One Stop Micro

Beam drift causing placement errors

Some micromachining projects require the accurate placement of features. However, Micromachining tools can drift over time. For example, if an IR laser heats up, the beam can drift, or effects like thermal lensing can occur. In a gas filled UV laser, the cavity is filled with gas mixtures under high pressure. To prevent leakage of the gas, the cavity mirrors (High Reflector and Output Coupler) are sealed to the cavity using polymer gaskets. To align these cavity mirrors, a set of screws is made available on the outside. If an unusual amount of heating occurs, it will distort the frame of the laser head or move the cavity mirrors since they sit on these relatively softer gaskets. I will go into the details of beam drift from laser cavities in another article, and only discuss errors introduced by the motion system here.

Placement errors caused by the motion system

Placement errors call for correcting the motion system to reduce errors. Although the logical thing would be to use stages that inherently have less errors- like a stage with a linear reference scale placed inside the stage- sometimes we might not have that luxury. Even if we do have accurate linear stages, further correction of the already low level of errors might be required.

Some motion system vendors provide the facility for lookup tables. Local errors can be found out by traversing around a reference piece (like a calibrated and traceable ruler or XY plate) . These errors can then be incorporated into the lookup tables. The motion system uses various mathematical functions to interpolate these discreet error values into a continuous XY area.

The question might then be asked: why go to the trouble of using external reference plates when there is already an accurate ruler inside each of these stages. After all, the ruler is already calibrated to 0.1 microns. The answer to this is:

1. The calibration ruler is one just one axis. For example: take a stage that has a perfectly rigid glass ruler inside it. The ruler has subdivision markers of 0.1 microns. All it promises is that placement precision will be around the 0.1 micron range at the temperature that it was fabricated at, along the perfect axis of the ruler. When the stage moves along its axis, say X, the carriage will also move, ever so slightly, in the Y and Z axis. When a second stage, one that travels in Y axis, is placed on top of the first one, it will move in the Y axis when the X axis is moving, even though Y has not yet been powered.
2. When you fasten two stages, there is a lot of bending, yawing, pitching and other deformations introduced. The reference scale inside the stages lose their relevance to a great degree- particularly for very heavily loaded stages and/or for those with larger travel. Of course, there is always the issue of orthogonality of the fastened stages, which, in an ideal world, can be corrected by simple techniques. The philosophy here should be to physically reduce orthogonal errors to a minimum, and correct whatever remains using software (lookup table). In reality, orthogonal errors may not be linear- they might be wavy. Mechanical correction is not possible in such cases- the software route must be taken.

Value addition through precision in placement

Micromachining projects can involve large parts (like 12 inch wafers), and on it might be features that are in the micron-size range with precision placement requirements. Service providers might have a world-class micromachining capability, but advertising placement precision, accuracy and all those fancy terms can add a lot of value to ordinary micromachining contracts and systems.

The Disadvantages

Another point to remember is the temperature dependence of calibrated systems, as also the overheads in maintaining a steady temperature. You can’t let operators make any adjustments. Assembling and disassembling in a job-shop setting is a pain. Hysteresis is a killer, so you end up having to track a lot of parameters. You cannot rest assured that by bringing back all the original parameters (after an adverse event) your system is back to its original calibration. Imagine what will happen if one your operators is not careful enough with your precision aligned stages- you have to start all over again.

Wasn’t your core-competency laser micromachining in the first place? Why get into these fancy setups if you don’t absolutely need it??

Exotic error correction products

There are some outfits that sell interferometry based stage precision tracking and correcting systems. There are a few others who sell other add-ons that can enhance accuracy of placement. Buyer beware: some of these are indeed exotic, and perform well under a very small operational parameter window.

Laser Micromachining Product Development Examples

Translating an idea on paper into an actual product is a multi-step process that includes lasers as well as auxiliary and pre/post processes. This is best illustrated by the two examples below.

Example 1: Monolithic True 3-D Lymphatic Implant

Description

• Size: 1 mm overall, with 100 micron conduits
• Parts, masks made on calibrated stages
• YAG laser to make a contact-masks
• Alignment holes for point-to-point registration
• View ‘through-laser-lens’ and stereo microscope

Process I

• 193 nm Excimer- machining of PMMA (contact mask on top)
• 355 tripled YAG for high-speed release of parts from sheet
• Ultrasonic cleaner (Phosphoric acid, surfactant, 100 F) to remove metal slag deposited on polymer.
• Megasonics to remove finer particles (outsourced)
• Problem: FDA objects to fine metal particles on implants

Process II

• Switch to ceramic mask, Polyimide parts
• 248 nm Excimer- multi-step machining of Polyimide
• Benefits: No metal slag, increased throughput of 248 laser.

