LDP for low cost flex: a fast reel-to-reel process allows direct patterning of thin metal films on a variety of flex substrates.

Low-cost PCBs and flex circuits are readily produced in volume by screen printing, but this process is currently limited to a practical resolution of about .00295" (75 microns). But in numerous applications, there is a growing need to push the resolution beyond this limit without adding excessive process complexity and cost. A new type of laser direct patterning (LDP) process meets this requirement and can already deliver resolution down to .00059" (15 microns). It has been successfully tested with several types of metals patterned on common polymer substrates including Kapton, Upilex, Kaladex, Melinex and Mylar. Because this LDP process is compatible with fast reel-to-reel web processing and batch processing, it is very cost-effective. It is already in production use by a medical sensor manufacturer patterning gold on PET films.

There are several methods available to pattern dense PCBs and flex circuits, including screen printing, photolithography and laser structuring. However, all of these have drawbacks that limit their use in low-cost, high-density flex circuits. Screen printing is limited to a practical resolution of around .00295" (75 microns). Microlithography can go far beyond this resolution, but involves substantial cost and complexity stemming from its multi-step nature, including wet chemistry. This makes it impractical for low-cost flex circuits, and particularly for disposable products.

Laser structuring has been successfully used to deliver line widths down to .00078" (20 microns) on FR-4/copper/tin PCBs. In this method, a focused, green laser beam is scanned over a substrate that is plated with copper and over-coated with a layer of tin that acts as an etching resist. The exposed copper is then removed with conventional chemical etching. The drawbacks of this process are slow speed, because the tiny laser spot has to be scanned over the entire surface, and the need for wet chemistry to remove the exposed copper. (1)

In response to these process limitations, engineers began to explore a quicker, mask-based ablation method, in which the time-consuming scanning and chemical etching processes are eliminated. This work eventually led to a new LDP process. (2)

The principles of this new LDP process are outlined in FIGURE 1. The process is based on excimer lasers, which produce pulses of high-energy, ultraviolet light that can remove target material by direct ablation (atomization), without causing thermal damage to surrounding substrate material. This is in marked contrast to the action of infrared and visible lasers that remove material by melting and boiling surface material. As a result, the excimer laser is capable of creating small features in thin or thermally delicate materials.

[FIGURE 1 OMITTED]

The first step is deposition of a thin (<150 nanometer) layer of metal on to the polymer substrate. The excimer laser beam projects the image of a photomask onto this metal surface. This photomask is a positive image of the actual circuit pattern and is several times larger than the target circuit. Several types of photomasks can be used, but chromium on quartz is a good match for this application. The projection lens serves to create a demagnified image of the correct final size.

Because the metal layer is very thin and metals generally don't have strong absorption in the UV range, most of the ultraviolet laser light passes through the metal film. But organic polymers exhibit extremely strong UV absorption characteristics, so all the remaining laser light is completely absorbed in the uppermost layer of the polymer, causing instantaneous ablation of the surface layer. This catastrophic vaporization, like a controlled explosion, also removes the overlying metal film, leaving a metal pattern that corresponds to the photomask in every detail.

Advantages of LDP

In extensive tests at LPKF AG we found that the LDP process can easily produce lines and other features as small as .00059" (15 microns), thereby meeting the growing market demand for circuits in this scale range. (3) Moreover, there is no reason why the process cannot be scaled for even higher resolution circuits as the applications evolve.

Most important, LDP requires only one dry process step to pattern a thin film metal layer instead of many wet chemical process steps in conventional photolithography.

