Scaling up emulsion recipes

In emulsion-making, like many other pursuits, once you get up the initial learning curve of following a recipe to produce a good photographic emulsion, a whole new world of exploration seems to open up in tinkering with the recipe. In my case, I’m interested in successfully making larger batches at a time.

I’ve had some hard-earned experience lately in what *not* to do, and can share some insight from that experience while creating a topic that I haven’t found any discussion about here on apug (or anywhere else for that matter).

When scaling a home-brew emulsion-making recipe to produce different volumes, the ratio of ingredients and temperature hold times should all remain the same. Scaling the amount of ingredients is straightforward, temperature hold times are fairly straightforward as well.

A tricky aspect especially for scaling up is maintaining the same cool-down times. For first ripening and chemical/second ripening, this means controlling the cool-down time after the hold at the specified ripening temperature(s). The actual cooldown times can be measured for the original volume, and then replicated for the scaled amount. For larger volumes, the emulsion can be divided into several containers of original volume and then combined as the emulsion is setting (mind the cooldown rate when it’s in the fridge) or active cooling can be used to control the rate of temperature change (i.e. putting tubing in the emulsion with cool water flowing through as the emulsion is constantly stirred)..

The other hiccup is how the precipitation stage is impacted by the larger volume of silver nitrate and salts that need to be combined. Kodak AJ-12, for example, calls for adding silver nitrate to the salt solution at a rate of 20ml/30seconds for 10 minutes followed by 10 minute ripening. Scaling up the volume 4x (for example) would mean precipitation iat 20ml / 30 seconds takes 40 minutes. This entirely changes the timing of the ripening and causes significant problems with the resulting emulsion due to the change in the distribution of the grain size as well as longer heating of the gelatin.

So obviously not only is the precipitation rate inportant, but also the total time required to combine the solutions. What I’m curious about is, when scaling the volume, can the precipitation rate be increased (shortening the total time of precipitation) to *approximately* result in similar grain size distribution as the emulsion made in the original volume, and what does the scaling relationship look like?

As I think about how to scale the recipe properly by adjusting the precipitation stage, my hypothesis are as follows:

I think to ensure the grain size is kept small enough to remain in suspension for the % gelatin in the recipe, the total precipitation time should remain at or near that of the original recipe. This will of course change the characteristic curve of the emulsion (my though is lower contrast and more delicate tonality in shadows and highlights)

I think you would also want to increase the amount of gelatin added after ripening. Or, perhaps start with a larger amount of gelatin in the salt solution to reduce grain growth and maintain a precipitation rate closer to the original.

Of course, to avoid these variables, the original volumes could be made in serial and then combined for washing. But then of course you’re not learning anything (even if you have to learn the hard way).

There’s other aspects to worry about: storage, preservation, etc., but my focus is on correctly scaling a recope to produce same or similar results. Or perhaps recipes cannot be scaled up, and doing so requires making one from scratch (or finding a different recipe specifically put together for the amount you want to make).

Thoughts from our learned emulsion makers?

-Jason

Jason,
Excellent thoughts.

Scaling is indeed a big bugaboo. As I understand the history of emulsion research, scaling is one the most tightly held secrets because it's so important for the research-to-production pipeline.

Here are my thoughts ("learned" or not ). Start at 2x. Double the amount of each addition proportionately. Cool down time can be hastened by a cool or cold waterbath. The characteristic curve should pretty much stay the same that way.

The only reason to increase the precipitation gelatin is to mitigate mixing problems. Another issue with scaling is the change in how evenly the ingredients combine. A magnetic stirring bar might combine 100 + 100 ml, but have "dead zones" if the amounts are 200 + 200 ml. At some volume, the mixing arrangement you are currently using will break down. If you change your method, at say 4x, and you get peppering, but all else is good, you might try increasing the precipitation gelatin. Changing the amount of precip gelatin will likely affect the characteristic curve. It might also slow the speed, so extra ripening might come into play there.

I know you know that this will take some trials. Fun! p.s. you don't need to worry about preservatives in your emulsion. If it goes bad before you get it all coated, you made too much .

Jason, all valid thoughts except for changing precipitation times and gelatin. All times remain constant, but the rate of change in temperature must change. As scale goes up, mixer speed goes up and up and up until you reach a point - believe it or not, that you must come down in speed and improve efficiency or it cannot be scaled. I have seen 10L makes with efficient mixing at 6000 RPM but the small 500ml makes work fine at 300 RPM with a stirring bar. I have had to use a bottom (bar) and a middle (propeller) positioned mixer to achieve proper mixing. I have seen 100L makes with high efficiency mixers made at 300 RPM again.

