Wednesday, April 18, 2018

LLNL Lab Successfully 3D Prints Optical Glass - With Some Tricks

I remarked a few months ago that we don't go a week without a story about something new in 3D printing in the trade magazines.  While I've become a bit numb to those, perhaps because of my interest in optics, this one made me say "huh?"

A group at Lawrence Livermore National Laboratory reports that they've successfully fabricated optical grade glass with a new printing technique.  Optical grade glass is tricky.  It's hard to convey just how clear and distortion-free optical grade glass is compared to other glasses you've seen in your life.  Eyeglasses, which are virtually always polycarbonate or a softer plastic, are nowhere near as  transparent as optical glass is.  The LLNL group isn't using a printer to produce a familiar eyeglass lens; the breakthrough here is the ability to print special mixes of optical glass with a different refractive index in each layer, which may allow more exotic shapes and performance.
Because the refractive index of glass is sensitive to its thermal history, it can be difficult to ensure that glass printed from the molten phase will result in the desired optical performance, researchers said. Depositing the LLNL-developed material in paste form and then heating the entire print to form the glass allows for a uniform refractive index, eliminating optical distortion that would degrade the optic's function.

“Components printed from molten glass often show texture from the 3D printing process, and even if you were to polish the surface, you would still see evidence of the printing process within the bulk material,” says LLNL chemical engineer Rebecca Dylla-Spears, the project’s principal investigator. “Using paste lets us obtain the uniform index needed for optics. Now we can take these components and do something interesting.”
Their goal is to improve the ability to manufacture difficult things, such as gradient index (GRIN) lenses.  The promise of the technique is to manufacture optical glass in novel shapes, reducing component count in some systems, and probably allowing new types of optical systems as well.  
For the study, researchers printed small, simple-shaped optics as proof of concept, but Dylla-Spears said the technique eventually could be applied to any device that uses glass optics and could result in optics made with geometric structures and with compositional changes that were previously unattainable by conventional manufacturing methods. For example, gradient refractive index lenses could be polished flat, replacing more expensive polishing techniques used for traditional curved lenses.

“Additive manufacturing gives us a new degree of freedom to combine optical materials in ways we could not do before,” Dylla-Spears said. “It opens up a new design space that hasn’t existed in the past, allowing for design of both the optic shape and the optical properties within the material.”

(Pictured: LLNL chemical engineer and project lead Rebecca Dylla-Spears and LLNL materials engineer Du Nguyen.)

As the article said, the lenses that they show in the picture are "small, simple-shaped" optics, and I'm not clear on how it's processed.  It sounds as if their paste will have to heated to the melting point of the glass, which means it will have to be held in a mold so that it doesn't flow away while it's liquid.  The treatment in the molten state is one of the things the distinguishes optical glass from regular slabs of glass.  Holding it at some temperature between molten and solid, annealing the glass until it's stress free and shows no swirl-like irregularities when looked through onto a flatly lit surface (or sky).  All those steps are still needed and still there.  I'm guessing the main interest here is the novel structures with different refractive indices in different stack-ups, or perhaps in rings or other shapes.


  1. I'm guessing if you can "print" the lens, you can also print a mold. That might defeat the whole purpose of "printing" the lens.

    And yeah, most people are astounded when they look through optical glass the first time.

  2. they look so young

  3. Worked on building a chip making machine back in the 80's that had a 1-to-1 lens system. That lens required glass that was as perfect as they could make, or at least as we could measure on a Zygo interferometer. 1/20th wave (red HeNe) surface figure. The Melt took about 9 months, and a total of a year to get ground, polished and coated doublets and prisms into stock to build the lens assembly. We could occasionally qualify a lens to map 1 micron geometry, as opposed to the normal 1.25 micron lens ($100k option!, on a $750k machine).

    You spend a lot of effort cleaning before cementing optics like that!

    One of the stories that I learned then was that the optical industry in Germany in WW2 had lost access to good optical cement (origin was Canada). They resorted to ringing their optics together. Periscopes and binoculars were mentioned. I've seen glass break rather than separate after ringing together while cleaning the mating surfaces. One lens was mounted this way on the lightpipe in the Illuminator Assembly.

    1. "Ringing" them as you can do with gauge blocks?

      I've "rung" gauge blocks together, and sometimes they're very difficult to get apart!

    2. One of the stories that I learned then was that the optical industry in Germany in WW2 had lost access to good optical cement (origin was Canada).
      Yeah, it was the sap of some specific tree, like turpentine. They called it "Canada balsam". Another one of those "how the heck did they find that?" things.