Example 2: Transmission Windows on metal-on-glass wafers

• Simple Process
• 2 inch wafers (10 nm gold layer)
• Requirement: Ablate 16 micron windows of metal at fiducial marks (make optically transparent)
• Process: YAG, no coolant, tip-tilt vacuum chuck

Successful Product Development Requires:

• Intricate knowledge of laser/supporting processes
• MicroProduct development life cycle experience.
• Plan for production while developing prototypes.
• Stable, repeatable, predictable-cost MicroProcess

As can be seen from the examples, processes can be simple, or they can turn out to be complex. When unexpected changes occur, it is good to have someone guide you through the process development phase.

Consider this scenario: You are a micromachining service provider, and you have a loyal customer.

This customer has been with you for several years, but with intermittent orders, and wants to continue getting parts micromachined by you. The orders are not monetarily significant to your company. You as the provider have been doing it out of loyalty to the customer. And the customer has presumably been with you all these years out of loyalty (and at ease in his comfort zone). There were no returned parts in the history of this job because the customer had built in some redundancy. Although the initial set-up was time-consuming and complicated, running the job in production-mode got to being routine after the first few runs. Orders came in a few times a year. Each run took a few days to complete.

At what point do you re-evaluate the relationship?

1) Your costs have been going up.

2) Setup time after each equipment failure is becoming an issue.

3) People who initially ran the job are no longer around. Every time an engineer leaves, he has to train his replacement.

How much is loyalty worth?

Let us look at the unspoken (and un-analyzed) but subconscious aspects of loyalty. First of all, we need to distinguish and filter out the concepts of social activism (not buying animal tested products), social fidelity (as in a marriage) and fealty (no, not to the feudal lord, but I use the term loosely here in the context of allegiance to one’s country, friends or relatives). Once we do that, our analysis can be narrowed down to a significantly smaller set of human traits.

So what are these traits that contribute to loyalty? Pardon me when I state this in an oversimplified and cynical manner: In a free-market society, I ascribe loyalty partly to human weaknesses or drawbacks (for example: laziness, fear, job-security) and partly to non-rational thought-processes (for example: superstitions, why fix it when it ain’t broken, let’s not jinx this one). I don’t claim chivalry is dead, nor that people have become meaner.

Loyalty vs. Market driven Micromachining Process Development

Should market driven process/production be so bluntly money driven? I think not, because these human traits that I call weaknesses, drawbacks or irrational thinking are what makes us human. It is the grease that keeps societies functioning in an egalitarian manner, the glue that holds us all together- the examples we set for our children. May be we need to find a balance between self-interest, practicality, and our responsibilities. Too much grease and glue can make a real gooey mess, and too little of it will grind up things and make it all abrasive. Besides, it helps nobody to have a failed company.

My personal belief (or is it a weakness?) is that fair-dealings should precede cut-throat business practices. But don’t let your heart totally rule your mid. After all, we live in a market driven economy.

Obsolescence

Micromachining is a tough job. With the progression of time, equipment and processes get outdated to the point that it is simply easier to replace them rather than upgrade or refine. Band-aid solutions will get you only so far.

For example, motion control boards for stages used to have dedicated controllers (and of course, separate amplifiers). A few years back, these controllers became boards that were plugged into ISA slot. Then they became PCI slot cards. And now with the advent of multi-microprocessor based PC motherboards, these motion controller boards will become obsolete. Two things changed here: slot-types and board-types. Similarly, laser firing controllers (that were not integrated within motion controller boards) went from serial port to USB. You can hardly find a PC with a serial port or an ISA slot anymore.

Is it time to re-think your strategy on legacy jobs?

(with apologies to Marvin Gaye 1939-1984)

Ooh, now let’s get down the lab
Baby I’m hot just like an IR
I need some machinin’
And baby, I can’t hold it much longer
The beam is getting stronger and stronger
And when I get that feeling
I want Micro Machining
Micro Machining, oh baby
Makes me feel so fine
Helps to relieve my mind
Micro Machining baby, is good for me
Micro Machining is something that’s good for me
Whenever blue beam paths are falling
And the cavity stability is inverting
There is something I can do
I can get on the Controller and turn up the beam baby, and
Honey I know you’ll be there to relieve me
The love you give to me will free me
If you don’t know the things you’re dealing
I can tell you, darling, that it’s Micro Machining

A long time back, a senior colleague of mine introduced me to the phrase “Superstitious Pigeons”. I will describe the real thing before I relate it to laser micromachining.

There was this psychologist named Skinner at Harvard in the 1950s, and he conducted some experiments with pigeons to study how superstitions developed. He put some really hungry pigeons in a cage, with an automatic timed food dispenser. The timing was independent of what the pigeons ate or did- food would pop out every so many minutes.

