Emulsion design software
I have been thinking (yes, I know, dangerous habit) that with all of the ingredients used in emulsion making, it should be possible to create a set of rules or algorithms that would represent each ingredient and its expected response (or something like it) when used in an emulsion formula. You could then create a java applet or some other type of program that would allow you to design the emulsion you want on the fly, before you started mixing anything. You could even have the applet display a picture and dynamically alter picture attributes as you change ingredient parameters (if you change temperature, concentration, amounts, mixing time, etc) to give a possibility of what the resultant emulsion would most likely produce compared to some sort of standard. Although I admit that might be a stretch. You might even be able to change the scale of the mix, but this might add more metrics than can be handled easily.
The real interesting possibility is that the same design program could be linked to control and feedback electronics and used to create the emulsion once a formula is selected. You could then just select the desired formula, hit a button "and go". I know this is way beyond simple emulsion making, but it should be possible to do. You might even be able to substitute ultrafiltration for the noodle-washing step and decant the finished emulsion into light tight containers ready to put in the fridge (I have to work on that one -- could be problems with that idea)
Not easy, maybe. But definitely possible.
Your thoughts are appreciated.
Such attempts were made by most major manufacturers and some even published. (Not the software itself, but the equations and basic algorithms. Polaroid people published such a paper with several measurement data obtained with their research scale plant) However, nothing works without a few tuning parameters obtained from the particular mixing vessel. Mixing vessel shape, mixer type, rotation speed, nozzle shape, and number of nozzles all have influence on the resulting emulsion shape. After that, different gelatin works differently. Some gelatin retard crystal growth more than others, etc., and the size of resulting emulsion can vary even if the formula and the procedure are identical. Feedback control with pAg measurement is useful, if it can be done easily, but it is not possible to remove variables associated with vessel and gelatin.
Last edited by Ryuji; 05-05-2007 at 01:57 PM. Click to view previous post history.
Bob, please note that Ryuji said attempts have been made. These things always sound relatively simple in concept but usually prove extremely difficult to actually do and very expensive. Not that I'm any expert in the photographic material manufacturing world, but I believe its probably beyond the scope and means of the individual.
Originally Posted by Ryuji
You have good sense. I didn't necessarily mean negative of such a project by using the word "attempts" but what I meant is that, even with a good theoretical models and numerical implementation, there are always tweaking parameters that need to be determined by experiments. In other words, simulations of emulsion making process is nothing like simulation of electric circuits, transmission lines or optics. In reality, amateur emulsion makers can't bother with such an iterative process to adjust the model parameters for the mixing vessel and other factors. Indeed, unitl 1970s lots of practical emulsions were made with gray-hair experience and trial and error.
At the same time, there is always some simple numerical calculation that needs to be done for each make. For example, if I do a simple single-jet make of simple AgBr emulsion for old timey bromide paper. I need to calculate the volume and concentration of KBr in the kettle, volume and concentration of AgNO3. Then I have to calculate the total volume and concentration of excess KBr at the end of precipitation. Geltin concentration is another factor. In case of single jet, I can get better emulsion if I use a large volume of more dilute salt solution in the kettle. But then I'll have to concentrate the emulsion by coagulation method (ultrafiltration is nice but I don't do it) more carefully. Calculation like this can be done with a simple calculator(preferably an HP one).
When I do double jet, the calculation is a bit more involved. Generally, the problem is like above, but I also have to calculate the flow rates of both jets to grow the crystal while pAg and supersaturation level are maintained. It's impossible to do this without VAg (pAg) monitoring and automatic control, but what a heck, we all have two arms. I adjust concentrations of each jet (always more concentrated salt solution) so that both solutions flow at the same rate, and I accelerate the flow rate during growth period. This, a bit more complicated than above, but not much more.
Either case, this level of calculation is not sufficient to "design" emulsion in any way, since there are a lot more important decision factors. Where in the crystal to incorporate iodide. Where in the crystal to incorporate a dopant, and what kind of dopant. How to desalt (wash) the emulsion. How to give it chemical sensitization. These factors have large influence on the speed, contrast, and reciprocity law failure. And these are getting better understood at the level of solid state physics. Actual control of the making process is a lot more dirty process.
emulsion making software was written before
I never take anyone's word without VERIFIABLE PROOF that something cannot be done, or is too difficult for an individual. This is not meant as a slam against you personally. Too often, "facts" are given which have absolutely no basis in reality, and are believed too readily by those who don't have as much experience or knowledge in those same areas. It is important to do your own research and checking. Often you will find the information presented as "fact" is just not so.