      I've never seen glass wrung like that, but it's pretty obvious it should work. If both curves are the same radius and match to within a tenth wave (or less) tolerance, they'd have to.

      And for drjim, I've wrung ferrite cores together for critical transformers (and I doubt those were flat to even 1/4 wave.

    3. Now that you mention it, SiG, I've had ferrite cores stick together on me. I used to think they stick together for some other reason, but seeing as these cores were lapped together, I guess they had a pretty "air free" interface.

      Royal PITA to pop them apart without chipping the ferrite....

  4. Yes, just like with gauge blocks. Works best with flat surfaces. Difficult to match curvatures as well as a flat can.

    With glass, you can clean them, and let them touch, and watch the air evacuate from one side to the other. What you see is the fringes of the wavelength of the light you are using. When you are cleaning matching surfaces, that is how you check for dirt specs or soft residues. You really want to stop the process as soon as you determine it is clean, because sometimes they won't come apart. The developing bond sometimes appears to be atomic, and one or both parts may break while trying to separate them with physical force. I've seen a fracture that crosses the mating surface on both pieces, and it rarely separates there. Really bizarre.

    We also used diamond dust to lap the reticle mount and the wafer chuck (both hard anodized alum) to get them very flat. We used an optical flat to read the fringes to measure the flatness over the entire surface of the part. The lapping tool was a mehenite cast iron block. A little bigger that a 123 block, IIRC.
    BTW, it appears that Privacy Badger was the culprit for the non-posting problem. Wasn't playing well with Blogger.

    1. Good to know about the Privacy Badger trick. I'll try to remember that. I was thinking about trying what Bayou Renaissance Man did with turning off Captcha. The big disadvantage to that is no more anonymous comments.

      Yes, just like with gauge blocks. Works best with flat surfaces. Difficult to match curvatures as well as a flat can.

      I was thinking about this yesterday and got confused thinking about the difference between measuring flats and measuring radius of curvature. Is it possible that the fringes still look like the lenses are matching but the ROC is different?

      The effect of it looking like the glass pieces bonded in some sort of self-weld is a bit stunning. First I've ever heard of that. I know some pure metals can do that, but doing it in a vacuum helps. Never heard of compounds or mixtures of compounds self-welding.

    2. Offhand, I can't recall the types of glasses that were used for the various elements. I suspect the ones that tried to join were the same material. The pair that we had the most problems with was the lightpipe (kaleidoscope) and the end lens. I know the pipe was not coated, but I can't remember if the flat on the lens was, or not. You would look through the lens, which would magnify the joint, and watch as the air would get pushed out. You were watching the fringes, and as they got wider, the final one would cover the entire surface and flash, and the joint would essentially disappear, and you would be looking into the pipe. We had an identical element on each end, and one was joined this way, and one was cemented. IIRC, you seldom found both ends good enough to give such a perfect bond. That may be why one got a drop of casting glue? Stuff was used for taking impressions, and was crystal clear. Two part, that was somewhat soft. De-air it in a centrifuge, and use a drop per assembly. You could remove it from the plastic spin tube, and it would have a perfect impression of the volume markings.
      The wrung together joint was the one that was closest to the lamp. The Illuminator was designed to provide as uniform a light supply to the wafer as possible. The lightpipe was used to multiply the light source over the window to avoid hot or dark areas.

      I don't know the details, but I suspect the German optics were flat joints they were ringing together. I'm pretty sure that a wrung joint doesn't require coatings for best function, like a glue joint would.
      I wonder if the Japanese had the same problem of lack of cement?

      If the ROC isn't nearly identical, you won't get fringes beyond the contact point. The potential problem is that, at least with doublets, the edges of the socket get thin, and the glass will bend while trying to conform to the matching part. This will stress the glass, and even if you aren't using the full surface, may still cause image problems. The dreaded Maltese Cross may show up when checking bi-refringence of the doublet after cementing.

  5. "I've never seen glass wrung like that, but it's pretty obvious it should work. If both curves are the same radius and match to within a tenth wave (or less) tolerance, they'd have to."

    The drawback is that it's never perfect, but they will distort themselves attempting to make a perfect match. You could see it happening when you fit the parts together, dry. This is why our doublets were a glue bond, to avoid that stress. Still, when gluing, if you let the two surfaces get too close, this would still happen. De-bonding was a potentially hazardous situation to the health of the glass, and the techs.

  6. That stuff is way beyond cool, Will!