Toner transfer for PCBs having BGA packages on them!!!

OK, I'm a glutton for punishment.

Quite some time ago I purchased some small chips designed to drive a couple of LEDs at a constant (programmable) current. I was silly enough not to take enough notice of the package to order them in a BGA package.

This has spurred me on to try to improve my toner transfer technique to produce boards with traces of 0.1mm in width. For the imperial amongst us, that's a shade under 4 thousandths of an inch.

My existing printer 300dpi could not get near this, and another 600dpi printer was better, but not "better enough". When I saw some 1200dpi laser printers on special for $42, I decided to see if that would be good enough. I ended up as the proud owner of a Fuji Xerox P205B printer.

Initial impressions of the print quality were good, aside from a worrying issue of 0.1mm gaps between traces closing up on some areas of the page.

My first tests were done using a heat sealer to try to transfer the toner. This was not successful and it was apparent that I was going to have to apply more heat.

Tonight I did a second test. The aim was to check the degree of smudging of the image, rather than to produce a board, so I did not follow all the board cleaning steps. Principally, I did not take steps to remove any oil on the board, and used a "handled" printout that was both a little dusty and may have had fingerprints on it.

I used three types of paper, and 2 were clearly not working, so I'll concentrate for the most part on the one that did (a "home brand" glossy inkjet paper).

OK, you probably want to see some pictures... I'll apologise for these in advance. They're pretty blurry.



This is early on in the process as I'm stripping away the top layers of the paper (it's pretty heavy paper). The thick blob of paper at the top is where the tape was attached that held the paper in place. It prevents the paper from wetting and should really not be near the image area.



Already you can see the toner through the paper and a closeup of the board shows you that the toner image is pretty crisp. The BGA package is in the middle of the pattern on the right (it's a 14 pin package) and the pads around it are for 0805 components. To the left you can see some text with approximate sizes printed. The fine line around the edge is a 0.1mm line.

Also worth noting here is that I've flood-filled the ground plane and I've specifically NOT removed small regions of copper that get generated. Normally I would, but it provides even more fine detail. Note the thin sliver of ground plane going horizontally near the lower right corner.



Here's the board in the water bath soaking away.



After a bit more soaking and some gentle rubbing, the board begins to look like you could etch it.

However...



You can clearly see from this image that there is still a layer of paper attached to the board.

In this image you can also see the two less successful papers. If you look closely at the two failures, you can see that the paper has actually come away from the board more cleanly, albeit with more of the toner.

I have actually found that the coated papers work quite well. The coating helps attach the toner to the board and also helps seal it. But it is porous enough for the etchant to get through.



At this point I was worried that the temperature had not been hot enough to properly melt the toner and so I heated up a portion of three of the patterns with a reflow tool.

Would this make a difference?



Into the etchant.!

This is after a couple of minutes. The etchant is ferric chloride and everything is in a water bath that was around 70C when I started.

You can see a couple of things here:

1) it's a double sided board and the other side is etching slowly.

2) the sticky patch left by the tape is acting as a pretty good resist.

3) there are a couple of "bright spots" in the copper indicating that the etchant is not getting through.

4) clearly the etching is proceeding slowly...

I took the board out and rubbed various slowly etching areas (with a gloved finger) several times and this sped up the etching in those locations.



There's the etching tank. Not very high-tech. The yellow colour of the water-bath is due to light passing through the Ferric Chloride.



After etching the board looks pretty good. Note the missing patch in the lower left corner. That's also visible in the upper right of the image in the etching tank. This came off as I was removing the last layers of paper. Clearly using the reflow tool was a bad idea. Oddly enough it did not have this effect on the other papers.

This is unfortunately blurry, but you should be able to see the effect of undercutting that has widened the gap between traces and narrowed traces.

The next step was to remove the toner. This was done with acetone.



Here's a transmitted light image of the board with toner removed. There are clearly some pinholes but I suspect this may have been die to dust on the board or the paper. Looking at the surface texture of the copper, it doesn't show a lot of evidence of etching.



