The Science Of Strengthening Glass

Strengthen Glass

[Ben Krasnow] is at it again. This time he’s explaining a simple method for strengthening glass. As usual, he does a fantastic job of first demonstrating and explaining the problem and then following it up with a solution.

[Ben] first uses a simple rig to place a controlled amount of force against a glass microscope slide. His experiment shows that the slide shatters once about 30psi of force has been applied to the center of the slide.

[Ben] then goes on to explain that current methods for producing glass leave many tiny impurities, or cracks, in the glass. As the glass slide flexes, the inside edge is placed into a compression force while the outside edge is under tension. The glass is more easily able to handle the compression force. The tension is where things start to break down. The tension force eventually causes those tiny impurities to spread, resulting in the shattering glass.

One possible solution to this problem is to find a way to fill in those tiny impurities. According to [Ben], most glass has sodium added to it in order to lower the melting temperature. [Ben] explains that if you could replace some of these smaller sodium atoms with larger atoms, you could essentially “fill” many of the tiny impurities in the glass.

[Ben] does this himself by heating up a small vat of potassium nitrate. Once the powder becomes molten, he submerges the glass slides in the solution for several hours. During this time, some of the sodium atoms are replaced by potassium atoms due to the natural process of diffusion.

Once the slides have cooled down, [Ben] demonstrates that they become much stronger. When placed in the testing rig, the stronger slides do not break until the pressure gets between 60psi and 70psi. That’s twice as strong as the original glass. All that extra strength from such a simple process. Be sure to watch the full video below.

24 thoughts on “The Science Of Strengthening Glass

  1. How do we know the slide is stronger because of the potassium atoms and not because it has been tempered.
    Hardened glass is made by heating up a piece of normal glass to relieve any internal stress points.

    This is exactly what he has done with the ‘potassium bath’, how about he runs a ‘control’ where he just heats up the glass slide WITHOUT the potassium bath.
    Then he tests THAT glass to see if there is any difference.

    1. Becaue potassium atoms are larger than sodium atoms, which is what induces the compression. The process isn’t exactly new, and as Ben notes the patent has long expired. You can look up Corning’s original patent online (US3410673 A)

    2. The glass transition temperature for soda lime glass (which is what Ben is using) is around 580 to 600°C. This is the temperature above which the atoms in the glass can actually _move_. Tempering glass involves heating it to well _over_ the melting point (according to Wikipedia, 720°C) and then rapidly cooling the glass with air.

      Ben is only heating to 300°C, so there’s no chance of the glass tempering.

    3. Because Ben isn’t *testing* a new method, he’s demonstrating an existing method.

      If it was a sales pitch then questions like that absolutely need to be answered, but in this case (as other commenters have shown) it’s a well-understood process.

  2. While I agree that a control should be done, glass is tempered at 600C – 620C and cooled rapidly with blasts of air from different directions. His oven was only up around 425C and cooled slowly, so not adequate for tempering.

    1. The quality of the writing on the articles that Hackaday is pushing has really been dropping off lately since the corporate push to have more content. I hope you guys are making more money, because this whole quantity being more important than quality thing is not so sexy….

      1. Agreed. Why rewrite the summary of a science experiment and get all the science wrong? Either be lazy and just reproduce the linked content verbatim, or do it properly and have someone who actually understands the content (or at least has enough reading comprehension skills to properly summarise it) write about it. Basic scientific errors were what drove me away from slashdot and HAD is moving in the same direction.

      2. It’d be dropping since well before that. For as long as I’ve been going here I’ve always seen people bitching about copy quality, proofreading, or editing. The problem is attributed to a different cause as time goes on. I get the impression that the real variable is likely the readers themselves. People come here when they first get interested in such things, and as time goes on and they become more knowledgeable, the apparent quality of the research, fact checking, and so on seems to decline simply because they’re able to spot the shortcomings that passed unnoticed before.

  3. I was cutting some mirrors a couple of weeks ago and, yea, glass cracks to easily. Anyway I wonder if you could use this technique on a cell phone screen to make your own “monkey glass”?

  4. A nifty trick you can do with chemical tempering is a two layer temper. (I read about this in an old Machine Design magazine) First add a deep compression layer, then untemper a thin surface layer. The surface layer then shows viable cracks well before the part fully fails so that the part can be replaced before it fails catastrophically.

  5. It is hard to take technical writing seriously when basic dimensional analysis is not done.
    -” … 30psi of force has been applied to the center of the slide.”

    PSI can denote pressure or stress – source:
    Force is a vector with the SI unit of newtons – source:

    Quality of writing is important if you want to captivate the attention of critical thinkers.

  6. The slide broke past the end of the scale markings. You can’t extrapolate past the end of the scale since that is the only calibrated section. That region is usually nonlinear. So,my point is that the pressure could actually be higher than 70.

    1. The “pounds per square inch” is bothersome and doesn’t tell me much. That would mean the force on the tiny area where the cylinder contacts the slide, which which would be a very small force producing high PSI. Better for this kind of thing is one of the lab testing metrics, like mm of displacement per a given span and thickness and the force needed to produce that (pounds hanging from or pushing down on the glass).

      No one mentioned annealing. What is the heat bath doing in this regard?

      1. Doesn’t matter what the units are – this is a relative comparison. Also, this testing rig is more accurately simulating what a piece of glass would experience in real life – being bent or impacted. A normal tensile/compression stress tester, like you describe (a clamped sample being pulled or pushed) probably wouldn’t give you much useful info, since the reason this glass is “stronger” is due to a surface effect, not a bulk property of the material.

        The “heat bath” is the bath of molten potassium nitrate which the slides are immersed in. The temperatures are too low for anything like annealing – besides, the glass will already be annealed after production.

        1. I’m not suggesting a different apparatus, just reading different parameters, like mm of displacement versus the span of the supporting dowels for a given thickness of slide. This also allows rough estimates of stress and strain because you can calculate a radius of curvature and a difference in length between the top and bottom of the slide (if you know the thickness – Look what I found Mr. Wizard! A tensor!).

          We always annealed lab glassware after blowing/joining/forming, adding electrodes/quartz windows, etc. I don’t recall the temperature of the annealing ovens. I thought they were no that high. I see that borossilicate anneals at 535C and KNO3 decomposes at 400C so definitely not enough for annealing.

    2. Yes, the slide broke past the end of the scale markings, but it looked like it broke just before the needle hit the end stop pin, so the needle hadn’t pegged.

      Looking at Ashcroft’s range of gauges (the one he’s using), the _worst_ accuracy that they quote for a gauge is “3-2-3” percent. That means +/- 3% in the bottom 25% of the range, +/-2% in the middle 50%, and +/-3% in the top 25%. This is the usual way it’s quoted. So at most, I’d guess you’d only be a few PSI out even extrapolating a little past the end of the scale. Certainly less than the error in guessing what reading it reached.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.