Another advantage is its compatibility with fast reel-to-reel processing (FIGURE 2). Providing the laser fluence (energy density) at the work surface exceeds the ablation threshold for the substrate material (50-500 milli-joules/[cm.sup.2] depending on the polymer's chemical structure), then this is a single-shot process--each part of the circuit only needs to be exposed to one laser pulse. This means that for many polymers, an area of hundred square millimeters can be patterned with a single laser pulse. Furthermore, since excimer lasers typically have pulse repetition rates of several hundred pulses/second, LDP is compatible with high-speed, repetitive processing. In addition, the excimer laser pulse is very short (< 25 nanoseconds), which means that LDP can be used to create flex circuits in a continuous feed, reel-to-reel process with no need for stopping (FIGURE 3). Even at the high reel-to-reel feedrate of 1 meter per second, 25 nanoseconds corresponds to a "blur" of only 25 nm (00098" or 0.025 microns)--the motion of the web material is essentially frozen on the timescale of the laser pulse.

[FIGURES 2-3 OMITTED]

What does this mean in terms of production throughput? Let's assume that each laser pulse patterns an area of one square inch, and that this corresponds to a single manufactured circuit. At a reel-to-reel feedrate of 1 meter per second, 40 parts per second or 2,400 parts per minute can be produced. A reel with 1,000 meters of tape can be processed in about 15 minutes, yielding 40,000 laser patterned parts. Conceding some time to the machine operator for performing a reel exchange, it can be realistically concluded that three reels can be processed in one hour, i.e., 120,000 parts/hour. If such a system is operated 24/7, then this single laser workstation can generate more than one billion parts per year. And this huge volume of parts can be produced without any trade-offs in resolution, precision and feature definition.

Coupled with the process simplicity of LDP, its fast reel-to-reel compatibility is estimated to deliver a 20% cost saving over conventional screen printing.

An additional benefit of LDP is that it is environmentally friendly, utilizing no wet chemistry. This eliminates any issues with the handling and disposal of chemicals, plus its associated government regulations. Furthermore, when using the LDP process to pattern noble metals, they can be easily reclaimed in the form of a highly pure micropowder, which can be readily extracted from the exhaust filter system.

LDP also offers flexibility. The circuit layout can be changed simply by exchanging the mask. This means the same laser workstation can be used to produce multiple circuit types, even during the same shift.

Material Considerations

LDP is applicable to a wide variety of materials currently in use for flex sensor circuits, as well as polypropylene, polyethylene and polyvinylchloride. The properties of the most common materials are summarized in TABLE 1. In addition, all common metals and metals used in circuits and electronic sensors can he patterned using the LDP process. These include Cu, Ni, Au, Ag, Al, Cr, Si, Mo, W, Ni/Cr, Ta, Pt, Pd, Ti/Al, Ti/W, and Constantan.

However, our tests have revealed that successful implementation of the LDP process requires that the metal film meet certain conditions. Most important, the film thickness must be 150 nm or less. With thicker films, too much of the laser power is attenuated by the metal film before it reaches the polymer substrate. The film also must provide thermal stability during the coating process, careful control of the deposition process, and good adhesion to the sub strate. For example, it should meet the IPC-4204 tape test following climatic cycling and life testing. The film should also be reasonably free of defects.

These requirements mean that the metal film cannot be created using traditional chemical processes or by the application of metal foils. Instead, take advantage of vacuum deposition techniques, such as evaporation, sputtering, chemical vapor deposition and P-CVD. These are all well established methods that are commonly available. Finally, to improve adhesion and to increase climate stability, the incorporation of a tiecoat or primer layer based on chromium, nickel, titanium or molybdenum has also proved advantageous. (4)

Clearly, a key to this process is to minimize laser reflection and absorption by the metal layer and maximize absorption in the polymer. Put another way, complete single-shot metal removal can only occur if the focused laser power is above the ablation threshold for the polymer but below the ablation threshold for the metal. The ablation threshold of polymer materials lies within the range of 50-500 mJ/[cm.sup.2], and depends on the polymer's chemical structure as well as the laser wavelength.

The process can be optimized for this ablation threshold requirement in several ways. First, the excimer can be operated at a wavelength of 248 nm or 308 nm (as well as deeper UV wavelengths) in order to best match the absorption characteristics of the particular polymer and metal being used.

