Most UV curing equipment manufacturers use a common sales pitch when selling UV-LED exposure technology: “It demands less heat, less money, and lasts longer” when exposing the UV-curing of inks, coatings and adhesives. Today, a German company, Radove GmbH, has introduced this technology to image PWBs, featuring a top and bottom scanning UV-LED array to expose photoresists. The selected UV-LED has a maximum energy output of 395 nm.

Compared to conventional mercury lamps, the potential advantages of UV-LED include instant “on” and “off” status, over 95% efficiency in wavelength output, the generation of significantly less heat (no IR emission of LED) and potentially long lamp life. In addition, arrays can be built providing a uniform energy distribution. By moving the array over the imaging area, the illumination can be equal throughout the total imaging area. Finally, new equipment with slim designs become feasible, since the UV-array can be placed in close proximity to the exposure frame.

Theoretical Aspects

Figure 1. Doped Mercury Lamp.

The radical-initiated polymerization of multifunctional acrylates is the base photo- reaction in the imaging transfer process of most soldermasks, etch- and plating resist.

The initiation of the acrylate polymerization by radicals is a complex reaction sequence. It starts with the absorption of a photon by a photoinitiator or sensitizer molecule. The photoexited sensitizer can transfer the energy to the photoinitiator. The exited photoinitiator decomposes under formation of two radicals. These are active enough to attack a carbon-carbon double bond present in the acrylate monomer and oligomers. Radicals formed by this reaction add to other double bonds and herewith start the chain reaction, ending up in a polymer network. The average chain length is restricted by recombination reactions of two radicals (including oxygen). The level of cross-linking strongly influences the solubility of the carboxy-functional resist matrix in alkaline, aqueous media. The carboxylic acid groups form a “soluble” salt in the alkaline environment of the developer.

In 1968 DuPont was first to market this chemistry as a dry film for lithographic applications under the trade name Riston®. In the 1990s liquid etch resists for the production of innerlayers and liquid soldermasks increasingly became popular. It addressed the challenging requirements for the emerging fine line technology.

The first step in photoinitiated polymerization is absorption of light energy by the photoinitiator and/or photosensitizer. This requires that absorption bands of the photoinitator/sensitizer overlap with the emission bands of the light source. The resulting absorbance at each wavelength represents a measure of the probability of the light absorbance. Currently used photoresists and solder masks contain photoinitiator/sensitizer system, which are optimized to absorb light in the 350 to 420 nm wavelength range. This corresponds to some of the major emission of high pressure mercury lamps used in exposure systems in the printed circuit industry (see Figure 1).

Figure 2. UV-LED.

The UV-LED light source has one single emission centered at 395 nm with a width at half maximum of 12 nm (see Figure 2). There are no other emissions neither in the UV, the visible nor the infrared region of the spectrum. This means also that the UV-LED’s emits a cold light.

The light of the lamp or LED should be absorbed by the resist layer. Using a conventional exposure system with broad emissions over the complete UV, visible and IR range, only a part of the light is absorbed by the resist. The light in the wavelength range below 350 nm is completely absorbed mainly at the surface of the resist layer because of the strong absorptions bands of the resist components in this part of the spectrum (see Figure 3). This light only cures the surface of the layer and does not penetrate into the resist. On the other hand, all the emissions over 420 nm will not lead to a photo-induced polymerization but will be absorbed and/or reflected by the substrate. The absorbed energy will subsequently be dissipated to undesired heat. The emission between 350 and 420 nm will partly be absorbed and mainly is responsible for the through curing of the resist layer.

The UV-LED equipment with its emissions centered at 395 nm fits very well the absorption properties of current dry films and liquid resists. The layers only absorb part of the light, which is optimal to cure the entire layer from top to bottom. Because of the absence of other emissions no additional heat is generated. Therefore the exposure with the UV-LED light source will much less distort the artwork than a conventional exposure system with multi-wavelength emissions.

There is another distinct difference of the exposure with the UV-LED system as compared to a conventional lamp. Light scattering at “particles” (micro bubbles and/or impurities) in the polyester cover foil of the dry film or in the resist itself, is dependent on the wavelength of the light. The intensity of the scattered light follows a l -4 law (l= wavelength of the light; Rayleigh law). This means the short wavelength light is scattered more then longer wavelength light. The UV-LED exposure system with no short wavelength light is therefore optimal to reduce light scattering. As a consequence, sidewalls of the resist may be smoother.

Application Tests

Figure 3-1. 19 Micron Thick Dry Film.