Scaling for me has been straightforward from 100 ml to 1L. I have had problems above that. The problems were related to heat exchange. Above 500ml I found that I needed thick wall beakers and/or a water bath.

PE

I’ll go through your responses in detail later but just wanted to quickly say thanks for your replies. I thought you two would have some thoughts to share. . I hope you are both doing well.

I’m currently mixing at 1.3L volume with good results. To meet demand for plates, I scaled up to 1 gallon batches. I kept the precipitation rate the same (so total time to combine the teo solutions was ~4x longer), and the cooling down times were much longer. Remember the other thread about swirlies that I had a heck of a time troubleshooting? I finally traced the cause to scaling to the larger batch size...they disappeared completely when I started making batches at the original volume again. So of course I am analyzing what changes with scale.

I think the actual swirls are silver halide crystals too large for the emulsion to hold in suspension and/or fogging. Hence my train of thought on proportion of gelatin.

Just to be clear on precipitation: increase flow rate to maintain the same overall timing, and compensate with increased agitation (and this only works up to a certain mixing speed)? This should be doable, as my mechanical stirrer gets up to 3600 rpm. It can stir fast enough to actually foam up the 1 gallon batch like a cappucino.

The side effect of troubleshooting the swirls is that, once they went away, the coating quality is much better and more consistent.

Sometimes you’ll see references to Eastman’s Dry Plate company almost going under in its infancy due to technical problems he had to overcome. I sonetimes wonder if he encountered these same problems, but I could never find a book that described his startup to that level of detail.

-Jason

From my understanding the form of a vessel or kettle plays a role as well as the stirrer’s. The deciding differences may come from the streaming patterns. You want as much volumes combining at a given rate in larger sizes as regular ones. That calls for multiple inlets and stirrers. I don’t think that whipping up like a cappuccino is most helpful but should rather seek for the best digestion, if I may put it so. A vertical helical and radial flow seems to be the thing to aim at. Then you can even increase precipitation rate. There’s one best spoon for every coffee cup. The cappuccino again . . .

Ok, this gives me something to work on. Good point about the mixing head and container. I can probably fabricate a different shaped mixing head and/or find a better one (perhaps a stainless steel paint mixer). I have multiple parastaltic pumps so I can support testing of higher flow rate.

Question: Would rapid agitation be maintained into and through first digestion?

Avoid foam at all costs. It can distort the emulsion precipitation reaction because it can force crystals to form on the surface of the bubbles which is ungood.

For mixing, we used a PEPA or Planar Emulsion Precipitation Apparatus.

Attached is a rough depiction. I have a better one somewhere, but can't find it right now. Basically, it consists of two parallel disks with holes in them. Solution is added above and below, and mixes in the gap between them and is then spread out to the sides. It is sometimes called a spreader. There were probably up to 4 generations of these used at one time or another.

PE

Ron, funny coincidence in that I was thinking of a similar approach if needed, whereas a tube (tubes, actually) would bring the silver solution from the multiple parastaltic pumps down to near the mixing head on my stirrer. I come here to find your illustration.

Before I get to that level of complexity, I need to overcome the primary hurdles of correcting the timing issues.

First, I need to set up multiple lines to add the required amount of silver nitrate within the required time. That’s fairly straightforward. I also acquired a stainless steel paint mixing head, which should be a more efficient mixer than what I have now (doing some food dye mixing timing tests tonight).

Then, I need to control the rate of temperature cooling (and heating after washing). I have those times recorded for the smaller batch size. I think for my water bath setup that will be straightforward as well...especially with snow on the ground outside.

I’ll start by scaling up 2x (from 1L to 2L). Once I can get those right, then I can see where I’m at with coating quality. My guess is fixing those two things will help immensely.

Have a look at your local HomeBrew supply house at the wort chiller's. They are normaly copper, but you could make one from stainless or plastics. You could even use them for heating a batch as well as cooling.

2x scaling setup. Two peristaltic pumps each set for 40ml / min to precipitate all the silver nitrate in 10 minutes. 2-channel ower supply on the right. The light source on the left is a red LED lamp that doesn’t fog plates after at least an hour exposure.



Not easy to see, but the tubing coming from the pumps are on either side of the mixing motor. You can see at least one of the streams of silver nitrate flowing into the solution. Thermocouple on the left for the PID controller which I use to control temp of the water bath and thus temp of the solution.



I was able to replicate the temperature timing of my smaller batch size with ease (in fact it cooled down slightly faster). Currently in the lab/beer fridge cooling down to set for washing tomorrow night.

Nice setup. I used a timed syringe addition or a pump, and was very careful. Things worked out for me.

See my setup in the book. If you missed that, I can post some of it.