The laser power at the work surface can be optimized by varying the magnification of the projection lens and/or by reducing the laser power with an attenuating optic. For a given laser power level, increasing the size of the projected area on the worksurface reduces the fluence. It is desirable to make the projected area large enough so that the fluence is just above the minimum level necessary for processing. This ensures that the processing area has been maximized, which increases production throughput and lowers production costs.

However, the issue of ablation debris has also to be considered. At certain laser fluence levels, the laser beam creates plasma conditions that actually inhibit the deposition of debris onto the surface. Wherever possible, this condition should be met, in order to avoid the cost and complexity of using an additional cleaning step.

For these reasons, it is advantageous to work with a laser system supplier who can fully evaluate the process in their applications lab before any capital equipment is purchased. Such a supplier will help you choose the optimum polymer/metal combination as well as determine the optimum laser parameters.

Application: Medical Sensors

Recently, we were asked to develop and optimize this new reel-to-reel LDP process for a manufacturer of disposable blood sensors, with a target linewdith of .00059" (15 microns). The goal was to reduce the size of their sensors to allow increased functionality, as well as to reduce the amount of blood needed for a given test.

Gold on PET had been selected as the material combination for these low-cost, flexible circuits. Because of the material absorption characteristics, it was clear that optimum results would be obtained with a 308 nm excimer laser. The next step was to evaluate the action of the laser on uncoated PET. At a wavelength of 308 nm, a focused laser fluence of only 50 mJ/cm_ proved to be sufficient to reach the ablation threshold. Moreover, we found that each laser pulse removed about 50 nm of PET thickness, which corresponded well with ablation values from the literature. (5) In addition, 50 mJ/[cm.sup.2] is far below the ablation threshold for gold at this wavelength (about 1000 mJ/[cm.sup.2]). Just as important, earlier LDP evaluation had shown that ablating 50 nm of substrate was sufficient to ensure 100% removal of an overlaying metal film of under 100 nm thickness.

The next step was to work with gold-coated PET to optimize the gold thickness and the final fluence. We evaluated sample types with both 50 nm and 100 nm thick gold coatings. For the thinner gold film, good ablation results were achieved at a fluence of approx. 75 mJ/[cm.sup.2], while for the 100 nm thick gold film, the optimum ablation was achieved only at a fluence of >120 mJ/[cm.sup.2]. As the material characteristics are the same, this difference must be entirely due to laser power absorbed by the gold. To put these numbers in perspective, an industrial excimer laser (e.g., Lambda Physik Lambda STEEL 2000) generates 1050 mJ/pulse at 308 nm. So, a nominal fluence of 100 mJ/[cm.sup.2] can be reached while still using a projected image as large as 400 [mm.sup.2].

In contrast, when the laser parameters are not fully optimized, the metal is not cleanly removed, leaving warped edges. FIGURE 4 shows this effect in a comparison between circuits created by an optimized setup with a non-optimized setup.

[FIGURE 4 OMITTED]

Based on this process development at LPKF, the customer began successful volume production in 2002, using a proprietary thickness of gold as well as a proprietary value of the image size.

Application: RFID Antennas

Some high-resolution circuit applications demand a greater metal thickness than can be produced directly using this LDP process. These applications may also require layers of multiple metals. For example, we have recently been tasked with evaluating and optimizing the LDP process for a manufacturer of RFID antennas. We successfully showed that these could be economically fabricated by LDP followed by electroplating, and that both these steps could be performed as reel-to-reel processes. Specifically, we first patterned a 100 nm layer of copper on a flexible PET substrate followed by electroplating with up to .00039" (10 microns) of copper. Subsequent plating with .000078" (2 microns) of nickel and .0000039" (0.1 microns) of gold has also been demonstrated yielding a surface quality suitable for bonding processes.

The electroplating process inevitably produced some lateral additive growth of the circuit lines. Nonetheless we found that line widths of .00078" (20 microns) could be consistently produced, with very high definition and edge quality (FIGURE 5).