Sorry for the blurry image, but this shows the effect of undercutting, which is essentially over-etching of the board. The thin traces near and under the BGA package are thinner than they were drawn, and the spacing of traces is significantly wider.



Here is a closeup of my test area for long fine traces. I'm going to have to look up the details, but I think they are 0.1mm, 0.2mm, 0.3mm, 0.4mm, 0.5mm and 0.6mm traces. The spacing is likewise 0.1mm, 0.2mm, 0.3mm, and so on.

The pads at the ends of the traces are there to allow me to measure connectivity.

All but the thinnest trace show effectively 0 ohms resistance.

The thinnest trace shows 0.7 ohms resistance which is an indication that it has some very thin spots -- I was actually amazed that it wasn't broken.

The finer traces here also show narrowing and the gaps show broadening.

Note also that the horizontal lines are thicker than the diagonal lines. This is something to watch out for with toner transfer.

If you're at all interested, those traces at the top left corner are for some 1206 resistor networks which contain 4 individual resistors. These are effectively 4 0603 resistors stacked up together, but with smaller solder pads. These pads and the test points are for soldering practice

That's all for this evening. The printer has proven that it can produce usable results.

The next test will involve properly preparing the board, and using a fresh, unhandled print.

I was pointed to a "howto" on the web describing the modification of a cheap laminator. I'm not sure if I mentioned that I had tried a laminator, but it didn't get hot enough. Anyway, here is the article.

That laminator is on special at Officeworks here for about $20. THe extra bits come to about another $20 including freight. A colleague and I modified two of these laminators today. I won't even bother with pictures because the link has great photos and they're trivially easy to follow.

Anyway, so the first test with this was done with the laminator possibly not heated up all the way, and with 4 passes of the 1.6mm board, with toner transfer being done to both sides. Note that for these tests we did not even bother to stick the print to the board.

I'll give you one shot of the laminator.



Looks totally standard. For those of you noting details, it's turned on, but the heater isn't on.

The board was roughly prepared by cleaning with a dry green plastic scourer (in 2 directions) and then wiped down. Two strips of transferred image were placed on the board and it was run through the laminator 4 times board inverted each time). Yes, it was really hot each time it came out.

After this it was dunked into water and left for about 10 minutes.

The excess paper war rolled away by hand before finally rubbing most of the excess with my fingers.

This image shows the result. The white blobs are bits of paper mache from the removed paper.



The image here is very crisp and the toner is really well stuck to the board. I was pretty rough in cleaning off the paper.

Here are some closeups of various sections.



The text is clear and sharp, and you can note that below 1.8mm, the decimal point doesn't print!

Note the vertical scratch to the left of the characters? Maybe an issue with the printer. Better see how this turns out!



In this image, you can see that there are what appear to be specs of copper showing through one of the tracks in the lower right corner. Better see how that turns out!



This image really clearly shows that there's still a layer covering the toner and most of the rest of the copper to be etched. This is a potentially serious problem.

The etching was carried out with heated Ferric Chloride and a brush since this longer board wouldn't lie flat in the beaker I was using to etch in.

The use of a paint brush was recommended by the article linked to above (I have also used a sponge to accelerate etching)

In retrospect, the use of a brush was probably not a good move, as you will see soon...



The "effort" using the brush has tended to either remove the toner or promote over-etching. The 1mm characters look pretty poor here, as do some other areas which were subject to slightly vigorous brushing.



There may be a touch of under-etching on the upper end of the 0603 pad just below and to the left of the BGA package. The pad may be bridged to the trace that leaves from the top and turns right. It's inconsequential here.

Note that the pads still look crisp, but there is undoubtedly some undercutting that has occurred (look at the narrow traces leading to the BGA device). However the mount of undercutting seems less than for the previous test (whether that is due to relative under-etching is a matter for conjecture at the moment).