The pigeons of course didn’t know the auto-timing part, them being bird-brained and all that. They associated the delivery of food with whatever action they were performing at that instant. Lets say Pigeon #1 was singing an Elvis song (this was the 1950s, mind you) and food popped out just when he started singing the song, he thought Elvis was delivering the food. So, he would sing the same Elvis song whenever he wanted food. Even if the food never appeared after numerous attempts, he wouldn’t give up. That is, Mr. Pigeon got superstitious. And this kept getting even more strongly re-enforced in the face of repeated failures. Wouldn’t you say there is a subtle difference between superstition and perseverance? Hmmm, may be a gray area separates the two.

So, what does this have to do with laser micromachining? Very often, customers send exotic materials or composites to be machined. To get good cut quality, there are numerous parameters that can be (or need to be) adjusted. The list includes: focal length of the final lens, laser energy, spot-size, machining rate, pulse-overlap, ambient gas, pulse-energy, average energy… Many of these are inter-related.

An inexperienced machinist might change multiple parameters at the same time, and after numerous attempts and a multitude of parameter-sets, would start to see what he believes to be a pattern of good machining. But he never is able to refine it and narrow it down. Of course, he never gives up too. He has a coffee break every hour, comes back fresh, believing intently that a particular parameter is the answer. By evening, he has tried so many processing parameters, and starts believing intently that he can solve this thing today. His faith in his ability to solve this problem has gotten stronger. He becomes more persevering, and has a book full of test results.

His manager, meanwhile, is distraught. Not at the micromachinist’s plight, but because of repeated promises that the solution is at hand. An experienced manager will, by now, realize that machinist has lost it, has become superstitious. Time to bring in the big honcho.

The senior guy has gone through the drill numerous times over the last many years. He calmly selects 3 parameters to change, changes them one at a time, and arrives at the solution in an hour. There you have it- the difference between a rational person and a superstitious person. Or pigeons.

Experience matters.

In a Scanning Electron Microscope, the depth of field (DOF) is inversely proportional to the both the aperture size and magnification. Although this is in some ways similar to an Optical Microscope, the DOF is much higher for an SEM for the same magnification. For a 100 micron objective lens with a 10x eyepiece (total mag of 1000x), the DOF is around 1 micron. For an SEM at under 1000x mag, the DOF is around 40 microns (with a 100 micron aperture, 10 mm Working Distance).

Lets say you have a part machined by Acme Laser, and they send you SEMs of the part. Testing reveals that something is just plain wrong with the part, but the SEMs look great.  When you check with an optical microscope, things start looking bad- the part has been made unevenly. The reason the SEM led you astray is that the part looks focused over a greater depth and hence conveys a mental image of good micromachining. When you put it under an Optical Microscope with a similar magnification, the depth of field is much smaller, and you will have to keep moving the focus over a greater distance to look at all the features. In essence, unless you are a computer and can store the image details at every 1 micron depth of focus, the part looks uneven- which conveys a mental picture of poor micromachining.

So which should you use? An experienced micromachinist can make do with an Optical Microscope, but for showing the same part to potential investors or to spice up publications, you are much better off using an SEM, even though the SEM provides black and white pictures (while the OM provides color).

There is a caveat, though: Often, a well micromachined part may not be what is really needed. You might actually end up needing a softer focused part to avoid sharp edges and for better fit- as in biomedical applications. Read about it in another article that will appear soon.

Another thing to remember: An SEM gives much higher resolution than an Optical Microscope primarily because the wavelength of an electron beam is much smaller- less than a nanometer compared to about 500 nm in the visible spectrum. But then, a laser can not micromachine features smaller than around 2 microns (usual case scenario for a Flourine laser at 157 nm). So what’s the point using a metrology/visualization tool with a much higher resolution- unless you are loking at post-fab debris and compostional changes. That’s another topic altogether.

Imagine a silicon wafer with some features already micromachined on it. You now want to make some vias to provide connectivity from the backside of the wafer. How would you do that? One option would be to laser micromachine these vias, but then you are worried about what kind of damage the laser would create to the structures already on the wafer.

May be you have this impression of a red-hot laser burning through the wafer in one quick second- been watching too many Bond movies lately?

The fact is, the heat affected zone is very small- provided you use the right kind of laser. An ultrafast laser will have the smallest HAZ, followed by the UV and then the IR. A CO2 laser will surely be unacceptable. So what kind of numbers are we talking about? For an ultrafast, the HAZ typically would be sub-micron, for the UV lasers with pulse-widths in the low nanosecond range, the HAZ will be a few microns at the most, and much more for IR lasers.

If the features are 10s of microns away you are safe from a thermal point of view- but not from debris. Debris is going to be scattered aruound much further, and cleaning it completely without damaging the sorrunding structures is going to be very difficult or impossible.