[FIGURE 5 OMITTED]

There are other advantages of LDP that make it attractive for many cost-sensitive applications. It requires only eight process steps, as opposed to 18 for conventional photolithography (TABLE 2). This simplification lowers process costs about 20% and also improves quality and yields.

On a broader note, this proven ability to produce thicker electroplated flexible circuits opens up a wide range of potential applications, from flexible chip-interposer to a broad selection of sensors. All these kinds of high volume products could benefit from the cost-efficient reel-to-reel LDP and subsequent electroplating process.

REFERENCES

(1.) "Direkte Laserstrukturierung--50um--Strukturen mit mindestens 80% Ausbaute," Krause, J., PLUS-Produktion von Leiterplatten und Systemen 3/2000, 395-398.

(2.) "Laser Structuring of Fine Lines," Meier, D. J., Agater, M., IPC Fifth Annual National Conference on Flexible Circuits, Conference Proceedings, Vol. I, pp. 95-99, Denver, CO.

(3.) LPKF Laser & ElectronicsAG, Osteriede 7, D30827 Garbsen, Germany, www.lpkf.com.

(4.) "The Effect of Moisture on Peel Strength of Adhesiveless Polyimide Laminates," Bergstresser, T. et al., Proceedings of the Fifth International Conference on Flexible Circuits, Sept. 15-16, 1998, San Jose, CA.

(5.) Srinivasan, R., Science 234, 1986, 559 ft.

DIETER J. MEIER is a senior scientist in R&D laser technology at LPKF AG; [e-mail]. ANDREAS BOENKE is a senior scientist in R&D laser technology at LPKF AG; [e-mail]. LUDOLF HERBST is product manager for Coherent Lambda Physik GmbH; [e-mail]. GERD SPIECKER is business development manager at Coherent Lambda Physik GmbH; [e-mail].

TABLE 1. Physical properties of common polymer substrates used in
high-density flex circuits

PROPERTY                       POLYIMIDE   POLYIMIDE     POLYIMIDE
                               (KAPTON)    (APICAL)      (UPILEX)

Tg [degrees]C                     385         500           500
CTE (ppm/[degrees]C)              15          12            12
Tensile Strength (Kpsi)           24         15-24          35
Water Absorpt. (% / Wt.)          2.9         22            1.2
Dielectric Strength (kV/mil)       7          9.4           6.8

PROPERTY                          PEN      POLYESTER   POLYCARBONATE
                               (KALADEX)    (MYLAR)     (MAKROFOL)

Tg [degrees]C                     122         80            153
CTE (ppm/[degrees]C)              20          20            70
Tensile Strength (Kpsi)           32         28-32         20-25
Water Absorpt. (% / Wt.)          <1          <1           0.35
Dielectric Strength (kV/mil)      3.4         3.5          2.8

TABLE 2. Conventional photolithographic
structuring vs. laser direct patterning

NO.   CONVENTIONAL            LASER DIRECT
      PHOTOLITHOGRAPHIC       PATTERNING
      FCB PRODUCTION          (LDP)
      STEPS

1     Cleaning/degreasing     Laser structuring
2     Micro-etching           Cleaning by high
        for resist adhesion     pressure water jet
3     1st rinsing             Rinsing
4     2nd rinsing             Activation resp.
                                removing Oxide
5     Drying                  Cu-electroplating
                                or electroless Cu
6     Photo-resist coating    1st rinsing
7     Curing/drying           2nd rinsing
8     UV-exposure             Plating of
                                functional
                                surfaces for
                                different
                                applications
                                (Cu, Ni, Au, Sn)
9     Resist developing
10    1st rinsing
11    2nd rinsing
12    Etching
13    1st rinsing
14    2nd rinsing
15    Stripping of resist
16    1st rinsing
17    2nd rinsing
18    Plating of
      functional
      surfaces for
      different
      applications
      (Ni, Au, Sn)