This is a very interesting image because it appears to show gross under-etching. It is very clear that the plastic coating of the paper remains totally covering the tracks here (there are a few places in the top right corner where it has peeled back.

And, doesn't it look like there is copper under there?



Here is a very clear break. These are pins with a 2.54mm spacing and tracks running between them. So the tracks are pretty large by the standards of what I use -- about 0.5mm. And this is a pretty big break. I should have spotted this before etching... or was it hidden?

Unfortunately I'll never know as I don't have an image of that side of the board prior to etching.



On the other hand, there are some 0.1mm lines that seem to have survived completely intact...



Here is a whole board that looks very clean.

Removing the toner was again done using acetone. The toner came off far more easily this time. Perhaps I had removed more of the paper covering?



This is the first look at a board. The fibres are from the cotton wool used to remove the toner (yes, I used my wife's nail polish remover)



This is the worse of the two of the boards of this design. The text is a little mangled, but aside from that it looks perfectly usable.



Here is a section of board that appeared grossly under etched. Clearly it's not grossly underetched, but it shows some signs of under etching. (It also shows some signs of overetching, and to see both is a bit of a worry)

There is some "cloudiness" in some of the etched areas and I'm going to get out a multimeter shortly to see if there is any conductivity in places there shouldn't be,

The overall conclusions are:

1) The new laminator works better than the iron!
2) Agitation with a brush may be a bad thing.
3) Care is needed to ensure correct amount of etching.
4) Need to increase size of very small pads/tracks to ensure adequate size after etching.

So I am on a continued path to find the best way to use this new cheap high resolution printer.

The previous tests show that the printer produces a very high resolution output and it is possible to transfer the toner.

The problems are:

1) getting it to stick
2) getting the rest of the copper exposed to etchant
3) reducing etching through the toner.

About three weeks ago I performed a number of other tests using a toner release paper and two different types of film.

Firstly a close-up of the image as it is printed. These are not the greatest, but will give you some idea of the resolution.

The first images are printed on toner release paper. The images are good enough in places that you can see the surface texture of the paper. Note that you do not see fibres. Also note that there are some areas where excess toner appears to have been deposited.




The images are quite mottled, and the toner layer is clearly very thin. It looks very much like there will be significant etching through the toner.

Next are the images printed on the blue film. On a normal laser printer you get a nice crisp black image. But not using the 205b. The images look pretty bad, very little toner appears to have adhered to the film. Also note that there appears to be much more toner in places that it isn't supposed to be.



Fine lines are not affected so much, but the thick lines here are only 0.4mm in width.



Here is the first toner transfer (still a little wet). As you can see the paper has come away completely, but that copper can be seen through the toner.



What you don't see is how the toner has come unstuck near the edges of the board.



Sorry about the fuzzy image. The fine line here represents the edge of my board, so these problems are outside the board area, but extend quite some distance from the edge of the board (up to 8mm or so).

This lack of adhesion may be due to adhesion, lack of cleanliness, or too low a temperature when transferring. Later investigation reveals that surface preparation is the major contributing factor (but I don't want to spoil the surprise just yet)

The blue film is an absolute failure. Note that the thinner lines are relatively unaffected, and the thicker lines have both more excess toner around them as well as voids within them. This is going to be an unhappy result, but I'm going to do it anyway.



Note that the blue film performs very well using other printers so I suspect the problem is something to do with the Xerox toner.

Here is another view of a board after the release paper had been peeled back (it actually lifts off). Note the adhesion issues around the edge of the board. This is not a particularly bad example!



This board was cleaned with steel wool, then washed in laundry detergent (good at cutting through grease) before being rinsed and dried using isopropyl alcohol.

The next step was to use the green film to seal some of the transferred toner. In contrast with the blue film, the green film is not printed to, it is placed over the toner image and the film is fused to the toner.



