Eric has been wanting to post a thread, with photos, about the importance of proper HT, especially normalization. I agreed to help with the photos, but please keep in mind I am a beginning knifemaker, and this is really Eric's tutorial - I'm just involved due to Eric's generosity and because I could get some photos that show what we're looking at. He'll come in and add text to the photos as he sees fit, and/or repost and discuss the photos later in the thread.
[Edit by Eric (knifesmith): Ken is being a bit too modest. While I did the basic HT, we couldn't have this thread without his spectacular photography, so it's definitely a team effort.]
This experiment helps illustrate something that I learned a couple years ago at an ABS hammer-in at Visalia (about this time of year - there's another one coming up pretty soon now). Even though I'm a neophyte, I'm going to state it here, because it was something I had never been told or had read before.
You can have two identically formulated and shaped pieces of steel, both having the exact same Rockwell hardness, and they will have very different toughness, strength, and edge holding ability, depending on the size of the grain in the steel. Proper heat treat controls grain size and hardness! [Edit by Eric: Heat treat pretty much controls everything. Hardness, grain size, carbide distribution, toughness/strength, as well as edge holding/wear resistance]. A knife identical in every respect but with larger grain will break more readily, and not have the edge holding ability of its fine grained counter part.
Eric decided to use an old file (I believe it was a Nicholson) for the experiment. The short version of the experiment is that he broke the file into reasonably equal size shards after each step of normalization cycles in a normal garage style quench, including an as from factory sample and a sample with purposely bad heat control.
Each sample or shard represents the cumulative effects of all preceding heat treating and quenching at that point. First, using a vice and hammer, Eric broke off a sample of the file as it came from the factory, then sealed it in paper and labeled the package (all samples were sealed immediately upon detachment from the file to make sure they didn't get mixed up). That sample shows the grain of the file as created soley by the factory's heat treat, or what would be considered optimum (labeled "Factory HT").
To show the effect of poor heat control on grain, he then heated the file far beyond critical, and quenched it in oil. I've labled that sample "Bad Heat Control". [Edit by Eric: This is really the sample that shows what happens with too much time at a high grain-growing temperature. Unless you're doing proper normalizations, this is likely the sort of grain you have, or at least something close to it. This sample will be hard, but extremely brittle and not tough/strong at all. It will also have significantly reduced edge holding capacity compared to its counterparts with the same hardness and smaller grain. The individual crystals in the steel are huge and easily visualized with the unassisted eye.]
Next, he brought the remainder of the file up to critical, judging by the decalescance on its way up (clickhere to check out the Decalescance thread in the Knifemaking Reference forum). The file was then allowed to air cool past black, then put back in the forge and brought to the correct temperature, judging by decalescance, and quenched in oil. Another sample was taken, which was the 1xNormalization sample. This process was repeated until we had a 2xNormalization sample, and a 3xNormalization sample. At the end of the heat treat portion of the experiment, each of the four samples other than the "Factory" sample had been quenched, and all three of the samples other than the "Bad Heat Control" had been normalized (either once, twice, or three times) prior to quenching. [Edit by Eric: As this experiment was meant to reproduce simple "by eye" home workshop heat treating, we did not soak this steel at critical or do anything other than the most basic heat treat possible. As noted in other threads, hypereutectoid steels like file steels generally benefit from soaks at temp. etc...]
Finally, I photographed the samples so we could see the relative grain sizes. Sorry for the quick and dirty photography, it's all I had time for. I'll briefly go over how I took the photos later in the post in case anybody's interested.
Following are the photos of the samples, preceded by some shop shots taken during the experiment.
The file in the forge, coming up to way over critical for the Bad Heat Control Sample. The piece of metal sharing the forge (on right) is a wrought iron guard I was working on while Eric was conducting the experiment.
[Edit by Eric: The file was left in the forge at this bright orange heat for ~10 min to represent the worst possible case of grain/crystal overgrowth. This is worst possible condition for the steel (excepting burnt and sparking).]
Eric judging temperature on the file:
[Edit by Eric: Actually this is me doing what I normally do while I watch something normalizing. I'm open to suggestions on alternate behaviors other than scratching my head in boredom. And a good note would be that I normally HT in the dark so that ambient light conditions do not interfere with seeing colors, judging when something is no longer radiating light, decalesence etc... It didn't fit our time frame or photography needs to try doing this at night or in a closed garage.]
[Edit by Eric: Yeah, don't be stupid like me. Wear your safety goggles. Hot oil is not good for your eyeballs.]
Who says science isn't fun?
After the first quench for the un-normalized "Bad Heat Control" sample, the file warped:
Eric used a dremel with a cutoff wheel to score the file, in an attempt to get the shards to break off evenly (which was semi-successful - less so after the file had been stress cracked, apparently). Getting the file to break off evenly was important for the photographs, since with magnification the depth of focus would be greatly reduced, and a jagged edge would not be confined within the narrow plane of focus.
The shards laid out after all the photographs. I did this to make absolutely sure I hadn't mixed up the samples (though Eric and I had been very careful not to). It looks like after the first quench, the file would rather break along the stress risers of the file notches than the score marks.