The BMA exposure systems operate with UV-LED with an emission maximum at approximately 395 nm. As discussed above, the absorption characteristics of dry-film suggest that the photoresists examined will cross-link upon UV-LED exposure.

A series of exposure and development tests were undertaken in order to find and document differences in the image transfer process The various photoresists were imaged in a side by side comparison with a standard exposure system, featuring a high pressure mercury lamp and the BMA system equipped with the 395 nm UV-LED. The standard exposure system was a Dynachem 520 printer.

The following test conditions were maintained in order to obtain meaningful and comparable results:

• Imaging of resist was accomplished within the same work shift (BMA and Dynachem)

• UV exposure dose was calibrated (Olec UVM-01 integrating energy from 330nm to 450nm)

• Resist were exposed with the same energy

• Hold times, prior and after exposure, were identical for each resist type

• Resists exposed on the two different exposure units were developed side by side in the same developing equipment under identical conditions

• Same test artwork was used for all exposure tests

The following three test artworks were used in order to identify potential differences between imaging technologies and resist types.

Stouffer Step Tablet T2115

Figure 3-2. 75 Micron Thick Dry Film.

The step tablet features specific areas of increasing optical density. UV light intensity impinging the resist decreases step by step. The optical density for each step and herewith the percentage of transmission is clearly defined [1]. For example, Step 5 only transmits 25% of the UV light onto the resist.

The use of the sensitivity guide is fairly simple. The results however, are extremely useful for this examination since the sensitivity tablet verifies the amount of cross-linking achieved with a particular exposure energy at given developing conditions. For example, if the resist appears unattacked (still glossy surface) under Step 5, this means that the cross-linking achieved with 25% of the UV dose of the total exposure energy was sufficient enough to cross-link the resist. Therefore it isn’t attacked by the developing medium under the conditions applied thereafter. If the same resist-type exposed with the same nominal energy, once with UV-LED 395nm unit and once with a specific high pressure mercury lamp, the step obtained directly allows to interpret, whether the exposure unit used, provoked less, the same, or more cross-linking. Clearly, the results are comparable only if the developing conditions were identical.

The results, summarized in Table 1, provide two read-outs from the Stouffer step tablet:

1. The highest step number, where the resist surface appears still glossy (not attacked, see paragraph above). The result is labeled SSG solid.

2. The first step, where the resist has been fully washed off from the substrate (Copper, Steel or Laminate) without leaving any visible resist residues. The results are listed in the SSG free column.

The first, fully developed step (SSG free) allows drawing additional information in respect to the resists cross-linking behavior. If for example, the UV-LED exposure would cause cross-linking on the surface of the photoresists, whereas the exposure with the mercury lamp would cross-link the entire photoresists layer, the SSG free value of the resist exposed with the mercury lamp would result in a higher number.

The difference between SSG solid and SSG free values also allows conclusions regarding the contrast, the adhesion to the substrate and the developability (solubility in developer media). However, the experimental setup has not been designed to investigate on this topic.

Stouffer Resolution Guide T-1

Figure 3-3. 10 Micron Thick Dried Liquid Photoresist.

This resolution guide features a pattern of linearly decreasing width of lines and spaces in a light - and a dark field. This pattern provides an indication of the resolution achieved in the image transfer process by identifying the line dimension, where the lines and space have been clearly developed. Interpretation of the resolution guide starts at 1 mil lines and space (25 micron) and allows sensible interpretation in 0.5 mil (12.5 micron) increments form there onwards.

In Table 1 the results are listed in [mil] in column SRG Resolution. The value noted is the minimal width of lines and spaces, where the resist pattern is cleanly developed.

For example, SRG Resolution 1.5 means, that where the artworks lines and spaces were 1.5 mils in width, the resist has developed, showing clearly defined lines and spaces. Below 1.5 mils, the space was not cleanly resolved.

It is important to understand that “SRG Resolution = 1.5” does not mean that the resist lines and spaces are 1.5 mil. All negative resists show a wider line and a narrower space. The results, however, are interesting and useful, if different resists are compared or by investigating on the effects of the two different light sources regarding the obtainable resolution of a particular photoresist. For example a 75 micron dry-film would show a significantly better resolution (lower SRG Resolution value) if exposed with a collimated light source as compared to a non-collimated one. If the BMA–UV-LED would for some reason be less focused or provoke more light scattering in the Mylar-foil or the resist itself, compared to the exposure with the Dynachem 520, this would clearly show a lower resolution capability in this test (higher SRG Resolution value). Additionally it is interesting to explore the difference in resolution capability of liquid etch resist versus dry-film, effects of different dry-film-thicknesses and the effects of the variation in exposure energy.