Note that there are some areas where the green film has not adhered, and that the green film tends to remain in small voids like dome of the very narrow gaps between tracks. Unfortunately because of the lack of adhesion of the toner to the board, removing this using tape risked removing the toner as well

Here is the green film that was removed



You can see some of the small pieces of toner removed and where the green film has incorrectly come away or remained on the board. The large pieces of toner that you can see are from another board, they are the large pieces of toner from around the edge of the board.

Some areas of toner transfer are particularly bad. The following area looks to have been affected by the paper moving slightly. It's not typical, but illustrates the requirement to keep the paper and board from moving relative to one another.



This problem is probably compounded by the fact that the board needs to go through the laminator at least twice before it adheres well enough not to very easily pull away. This suggests that a slower transport speed or higher roller temperatures (perhaps better controlled roller temperatures) would be an advantage.

Here is a wider angle view of the board. It's pretty clear that something happened at the right side of this board (which happens to be the "back" -- i.e. last to go through the laminator -- of the board as it passed through the rollers). The edges look pretty good on this board, but don't be fooled, huge chunks still fell off.



Those problems are in contrast to areas of the board where the paper did not shift.



The pads in the middle are for a BGA device that is 2mm x 2.5mm in size (It's an LM2795TLX if you're interested). You can see some copper through the toner, but the toner image looks sharp.

The etching was done with warm Ferric Chloride (around 40C) in a bubble tank.

Here are two boards that had a simple toner transfer image. They look pretty good.



Until you remove the toner...



Here is where you can see the extent of the problems with etching through the toner.

Here are two "film" boards. There are significant problems with both and for once, perhaps, the lack of focus is a blessing.



Once you remove the film and toner, the results are as poor as you might have expected. The press-n-peel blue film looks pretty nasty. As mentioned above, this is a toner issue, not a problem with the film.



The green film appears much, much better at first glance.



However, the areas where the green film remained on the board exhibit some significant under-etching (although not as bad as you might have expected having seen the image before it was etched)



Something that's not obvious here is how tricky it can be to get the green film applied without dust or creases. It's VERY thin and sticks to itself via static electricity -- which also attracts dust.

But there is no doubt that where the image is good, it is very very good.

The problems with creases can be seen here. A significant number of voids (possibly caused by dust are also visible.



Surface Preparation.

Several tests were done with each method (I haven't shown them all). One perplexing problem is that tests done on dirty boards always seemed to work better than the most highly cleaned boards. Whilst the former may have areas that totally failed (due to greasy fingerprints usually), they worked better in the "good areas".

This lead me to consider what I was doing when I prepared the surface.

One quick test was to dip the board in the ferric chloride very briefly (until the board turned salmon pink) hoping that the start of etching would produce very small surface pitting that the toner could stick to.

The first board turned black and mucky after being left to dry so I tried to clean off the muck. This muck proved to be very hard to remove. However, it made me think about boards that had worked -- they also had a darker colour.

I dunked the board in the etchant again and then left it in the sun to dry, determined to leave the board otherwise uncleaned. At that time the sun came out and the board darkened very rapidly -- clearly the etchant had made the board photo-sensitive (possibly due to copper chloride?) and the light was causing a reaction that produced a darker chemical (perhaps copper oxide?).

Here is what it looked like after only a few minutes



This looks pretty shocking, but the results after etching a toner only image look much better.



Note that the right end of the board was under kapton tape and did not get hot enough to melt the toner.

But also notice the edges of the board. The adhesion of the toner to the board is much better. In other areas of the board there was some failure to adhere, but it was limited to 2 or 3 mm rather than the 8 (ish) that was seen earlier.

The copper has clearly been etched through the toner, and there is some evidence that the toner has smudged a little.

The next tests will be to try the oxidised board with green film over the top. The better adhesion of the toner should allow more of the excess film to be removed.

This also suggests that some of the earlier techniques using plain paper could be revisited, however the time it takes to remove the paper backing tends to suggest that toner release paper plus the green film may be the way to go.

This is just here to hold the 2 missing images from the previous post