If you ask a service provider if there is going to be heat damage to structures that are 100 micro away from the vias, the answer is going to be a no. But remember, debris and cleaning issues will have to be addressed

If you are new to Laser Micromachining, how would you know what your project might end up costing? Lets say you approach AcmeLaser Inc (of course, a fictitious company), would they quote on your project? And how much? Well, it all depends on what kind of a laser micromachining project you have.

There are the ordinary, run-of-the-mill laser micromachining projects. And then there the intermediate-difficulty ones, where fixturing, setup and micromachining might be complicated, but not too exotic- something similar has been done before by the AcmeLaser. Slight variations in operational or processing parameters will not kill the results.

And then there are the special ones, where every little detail has be taken care of. I am not talking about finicky or temperamental jobs. These are jobs that require an expert in the field to set up, and a good team to run it.

If your project meets 8 of the 10 criterion mentioned below, then your project might probably fit into the exotic or specialized category. Some of the characteristics of these difficult projects are:

1. Technologically Challenging: One or more of the following may apply. A few examples are given.

• May require the use of more than one type of laser. For example: A 2-step process.
• A laser with very specialized beam characteristics. For example: Perfectly circular and homogeneous energy beam.
• A laser with near-constant power per pulse, even after several hours of continuous operation.
• A motion system that is precise and accurate to a very high degree. For example: If you want +-0.5 microns over 200 mm of travel, you have to pay for it.
• Beam delivery components that are complicated to design or integrate. For example: You want the beam waist to be the same size over 2mm (nearly parallel).
• Requirement of very high energy density on work surface, but limited by wavelength (generally, lower the wavelength, the lower the average as well as peak power and higher the cost, maintenance, beam delivery etc). For example: extra pure glass, or diamond.
• Unusual fixturing or parts handling requirements. For example: Direct-load from cassettes, with double side registration and machining.
• Unusually contoured parts. For example: Some medical devices fall into this category. Need to map a 2.5-D surface and then machine.
• You want a machine 10 widgets a second- Super-fast leads to super-cost for setup, ROI after millions are made.
• You can’t afford even one single mistake on any part. Technology has to be developed to make the process perfect.
• Extremely tight tolerances. For example: specifying a 200 micro hole to be +-1 micron on a 200 micron thick steel shim.

1. Requires Precise Inputs. For example: Liquid coolant or gas flow on machined parts to be very precise? 10 micron Polymer sheets to be within 0.25 thickness, else holes go out of spec.
2. Environment-Sensitive. For example: Even a 1 degree Centigrade change in room temperature throws off your beam or motion system, in turn affecting placement accuracy.
3. A different generation of metrology equipment might be required. For example: Exit and entrance quality specified on holes that are on a ledge, requiring Extra Long Working Distance microscope lenses. Or, you need to look for debris spread-area using dark-field lighting. Maybe an SEM is required to visualize the finished piece.
4. Monitoring of inputs and results, extensive record keeping, statistical analysis. This puts a heavy burden on the provider. It may be very important to you as a customer, but not for the provider unless they are paid to do this QC, process management and book-keeping.
5. Involvement of mature process managers with strong fundamentals in technical skill sets. AcmeLaser needs a manager with a lot of experience. This person is usually very busy with other things in the company, so unless you are willing to pay a lot, you can’t have a lot of the expert’s time.
6. Reliable and mature technicians and operators. AcmeLaser has to allocate the best of its staff to your project because it is too difficult and sensitive.
7. Excellent overall management of project. AcmeLaser doesn’t want its managers to micromanage- just get it running and move on to the next customer.
8. Willingness, appetite and ability to bear risk involved in spending resources on an untested process. You mean setting up this job is going to cost 10% of AcmeLaser’s annual budget?

The truth of the matter is that no service provider has every type of laser or motion system in-house. What for one is a specialized job might not be the same for another. When you send out a RFQ to several vendors and most of them respond, then you can be sure it is basically a simple project or it is a large enough project for providers to spend their resources to recoup their costs.

Executing Micromachining projects that are not run-of-the-mill type is difficult. Many companies turn down some of these projects, the rationale being that it is technically unfeasible or too expensive. Being unprofitable and few and far in between, they rightfully choose not to make it part of their core activities.

The knowledge base needed to execute many of these projects exists in disparate fields in the public domain, which has to be collated and used in an intelligent manner. If only the volumes justified the investment, more companies would be willing to give it a try.

Some Specialized Micromachining Projects

• True 3-D, true monolithic microparts without micro-joining
• Ultra small holes (micron and sub-micron) at high speeds.
• Motion calibration for precision Micromachining over large areas, done to traceable standards. Price becomes exponentially higher as you approach a +-1 micron error.