[Edit by Eric: No giant surprise there. In fact, you can see dark areas in the later normalized samples that are where oil from the quenches got into a stress crack induced by the notches of the file. This is why not having stress risers heading into your quench is so important. Had we not been shattering this file, those partial thickness stress riser cracks would have gone unnoticed!]
Here's the first photo of the samples, all together. This is uncropped and resized from 4368 pixels wide to the approved and PaleoPlanet friendly maximum of 750 pixels wide. The cropped in version and higher magnification shots following are labeled.
Cropped in tighter. I'm sorry the order is incorrect, but I didn't have time to reshoot. Correct order as done in the experiment would simply place the factory sample on the very top of the stack.
[Edit by Ken: I think it is important to realize or make clear that every sample except the Factory shard went through the stage of being exactly like the top sample - Bad Heat Control. Subsequent normalization cycles are what took the grain closer to that of the Factory HT sample.]
[Edit by Eric: The above photo makes the stress riser induced cracks very obvious. You may even note that the crystal/grain size in the cracked (dark) area of each sample corresponds to a crystal size in one of the previous samples. The crystal size in the stress crack of the 1x and 2x normalized samples seems to correspond to the crystal size in the non-normalized sample. The crystal size in the stress fracture of the 3x sample looks to correspond to the non-normalized, or perhaps the 1x normalized sample.]
The following shots were taken through a reversed lens to increase magnification. This increase in subject size is a trade off since depth of focus decreases, and lens distortions and aberrations are introduced. The second image in included since the whole stack couldn't be completely shown in one shot, and I didn't have time to combine it into one image.
These following are in images of the individual samples in order of how they were produced in the experiment:
[Edit by Eric: Note the super fine silky grain. You can't see the individual crystals at all, even at this magnification.]
Bad Heat Control
[Edit by Eric: Note the giant crystals, easily seen. This looks rough and sparkly even with no magnification.]
[Edit by Eric: Starting to look good. The crystal size is beginning to get small, but it still looks a bit "rough". At no magnification it is still difficult to identify individual crystals. The bad grain at the lower edge of the sample is an artifact of having the surface texture "locked" in place by a stress fracture occurring at the previous quench.]
2xNormalization (The lighter colored triangular or flame shaped portions are tiny ripples where the file broke unevenly, and the appearance of finer grain within them should be ignored - it is a focus effect. Grain structure is similar in adjacent same-colored areas):
[Edit by Eric: This is starting to get really silky/small grain/crystals. At no magnification, this looks really velvety, and it is impossible to see individual grains/crystals. For all intensive purposes, this is an acceptable grain structure for a knife. Again the darker band at the bottom of the sample is bad grain locked in by a stress fracture at a previous quench.]
[Edit by Eric: Not really too different looking from the 2x sample eh? Again, darker area, stress fracture, blah blah.]
And then the same series again, cropped in:
Bad Heat Control
The photos of the samples above were taken with a Canon 5D, using an EF 24-70mm F2.8 L lens. The reversed lens for the latter shots is an old Pentax normal lens I got very cheaply (it is an old screw mount rather than bayonet mount), but just about any lens would work for this. A cheap lens is a better choice if you are going to clamp it like I did; and because it is necessary to get very close to the subject, where there is a risk of scratching the delicate rear element of the lens. Buy the way, an old normal lens held reversed (front element to your eye) makes an excellent loupe (much cheaper than buying a good loupe, but made to similar standards) for examining fine detail (like looking for scratches in your blade, etc.)
The snapshots above and following were taken with an old (approx 4mp) Sony Cybershot. The first one is a test using the macro mode, hand holding the camera in sunlight to see if the grain showed up on the Bad Heat Control sample.
Here's some snapshots of the set up for close up photos of the samples through the reversed lens. The stands are called "C-stands", and are very common and useful studio equipment that can hold lights or other objects. One is holding a telescoping magnet, which allowed for the file shard to be easily moved in incremental adjustments (very helpful working at magnification), and the other is holding the supplementary lens in place, utilizing an improvised clamp made from metal strap (covered with duct tape to help protect the lens) and a C-clamp.
Thanks for including me in this experiment, Eric. I hope the photos help. It'd be nice if I could pay forward even a little of all I've learned from hanging out with you...
[Edit by Eric: Again, Ken with the modesty. This couldn't have been done without Ken's photography skill and efforts. This was definitely a team effort.
I'd like to repeat that this series by no means represents an ideal heat treat for what this steel likely is. It was meant only to show the relative sizes of grain with no normalization, and with successive normalizations in the most basic of HT's. In fact, the more metallurgically savvy of the folks here can probably look at the closeup photos and note some issues with grain size and carbide distribution. Also, I'd like to repeat that the fact that we had serious issues/problems with stress fractures is likely due primarily due to the abuse that the steel had being overheated, not normalized, then quenched in parks 50, a fairly fast oil. That being said, you can get unseen stress fractures (even if they're not this bad) under file teeth, the "points" of rasp gouges, and even under significant spots of firescale. Some of these are going to be unavoidable if you do more "rustic" work, but this hopefully will show the logic of minimizing potential stress risers as much as possible. After the first quench, this file was riddled with partial thickness stress fractures that weren't obvious until we re-ht'd and re-fractured the file.]