo mirror speed must be considered). Figure 21 shows the corresponding parameter grid with a pulse divider of four. As the level of the accumulated fluence profile was much lower as compared to the case of pulse divider one (in Fig. 20) the single spot fluence could be much larger without causing substrate defects (therefor the grid was extended to the right). In this case five regions could be identified. Right of the flaking zone appeared a dotting zone (green area). “Dotting” means, that selective ablation was achieved only in the center of the pulses, but pulse spacing was such, that unveiled circles did not intersect and coating residues remained between them. There was a rather smooth transition from the dotting zone to the selective ablation zone. Laser parameters in the selective ablation area allowed flake free ablation of coatings without damaging the substrate. If one compares the substrate coloration in the flaking region (4/1) with the ablation results in the selective region (e.g. 11/2, 4/3 and 7/4), there is hardly any difference. No selective ablation results can be detected in the bottom line of the grid because the mean number of pulses (in this case ${N_P} = 73$) was too large, such that selective ablation threshold of the coating and destruction threshold of the substrate overlap in the incubation plot (see Fig. 14). As a result, flaking and destruction region are direct neighbors in the bottom row.

 

Fig. 21. Areal ablation grid with pulse divider 4 at 600 kHz pulse frequency (see Visualization 3 for high-resolution image). Fluence is increased left to right. Mean number of pulses ${\tilde{N}_P}$, pulse spacing, pulse overlap and mark speed are indicated on the left. Five ablation zones can be differentiated: no ablation, flaking, dotting, selective ablation and destruction regime. In the dotting regime the coating shows a uniform pattern of holes but retains areal conductivity. Two criterion boarder lines are marked inside the grid. The ten micrographs at the bottom enlarge fields inside the grid which are of particular interest.

Download Full Size | PDF

The first criterion, validated for the line experiments, directly compares the fluence with the pulse number dependent selective ablation threshold obtained in stationary beam experiments. The boundary given by this criterion is indicated in the areal parameter grid (Fig. 21) by a pink line. However, it may be seen that this criterion does not allow an exact delimitation of the selective ablation range. One reason amongst others for this deviation is a simplification in the calculation of ${\tilde{N}_P}$. The Gaussian profile of the laser was disregarded so far. Therefore, the accumulated valley fluence ${\mathrm{\Gamma }_{\textrm{AV}}}$ as defined in section 3.1 is considered to account for the Gaussian beam profile and will be used for a more precise criterion in the following.

Figure 22 shows a plot where the product of ablation threshold and the number of pulses is related to the accumulated valley fluence. The ablation threshold ${\phi _{\textrm{th}}}({{{\tilde{N}}_P}} )$ defines the minimal fluence which is needed to achieve selective ablation with a number of pulses ${\tilde{N}_P}$. Thus the product of the experimentally obtained selective ablation threshold and the mean number of pulses is equal to the minimal accumulated fluence level needed to achieve this ablation. This value can therefore be compared to the calculated accumulated valley fluence ${\mathrm{\Gamma }_{\textrm{AV}}}$. This calculation yields the second criterion in a rearranged form:

 

Fig. 22. Accumulated fluence plot. The accumulated valley fluence is depicted for different mark speeds in relation to spot fluence (blue dotted lines). Yellow shaded area marks the selective ablation zone. If accumulated valley fluence is larger than the product of mark speed dependent pulse number and the corresponding ablation threshold then selective ablation is achieved. (Refer to Fig. 19 for correlation between pulse number and mark speed e.g. ${N_P} = 18$ for 1000 mm/s. The corresponding threshold fluence can be drawn from the plot in Fig. 18.)

Download Full Size | PDF

The second criterion is added to the areal ablation grid indicated by a red line. As expected, it precisely meanders along the transition region between dotting and selective ablation. Thus, it can be used as a boundary condition to find suitable selective ablation parameters without the need for a large parameter grid. Furthermore, a set of parameters close to this calculated boundary can be considered to keep a safe distance from substrate damage. This provides a numerical value (distance to the boundary condition), which can be used to evaluate the ablation results in parameter optimizations when optically hardly any differences are perceptible.

In conclusion, we have experimental evidence that proves functionality of two simple criteria to predict ablation behavior in selective metallic film removal from polymeric substrates. The well-known first criterion, which compares the laser fluence with the pulse number dependent ablation threshold is valuable to forecast the ablation results of single spot and line ablation. In case of areal ablation this criterion is not piercing enough though. To predict at which laser fluence level areal selective ablation can be achieved we formulated a second, sharper criterion.

The described method for areal selective ablation promotes a combination of high-energy pulses and rather large pulse distancing to reduce the risk of substrate damage. However, a small pulse overlap can lead to noticeably rippled edges (as can also be seen in Fig. 23). Therefore, keeping the overall accumulated fluence low with this method is an appropriate way to achieve selective areal ablation on susceptible substrates but a subsequent smoothing of the contour with low fluence but high pulse overlap might be necessary.

 

Fig. 23. Micrograph of a pattern generated by successful areal ablation. Left side shows the sample directly after laser irradiation with parameters chosen according to the second criterion. Right hand side picture was taken after additional ultrasonic cleaning (to remove coating dust) at through light illumination only (thus areas with remaining opaque coating appear black). Any substrate damage would appear dark on the light yellowish polyimide substrate, but such destruction has clearly not occurred here. Because of the relatively small pulse overlap, minor positioning errors of the laser spot left some tiny coating flakes behind. Edgeline quality can be enhanced if areal ablation is followed by a contour line ablation with a larger pulse overlap.

Download Full Size | PDF

6. Conclusion

A method has been established to provide laser parameters for full selective ablation of a metallic thin film (NiCr) on a susceptible polymeric substrate (polyimide) in an efficient and reproducible manner. Thus, it can improve process optimization in research as well as for industrial applications whenever laser induced ablation of thin films is used for system fabrication. In particular, pulse overlap and pulse fluence have to be carefully controlled. The methodology is based on experimentally derived ablation thresholds and a calculated accumulated fluence profile to minimize overall energy input into the substrate. This may greatly reduce the risk of substrate damage. In practice, it may be sufficient to calculate ${\tilde{N}_P}$ for simple line ablations and work with the incubation dependent ablation threshold. In case of areal ablation, the second criteria must be considered. This allows to predict suitable parameters for an areal selective ablation (for methodology overview see Table 1).

Table 1. Methodolody of selective metallic thin film ablation

View Table