Figure 4. Picture shows the UGRA Wedge 1982 (Visit: www.ugra.ch).

The Stouffer resolution guide also features a dot pattern. This pattern has been assessed for this study. The resulting resist dot pattern is closely related to resist adhesion. It has 7 different arrays of dots, where the smallest dot array consists of an area of 1 mil (25 micron) and the largest is one of 2 mils dots. In Table 1 the results are listed under SRG dots present where 1 means the largest dots and 7 the smallest. If adhesion is poor for some reason (e.g. under-exposure, poor adhesion of the resist, not enough cross-linking on the interface resist to substrate), the finer dots 7 and 6, in severe cases even 5 or 4, will be lost.

The dots are a more sensitive indicator than a line pattern since the attack of the medium in the developer is from all directions. The result “Dots present 4-7” means that the dots are fully developed (standing as singular pillars) from array 4 up to array 7.

If the exposure energy is too high and/or the resist is insufficiently developed the arrays with larger dots are not cleared (normally at “SSR free” 13). The reason for not clearing the area between the dots below a certain size is that as the diameter of the dots increases, while the distance between the dots decreases.

The UGRA Plate Control Wedge 1982

Figure 5. SEM 10KV. DiaEtch 102, 10µm thick exposed with UV-LED at 100 mJ/cm².

Though the Ugra plate control wedge 1982© (PCW) is intended to control the plate-making process in the lithography, it features some geometric patterns, which are extremely useful to test the resolution capabilities of photoresists in the image transfer process.

In the middle section (Figure 4) the UGRA wedge features a series of highly defined micro-lines. There are 12 patches of positive and negative micro-lines between 4 and 70 micron. The widths are 4, 6, 8, 10, 12, 15, 20, 25, 30, 40, 55 and 70 micron.

Micro-lines created with a negative working photoresist give an indication on the smallest reproducible, fully intact line, shown in Table 1 as “UGRA best Line.” The positive micro-lines provide information on the smallest, truly developed space, shown in Table 1 as “UGRA best Opening.” The high variation of the values listed (best opening and best line) indicates that it is a very sensitive marker for differences between various photoresists and exposure conditions. As with the Stouffer Resolution Guide, but with even higher sensitivity, the reported UGRA values allow pointing out differences/similarities in performance regarding resolution capability. It is ideally suited to compare the two different exposure systems in question (BMA UV-LED versus Dynachem 520 printer).

Discussion of Results

Photosensitivity

Figure 6. SEM 10KV, 30° tilt, Magnification 500x. Hitachi Photec HN 340 Dry-film on stainless steel. This resist is used for Photo Chemical Machining (PCM).

The “SSG solid-” and the “SSG free”-values indicate whether a particular resist is equally sensitive to the exposure with the UV-LED or the mercury lamp unit.

Equal SSG values are an indication that the exposure energy creates a similar cross-linking within the resist matrix, even so the spectral outputs are different. DuPont PM 130 and Hitachi 6238 show 1 Stouffer-step more when exposed with the UV-LED unit. An increase of “one step” on the 21-step Stouffer tablet theoretically equals a 40% higher sensitivity.

This conclusion cannot be drawn regarding solder masks since the comparison--BMA vs. Dynachem 520--has not been included in this test series. The four solder masks could be sufficiently cured using the BMA exposure unit within a reasonable range of UV energy, i.e. 100 – 300 mJ/cm², depending on the type of solder mask.

It is no surprise that dry-films are more sensitive than liquid etch resists (LPR). This phenomenon is mostly due to the oxygen inhibition; LPRs have no cover sheet.

Adhesion & Resolution

SCHEMATIC

Great emphasis has been put on finding differences in the resolution capabilities between the two exposure systems. Behind these tests was the question whether the different spectral output of the UV-LED would lead to differences in curing the photoresist layer, particularly within the vertical cross-section. Such differences would immediately become apparent in differences in adhesion and resolution.

The results show the expected significant differences between photoresists, but could not be observed between the two exposure units. The thinnest dry-film was 19 micron, the thickest, 75 micron. It also can be noted that the LPR provides by far the best working window and resolution capabilities, even if compared to the 19 micron dry-film.

Again, the comparison between the two exposure-units reveals that no difference can be observed regarding adhesion and resolution. As far as the various values suggest, the achieved photo-polymerization within the resist-layer, as well as the degree of collimation of both exposure units, are found very similar.

