Regarding Starch:
Now that I'm mentally organizing everything, I've gotten a lot done in the last two weeks! The starch I use now is dyed according to the original autochrome recipes. The colors now look very reminiscent to pictures of the original screen (especially the violet grains!).
The new starch fails to achieve a neutral tone on the screen, leaning toward yellow instead -- I think this is due to the green grains being rather weak. I probably should have tried to balance this back out with more violet, but I had already went ahead and made a bunch of screens with it and it was too late. The T-Shirt dyed starch would probably have worked fairly well, but right now there are enough variables to deal with. It's something I'll probably revisit in a year or two.
Regarding the second varnish: I'm able to make completed, water-tight screens regularly now. I tried experimenting with nitrocellulose varnish, which for a time seemed to make a better seal. Unfortunately, things did not pan out - the photographic emulsion did not stick to the nitrocellulose varnish at all, and simply flakes off when dry. The second varnish ("my" version) performed well in this department, and I'll be sticking to that for the time being. Lately, as the plates dry, the varnish leaves small "wrinkles" which block light and seem to pull some of the carbon black out of the interstitial spaces between the starch grains. This isn't that much of an issue going forward, as the wrinkles are somewhat small. My technique for sealing the screens relies on setting the plate on a PERFECTLY flat surface. I use a bubble level attached to a ball-head tripod, and meticulously check it to ensure it's flat. Before I pour the varnish, I take a blank glass plate and pour some water on from a shot glass, ensuring it spreads evenly across and doesn't bias in a particular direction. Using the 2nd Varnish recipe I outlined in my previous post, I pour one teaspoon of varnish on the plate and gently tip the glass, allowing the liquid to spread evenly across the entire surface. Occasionally I use my finger to help it across the screen, but be careful - the presence of the 2nd varnish softens the first varnish considerably, and the starch is fairly loose. When finished spreading, the varnish is allowed to sit on the plate as a meniscus and dry. The ethyl acetate will evaporate quickly, leaving behind the amyl acetate, which takes a considerably long time to dry. If you make a mistake and one part of the plate is a bit thicker, it could take days for it to dry completely. After a day or so, test the plate out by submerging it in a tray of water for at least 1 hour. If it fails, you should clearly see areas where the dye bled.
I'm currently working on a formulation of the second varnish using no 'shortcuts'. Currently, I can't get all the nitrocellulose to dissolve completely, which has more or less turned the solution into jello. According to the original notes, it takes upwards of a week to dissolve fully, so we'll have to check back on that in a few days...
Regarding the emulsion
I now omit ammonia from the recipe. This results in considerably slower plates, but it also has eliminated my issues with fog. In the future, I'll probably experiment with faster emulsions, but this speed is fine for now. With this formulation, I get ~1 ISO with no screen. With the screen, maybe closer to 0.25 ISO. Solution A:
25mL H2O
2.4g KBr
30mg KI
150mg gelatin
Solution B:
15mL H2O
3g AgNO3
Solution C:
15mL H2O
3g gelatin
Heat A to 45C, then under a redlight (and with magnetic stirring), slowly add B to A. Allow to sit for one under with continued heat and stirring. Add C with a small spatula. Allow to cool and solidify, then perform a standard noodle wash. Re-melt, and heat for one hour. Before pouring the plates, add: 1mL 1:1000 alcoholic solution of erythrosine
1mL 1:1000 alcoholic solution of pinacyanol
2mL 1% chrome alum solution
2 drops of photo-flo
Coat plates with 20mL of emulsion per 4x5.
First Results
Only a single, slightly underexposed plate made it from the first batch of 7 (the other 6 had used the nitrocellulose coating, so the emulsion did not survive). The plate did exhibit the slightest bit of color response, showing that the second varnish layer was thin enough. The plates from this set still had issues with leaking (deep purple areas) so processing on this was rushed. Reversal processing is out of the question until the leaking is fixed.
The second batch shows a much cleaner color response. It's hard to tell if the plate actually has a blue cast, or if the digital inversion is actually being thrown off by the yellow screen color. I neglected to filter the emulsion before coating, which is where the black dots likely came from.
Attempts at reversal processing
I won't say I've officially made an autochrome until I can manage to reversal process one of these plates. All additional tests with this batch have been attempting to reversal process a plate, with little to no success. Reversal processing these plates ends in a completely opaque coating covering the entire plate. I believe it is due to excessive emulsion thickness -- as the rays of light pass through the screen and into the emulsion, they are not able to penetrate far enough to convert silver halides towards the "back" of the emulsion. When it comes time to reversal process, this leaves a blanket of silver covering up the reversed image. See my highly technical drawing below. Reversal processing was performed with a dichromate based bleach. The dichromate bleach, though toxic, is much easier to work with than the original permanganate bleach the Lumiere brothers used. The permanganate bleach, in my experience, is a purple nightmare that will destroy your emulsion and etch away at the second varnish. My dichromate bleach is as follows:
1000mL H2O
9.5g Potassium Dichromate
12mL concentrated Sulfuric Acid
After the bleaching step is performed, the plate must be cleared. My clearing solution was:
1000mL H2O
50g Sodium Sulfite
A single damaged plate with "craters" where I monumentally screwed up coating the emulsion ended up being the only plate that I was able to reverse at all, due to the emulsion being particularly thin in those areas.
Looking forward
I'm not really sure what exactly is a good method for producing super thin coatings from home. The first thing that comes to mind is to dilute the emulsion considerably before coating, meaning less gelatin / mL of H2O on the plate. I assume there's an upper limit as to how much water the gelatin mixture can have before it just won't set up -- but I'm betting I'm probably pretty close to that already. I guess there's only one way to find out! While I work on that, I also need to re-mix my starch, hopefully to achieve a more neutral color balance.