It is important to verify information and pronouncements before deciding on a course of action. To make any statement regarding emulsion making and coating, for example, without the benefits afforded by verification, to me, simply cannot be trusted and is essentially discarded out of hand. Verification as defined here can be demonstrated by charts, graphs, clear references to published and accessible reports, patents, research disclosures, or actual work done to demonstrate visually or physically limitations or failures or successes. I guess the short version is "Prove it".
I would like to draw your attention to patent number 5422825. You can look up patents for free at www.uspto.gov. This particular patent is titled "System for monitoring and controlling supersaturation in AgX precipitations".
The abstract of the patent says this:
"This invention is a system and associated method for monitoring and controlling independently the supersaturation and halide ion concentration, during the precipitation of a silver halide emulsion in a precipitation vessel. The introduction of a silver salt is controlled in accordance with a predetermined program, which in turn is responding to the supersaturation signal measured in the vessel by a supersaturation sensor. The sensor is comprised of a true silver ion electrode and a silver-silver halide electrode. The introduction of the halide salt is controlled in accordance with the predetermined program, which in turn is responding to the halide ion concentration signal measured in the vessel by a silver-silver halide electrode and a reference electrode."
If you look at the first drawing of patent 5422845 carefully, you will find a very good drawing of one example of just such a system.
In patent 5248577, titled "Reactant concentration control method and apparatus for precipitation reactions" the abstract has this to say:
"A method and an apparatus for controlling reactant ion concentration within a precipitation vessel to optimize the reactions occurring therein is disclosed. In one important implementation, the reactant controlled comprises halide ion in a reactor during the precipitation and growth of silver halide grains. In this embodiment, the method includes measuring the amount of halide ion within the precipitation vessel; measuring the flow rates of the halide feed stream and the silver feed stream; storing the measured halide ion concentration and measured feed stream flow rates; computing an estimated smoothed halide ion concentration value as a function of the stored measured halide ion concentration and measured feed stream flow rates; and adjustably controlling the amount of halide ion added to the precipitation vessel as a function of this estimated halide ion concentration value. Preferably, N most recent measured halide ion concentrations and measured halide and silver feed stream flow rates are stored and used in determining the estimated smoothed halide ion concentration. Mathematical algorithms are derived herein to facilitate the computation of the estimated halide ion concentration and the necessary flow rate adjustment. Analogous apparatus is also provided herein."
Other patents that are referenced go back decades. My point here is that it has been done before with more primitive equipment, more primitive PC's and software on a small scale. They did it, we can do it too. Definitely. And you may find it is not as hard as you were told. Just detailed.
Last edited by rmazzullo; 05-05-2007 at 07:25 PM. Click to view previous post history.
Reason: forgot to add title to post
Sponsored Ad. (Subscribers to APUG have the option to remove this ad.)
That's excellent Bob. I hope it works.
Originally Posted by rmazzullo
I don't know about that. US Pat 5422845 is Method and device for improved programming threshold voltage distribution in electrically programmable read only memory array.
5248577 is as described.
Anyway, the sensors mentioned in these abstracts cost at least a couple of hundred bucks each. Durable and accurate ones typically used for process control can exceed $1000. A good problem is that these probes are rather bulky in small reacting vessels. But they must be immersed in the bulk solution in such a way that they don't create pockets where the solution doesn't get mixed well. The nozzles of silver and halide jets in the diagram of the second patent is above the solution surface, but that's intentionally deformed. I can't imagine modern Kodak plant using jets that drip solutions into the vessel like that. Usually the nozzles are immersed, often with multiple fine endings, and they are always located very close to the mixing propeller(s), often in a smaller immersed chamber. These silver ion sensor and halide ion sensors are best situated in the immersed chamber, and another set placed in the bulk solution. I wish I had a small research mixing vessel like that but making such thing with 316 stainless or titanium is very expensive.