SEM Analysis

Figure 7. SEM 10KV, 30° tilt, Magnification 1000x. Hitachi RY 3219 on chemically pre-cleaned copper.

It was of particular interest to examine the sidewall characteristics of the resists after exposure with the BMA exposure unit. Resist structures developed under different conditions were viewed in a scanning electron microscope (SEM).

The three resist types were exposed with the same UV radiation energy, 75 mJ/cm² for dry-film, 100 mJ/cm² for LPR, and 200 mJ/cm² for solder mask. It must be considered that the selected processing conditions as well as the developing equipment itself may not provide the best performance for a particular photoresist.

Figure 5 demonstrates the high resolution capability of the liquid etch resist DiaEtch 102. Isolated 30 micron lines can easily be achieved on a chemical pre-cleaned copper surface, while at the same time openings/spaces of less than 15 micron can be produced.

The sidewall appears to be perfectly straight. The waviness of the line may well reflect the quality of the first generation silver halide artwork.

The resist has a nominal thickness of 38 micron. Because of the 30^o tilt, this dimension appears to be about 50% of its true value only. The SEM photo was not electronically altered (no tilt correction). This fact applies to all following SEM photos and is explained by the following schematic:

This high resolution dry-film is 19 micron thick only (Figure 7). The sidewalls appear clean and straight. The high magnification reveals some waviness. The resist surface shows no dimples or structure. This is a good indication, that the applied exposure energy of 75 mJ/cm² caused sufficient curing.

Figure 8. SEM 10KV, 30° tilt, Magnification 500x. DuPont PM 130.

The DuPont PM 130 dry-film (Figure 8) looks rather impressive in respect to the appearance of its sidewall and the clean space. Again, due to the 30^o tilt, the sidewall is about 2 times higher than it appears on the SEM photo. According to the manufacturers product data, the resist thickness is 75 micron (3 mils).

The copper and the resist surface being well in focus (Figure 9), this picture demonstrates that the sidewall is homogeneous and of high quality.

Compared to other SEM photos of dry-film, surprisingly the sidewalls do not predominantly show the regular vertically streaks, normally observed with aqueous developed dry-films. It is believed that the vertical streaks are created from the light-diffusion in the artwork and the Mylar cover-sheet. It may well be that the narrow and defined spectral output of the UV-LED reduces the amount of light scattering in the artwork and cover sheet, leading to cleaner looking side walls with less vertical streaks. The resist shows good adhesion to the chemically pre-cleaned copper surface.

The photo (Figure 10) shows developed structures of the solder mask NPR 80. It was applied onto copper by screen-printing at a thickness of 20 micron (dry). NPR 80, Taiyo 4000 and Elpemer SD 2467 could be photo imaged using the BMA UV-LED without any surprises. Particularly the NPR 80 demonstrated good resolution capabilities, exceeding 50 micron lines/spaces.

Summary

Figure 9. SEM 10KV, 30° tilt, Magnification 1000x. DuPont PM 130.

All results indicate that the exposure with UV-LED @ 395 nm is effective with all tested photoresists. There is a clear indication that with equal energy doses from either exposure systems, the sensitivity of some dry-films is higher with the UV-LED exposure, meaning less energy is required to achieve a desired step-wedge rating.

It is also documented (see SEM photos) that the sidewalls of developed resist structures are lacking the vertical streaks when the LED light source is used. The current setup of the BMA UV-LED exposure equipment has demonstrated equal or better resolution of developed resist structures when compared to the Dynachem 520 exposure unit.

Outlook

Figure 10. SEM 10KV, 30°tilt Magnification 500x. Nippon Polytech Corp. NPR 80.

In the future it will be possible to increase the collimation angle of the individual UV-LED diodes. This will allow it to compete with standard collimated exposure systems in areas were collimation is required. Total imaging time now heavily depends on board size and the UV-LED array-area, which is moved over the area to be imaged. Coming array-areas will be larger, thus reducing the imaging times. It can be predicted that UV-LED will be commercially available, featuring UV-LED with shorter wavelength (365 nm) and higher emission energy.

The state-of-the-art BMA-exposure equipment capabilities are already impressive. With the above-mentioned upcoming improvements, UV-LED exposure units will find acceptance in the image transfer process in various areas, including the manufacturing process of printed circuit boards.

Acknowledgments

Advertisement

We would like to thank Dr. Piotr Domanowski, the inventor of the BMA-exposure equipment, for supporting this research. A special thanks to our customers who contributed substantially by providing the test materials and their expertise.

For more information contact HTP HiTech Photopolymere AG by visiting www.htp.ch.