PE

Thanks, Ron. I was too late in acquiring a copy of your book, so I would love to see your setup if you are willing to take the time to share.

It's always satisfying to observe when engineering principles appear across disciplines. I'm guessing that -- like simple optics -- simple emulsions are rather forgiving in parameter tolerances and a timed syringe works fine. I originally started with a timed syringe, but quickly went with the pumps so I wouldn't have to stand there. You might have caught on, but I only have a relatively short hop now to go from a single-jet setup to double-jet precipitation. If I can detect the concentration of silver ions in the solution (i.e. measure conductivity), a controlled double-jet precipitation wouldn't be too far behind.

Another question: As you can see, I used two pumps to deliver the silver nitrate through two streams, each at the required flow rate to deliver both pictured batches in the necessary time. What differences arise if a single pump is run at twice the rate to deliver all the silver nitrate from a single container in the required time period? Does the agitation change between the two methods of delivery?

-Jason

JThis evening I washed, chemical ripened, filtered, and added finals for the 2L batch. I’m now up waiting for a couple test plates to dry so I can inspect (a 4x5 and couple of 35mm test plates). So far the drying emulsion looks clean and sexy. I’ll take some test shots tomorrow.

-Jason

Jason, we often used multiple pumps or delivery tubes to make an emulsion. The trick is to deliver the same # of moles / minute with respect to the scale you are working at, so that the scaled time vs moles is the same across the entire scaling sequence. Agitation should go up, and heating and cooling rates should be consistent with the scale.

We took things very seriously regarding enthalpy when scaling. Adding too much too quickly could cause a huge temperature spike and ruin a scaled up system. Often, a large scaled up batch had to add cooler solutions due to heat of reaction. We had a calculator for enthalpy to correct temps for us.

PE

Here is the whole thing. Included in this is the vAg electrode setup with salt bridge and the pump with holder for source beakers. No mixer is shown here.

PE

Photo Engineer said:

Here is the whole thing. Included in this is the vAg electrode setup with salt bridge and the pump with holder for source beakers. No mixer is shown here.

PE

Click to expand...

Thanks for the photo.. I'm proud to say I can actually make sense of that setup.

Test image revealed I lost about 2 stops of speed (I'll do a more accurate test tonight), but the quality of the coated emulsion itself didn't have any problems. Theoretically I can make up for that speed loss with the addition of an appropriate amount of dilute sulfide or gold sensitizer, both of which I conveniently have on hand. So I can salvage this batch for sensitization testing.

Based on your statement regarding addition of mol silver / minute and making sense of my batch results ... if the end-goal is to double the final volume of emulsion, to a first order one should double the total precipitation time, temperature hold time, cooling rate, and heating rates... and then account for enthalpy?

I say again.... Do not change overall time!!!!!! Just change the rates of addition, heating and cooling.

PE

Thanks for the clarification. I thought that’s what should be done, but I misinterpreted your wording and it had me confused.

-Jason

For the latest batch, I rigged up to do under-the-surface precipitation. I connected my paristaltic pump tubing to straight glass tubes so I can inject silver nitrate right at the stirrer.

Sub-surface precipitation seems to be a good move. i have scaled from 100ml to 500ml without problems that way. Did you ever figure out how to measure vAg? I built the salt bridge and electrodes described by Ron and had success with it working. Controlling vAg is another matter entirely!

I took some test plates but didn’t have a chance to scan them before I went on travel. They are a little slower than previous plates..I think my precip/first ripening temps might be a little low. My temperature sensor needs recalibration.

I haven’t started trying to measure vAg yet, although I have been looking into it. At this point it’s not quite a necessity for the emulsion I’m making. Plus I have about 3 other projects on the list before that. Another low hanging fruit now that I’m doing subsurface precipitation is the mixer head..I’m going to replicate the one in Ron’s book. I’ll use a silver rod and I can turn the head on my lathe.

So much to do and so little time.

Controlling vAg is quite difficult and requires a proper control algorithm to run the salt pump plus a PID control algorithm. (ask Jason about PIDs)

PE

Photo Engineer said:

Controlling vAg is quite difficult and requires a proper control algorithm to run the salt pump plus a PID control algorithm. (ask Jason about PIDs)

PE

Click to expand...

Yep... that’s why measuring vAg is down on the list. It’s more than just measuring the value. You need to do something with it. Not insurmountable from an engineering standpoint (feedback control systems tend to have a lot of familiar characteristics ), but not a weekend project either.

Don't forget that pump chatter (or on/off cycles as well) will introduce irregularities into any salt being controlled. This becomes a huge problem in some cases. Also, don't forget a salt bridge to prevent contamination and other problems.

PE

For pump chatter you can use back pressure or switch away from a paristaltic pump.