Anyway, you can go through many of these things, and they are good as a making process controller but not as an emulsion design software. There are too many tweaking parameters taht require experimental measurements. Without doing this, the system can't answer questions like this:
What is the best flow profile or model to make a tabular grain emulsion of average diameter 0.6 micron and average thickness less than 0.04 micron, and optimize everything to minimize reciprocity law failure but the emulsion has to have a shadow speed exceeding equivalent of ASA 100? This is by no means a trivial question but experienced emulsion engineers can tell you a couple of ways to do this in the equipment, gelatin and a host of additives they are familiar with.
Well, like any project, this is doable if you set a realistic goal. But to make an emulsion design software without tedious experimental calibration, you'll need something Kodak and Fujifilm do not have.
Last edited by Ryuji; 05-06-2007 at 12:19 AM. Click to view previous post history.
And what on Earth could that possibly be? Are you implying that you know, and can do better than, Kodak and Fuji? If so, then you must have done this work before. Can you provide some evidence to back this up?
Originally Posted by Ryuji
Last edited by rmazzullo; 05-06-2007 at 06:16 AM. Click to view previous post history.
Reason: incomplete statement
The patent number I mentioned in my earlier post is 5422825, not 5422845.
Originally Posted by Ryuji
This work is doable and has been done
At Kodak, the first complete automation was constructed at plant, pilot and research scale in the early 70s using simple controller boards.
In 1982, the first front end CAD/CAM type program was built to help the emulsion engineer design an emulsion sitting in front of a computer. This program included a vAg model, a halide distribution model, a mixer model, and a volumetric to gravimetric conversion program.
This latter meant that we could use either volumetric or gravimetric measures and could vary the temperature and even mix salts and still have density-volume measures accurate to 4 or 5 decimal places from 20 - 40 deg C.
There were warnings built into it to enable the emulsion maker to design an emulsion that could not be manufactured, but still let him make the emulsion at research scale.
At this time, there was a set of programs developed to run the making station computers that accepted the code from the CAD/CAM program allowing total operation of the system with no operator intervention except for making the chemicals up, filling the containers and pressing the start button. On screen color display showed the process and plotted the vAg, flow rates and temperature.
In 1988, the first Kodak wide standards committee met to plan the second level of this CAD/CAM program and upgrade it. These all ran in DOS or DOS like programs on a PC or Taylor computers depending on the scale of the project. The CAD/CAM program and process control software were designed to accept the standards at Kodak Park, Harrow and Chalon. There were 4 different process types in use at Kodak Park, and two each at the other sites.
The emulsion maker had a simple nomograph to pick the proper vAg values and the starting flow rates for a given emulsion type.
After the run, the program then allowed the user to attach sizing plots, photo micrographs, run time data and a graph of the process as it actually ran, compared to the pre-run prediction from the model.
The program had a set of pop-up tool kits that allowed instant calculations to verify that the starting information was indeed correct, This included vAg to pAg conversion, pBr, pI and pCl calculations, the effects of nitrate ion on vAg and the enthalpy (heat of reaction) taking place at the various scales so that the experimenter could back calculate the desired temperature of the starting solutions was correct.
The mixer model was so precise, it gave not only the mixer rate for the volume, but for the kettle as well and included a differential for the change in volume over time. It also included a splash guard so that if the speed of the mixer was predicted to be high enough for the volume of solution in the kettle, the user was warned. Otherwise we had emulsion everywhere.
The model also included valving positioning information and timing.
I think that this shows, to some degree, the level of sophistication of the Kodak software. These were all linked on a private 'internet' so that formula information could be linked between all emulsion makers.
The Kodak patents on this issued in 1993, but comparable Agfa and Fuji work was less extensive and some did not appear until much later. The Lin patent shows a diagram with a schematic of the entire automated process but it shows his improvements being implemented at that time. As you see, it shows up to 2 silver and 2 salt pumps. This is not all inclusive, as you can imagine.
Version II, was begun in 1995 with the introduction of Windows 95, it being felt that the Windows 3.1 system was too chancy for our level of complexity. We used DOS and wrote our own pop-up windows and windowing utilities. Kodak is now at version III, I would call it, but there have been 3 different names for the processes and programs.
I am including the Version II desktop icon which I kept a copy of. I knew it would come in handy somewhere. I had to convert it from a .ico file to a .jpg file to upload it.
I hope you find this information interesting.
Oh, BTW, from 1988 until I retired I was the project leader! I was a member of the project from 1982 though. During that time we worked with every emulsion and emulsion maker at Kodak, world-wide, and literally thousands of emulsion formulas at all scales and of all types were handled by our group.