Dicing Solutions

As you may have read on this publication or perhaps noticed in the current trend in our technology, everything is becoming smaller.  There are however certain applications that require the dicing of extraordinarily thick materials.  This is the case of such materials as glass, quarts, ceramics, and some silicon.  These applications range from making imprint masks, to making heat shields, and substrates that require an unusual or uncommon thickness.  To put this in perspective the current standard thickness for any given material in the industry is about .021 – .029″.  We are in this article talking in specific of thicknesses of .100 – .250″.  When encountered by these types of thicknesses one faces a new set of challenges.

The first challenge, from the dicing point of view, is the blade thickness and exposure.  It is likely that when dicing something that is 1/4 of an inch in thickness, a 3 or 4″ diameter blade is needed.  Also the thickness of the blade needs to be relative to the thickness of the material.  The blade needs to be fairly thick in order to accommodate the spindle load that will be generated.  A thickness of .020 – .030″ is recommended for the blade.  Now there are a couple of different techniques that one might use to accomplish this task and they will require different procedures  as well as yield different results.

Multiple passes

In fact, it is recommended that when dealing with thicknesses of .060″ and over, the dicing be done in multiple passes.  This technique requires the blade to make several passes while gradually cutting the whole of the material.  In other words the blade makes the first pass at say .015″ into the material; on the second pass it knocks out another .015″ and so on until the blade reaches beyond the tape and the cut is completed.  This is very time consuming since for every cut that needs to be made, it has to be multiplied by the number of passes.  Thats the obvious problem but another is that, when dealing with extraordinarily thick materials, the blade tends to deviate during the first passes because of lack of support.  Along with the spindle load created while trying to cut through this material and the amount of exposure that is not being used, it is the perfect recipe for wavy cuts.

This can be remedied by using a different scribe blade.  Instead of using a blades that gives you the entire exposure needed, we recommend using one that has only about .050″ of exposure.  Scribe .040″ into the material and make the trenches for all the cuts that need to be made.  Once this process is done, switch to the blade that has the entire exposure and dice in multiple passes over the trenches that were created earlier.  The second blade will tend to follow the scribe cut alleviating any waviness.

Another issue is the bottom of the cut or the back side chipping as we call it.  It tends to be augmented when dicing thick materials.  Some of the causes are, improper cut depth,  improper RPM or feed rate, and vibrations that occur while the last chunk of material is being knocked off on the last pass.  There are no real cures for this, backside chipping is an immortal archenemy of dicing.  To improve these unwanted happenings the right thickness of tape is required in order to acheive proper cut depth.  Preferably a tape that is sturdy and does not allow for much vibration.  We recommend using a RPM of 15K for 3 and 4″ diameter blades.  20K for 2″ blades and a feed rate of about .02″ per second depending on the specs and material.  For glass and quarts we recommend using 325 grit size, for ceramic 270;  and for silicon 3000 or larger.

Cutting from both sides

This second technique is our finest weapon for fighting backside chipping, but in turn it requires extreme precision and it often produces results that might or might not be an issue to the final product.  This process essentially requires that the material be scribed from both sides.  This, as I said, is very efficient at reducing or in some cases eliminating back side chipping simply because the blade never dices through the entire thickness at once.  In this process there is only need for enough exposure to dice say 2/3 of the way through the material.  The wafer is mounted normally at first, aligned and scribed past the half way point of the wafer thickness, making all the cuts necessary.  After that part of the process has been done for the whole wafer, it needs to be unmounted, flipped, then mounted with the scribe cuts facing the tape.

Here comes the tricky part!  Then the cuts need to be finalized from the other side of the wafer.  Once the wafer is at this stage the blade can easily knock off the rest of the material without having to go all the way through the entire thickness, hence eliminating the horrid back side chipping.  There are several ways this can be done accurately.  The easiest way is if the material is glass.  In this case it is not difficult to focus on the bottom of the cut with the optics so that proper alignment is achieved.  In case of non see through materials there are 2 steps that can be taken.  1) While the wafer is still on phase 1 of this process, precise measurements need to be taken from the edge of the wafer to the center of the first cuts both on the X and Y directions.  When the wafer is flipped these measurements can be used to blindly find the center of the first cut and from there the rest of them.  This poses two problems, first, that the measurements need to be very accurate and second, that the machine has to be very accurate as well.  This ensures that perfect alignment is made with the initial trenches which are invisible because they are now on the back side of the wafer.  2)As an alternative, when the wafer is on phase one, the first cut of both the X and Y directions are completely made while the rest of the cuts can be made as previously expained.  The edge of these first cuts can be used as reference points when the wafer is upside down.

Dicing this way I have to admit, and worn, is not always advantageous or simple to do.  However there are many applications that could benefit from this technique, especially because if it is done correctly it is a sure way of getting only front side chipping both in front and back of the material which is always significantly less than the back side chipping.  A major issue with this process is that if improper alignment is done the cut profile will not be adequate, now in some applications this is a don’t care and getting the backside chipping diminished is more important.  Also this technique will definitely not work for any wafers that cannot be touched on the front side for it requires the wafer to be mounted top side down on the tape.

Having used both these techniques on many projects for many of our customers, whom will not be named, we have confirmed that they are sound techniques.  However it cannot be stressed enough that at least the second one is not for everyone and encourage the reader to look closely at their requirements to see if these procedures are for them.  A lot of dicing houses will not even consider dicing this way due to the amount of time and personal attention that these type of jobs require.  It is likely that a house that is focused on mostly production runs will not take their time to process a small lot of wafer this way.  If the reader has a question whether these processes will work for their application I would be happy to help asses this issue and can be contacted through my email.

I would also like to thank the readers once again for their time and feed back.  It is true that words are meaningless unless they are thought to have meaning and relevance by the reader.  Fortunately there are enormous amounts of people who are curious as to how their favorite devices come to be and the processes that make them possible, even if they are hidden.  Thank you.

This entry was posted on Thursday, January 27th, 2011 at 12:53 am and is filed under Techniques and methods. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.