Effect of emulsifiers on LLVE?

[TiCl4]

I have emulsion polymerization reactors that have a various amount of water, miscible and immiscible monomers (styrene, vinyl acetate, various acrylic polymers), and emulsifying surfactants. In the past, to evaluate peak pressure from runaway scenarios I have treated the system as a completely immiscible system for the liquid-liquid-vapor equilibrium, comprised of water in one phase and the organic monomers in the other. This was the most conservative approach, as each phase contributes independently to vapor pressure and gives the largest potential pressure at a given temperature.

However, in recent calorimetric sizing for vents from our reactors, developed pressures were significantly less than I had calculated using the completely immiscible assumption. Even the small amounts of miscibility that are normally associated with water-monomer mixtures (perhaps 1%) would not cause this significant deviation from that assumption. I can only surmise that the emulsification of the monomer somehow affects the vapor pressure. Does anyone have any theoretical background on what effect emulsification agents have on LLVE?

It would be nice to have a better means of estimating the LLVE, as I could then start exploring heat loads on overhead condensers theoretically as well as emissions speciation through our emissions control system.

 

[Latexman]

The emulsifiers affect the liquid activity coefficients, not the pure component vapor pressures, but the end effect is the same.

I view it as follows. Most emulsifiers have a "water-like" end and a "organic-like" end. The water-like end is "anchored" to the water, and the monomors are held by the organic-like end and the molecular chain in between both ends.

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[TiCl4]

Agreed - my reference to vapor pressure was regarding the total vapor pressure for the organic phase, not the single component vapor pressure. Since emulsification is not the same thing as solubilizing (or is it from a thermo perspective?), I was curious if this could be treated theoretically by any activity coefficient model, or if this is more suited to empirical determination.

Note: I don't have access to ASPEN or any other modeling software. I can get DWSIM if it would be helpful.

Check my understanding of the thermodynamics at work here:

With LLVE, the total pressure of the system is the combination of the vapor pressures of each phase. Things are more complicated when components are partially miscible in each other. However, for completely immiscible phases, the total pressure is simply the sum of the individual phases' total vapor pressure. This means that total pressure in the vapor space can exceed atmospheric pressure well before reaching the boiling point of one of the individual components. I.E. a mixture of ethyl acrylate and water at 180 F has pure component vapor pressures of approximately 8.4 psi and 7.6 psi, respectively, for a total vapor pressure of 16.1 psi of the vapor above this mixture. If the reactors are operated at atmospheric pressure (with a vent), this will lead to continual evaporation as the system tries to equilibrate at 16.1 psia, correct? If that is correct, how is the rate of evaporation determined?

 

[PaoloPemi]

for liquid-liquid equilibria any suitable thermodynamic model starting for example from NRTL can give accurate values once properly tuned,
for medium-high pressures you can prefer a combinations of a EOS + liquid activity (for example Peng-Robinson NRTL) with complex mixing rules,
I have access to PRODE PROPERTIES (there is a free version available to everyone) which includes a VLE-LLE-SLE data regression utility,
for liquid-liquid equilibria BIPs have a large impact...
I have found good LLE data points (required by data regression utility) in literature or well known databases as DDB

 

[Latexman]

Your understanding is correct. My company never got into modelling emulsion polymerization reactors. It is very, very complicated. Our largest raw material is water, of course, but then we have 5-10 major monomers, 50-ish functional monomers, 100-ish emulsifiers, and numerous initiators, reducers, buffers, defoamers, biocides, etc. Several hundred chemicals. The sheer volume of pure component properties, VLE data, LLE data, binary interaction parameters (BIPs), etc. would be quite time consuming in itself. Then you throw in reaction kinetic models and it gets immense. And then, it's doubtful all this would be realistic enough to make business decisions. So, we have a good many 1 gal. glass lab reactors and technicians. From a product development standpoint, we have found that to work well. Don't get me wrong, we have not abandoned thermodynamics totally. We just don't go crazy into the details. We do enough real time mass and energy balances and some basic thermo around the reactors to keep the accumulation of monomer under certain limits.

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[TiCl4]

Latexman,

We have a similar number of components that makes the approach you talk about (developing the interaction parameters alone!) an overwhelming task. I was hoping there were some major simplifying assumptions (i.e. grouping surfactants by type, i.e. anionic, non-ionic, and cationic, ignoring biocides,initiators, etc) that could provide a reasonable way to pare down the list of parameters that need to be developed.

This isn't just for theoretical hijinks, either. I was one of the main people specifying a new condenser for a large reactor. Having a means to predict the load requirement directly for the condenser would have been a huge help during the design phase (we had to work backwards from the data from our emissions system on a currently operating reactor/condenser system). As it was, the only design basis I got from other engineers in the company is that they have historically found the X sqft heat exchange area per gallon of reactor size is sufficient for certain chemistries. This reactor was to handle different chemistries that would require much less heat exchange area, and they had no historical basis for this.

I wonder how their first reactor condenser's were sized without that historical basis all those years ago...

 

[Latexman]

Maybe it was scaled up from the glass condenser on their 1 gal. glass RX!

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[TiCl4]

The image below seem appropriate.

 

[georgeverghese]

"I.E. a mixture of ethyl acrylate and water at 180 F has pure component vapor pressures of approximately 8.4 psi and 7.6 psi, respectively, for a total vapor pressure of 16.1 psi of the vapor above this mixture."

In a mix, the vapor space partial pressure from each component is in the ideal case, proportional to the mole fraction of that component in the liquid phase also

P = PP1 + PP2 + ... + PPn = Σ Yn. P = Σ Xn. Pn, where Pn is the pure component vapor press of component n at that temp - this is what we know as Raoult's Law.

Another reason for non ideal behaviour may be azeotropes, in this case negative azeotropic behaviour.

 

[Latexman]

Yes, but ethyl acrylate and water are immiscible and there is two separate liquid phases, each exerting it’s own Σx_nP_n

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[georgeverghese]

Okay,agreed, that would be the case if the organic phase is all ethyl acrylate and is immiscible with the all water phase. From the OP's first post, I got the impression there was more than one component in the organic phase.

 

[georgeverghese]

The Dechema database on VLE / VLLE is probably the largest of its kind - see if you can find these components in this database. The full set is only available by subscription.

 

[chemlite]

I have the same feeling, the OP first post said "water, miscible and immiscible monomers (styrene, vinyl acetate, various acrylic polymers), and emulsifying surfactants." but then he proposed to consider components as immiscible...
That assumption can produce large errors and many authors (see for example Prausnitz's books) have proposed specific models for liquid-liquid, see for example the NRTL mentioned in previous post,
also there are many books and papers discussing properties of emulsions as for example fugacities, which are affected due to the different volumes, one can find many papers discussing stability in emulsions with obviusly many references to water-oil but also different fluids.
I agree with PaoloPemi that for complex phase equilibria one should start from measured values or reliable models in order to get reliable results, there are not alternatives.

 

[pierreick]

Hello,
my experience with latex and resins : scale up from glass reactor and ancillaries ( few liters ) to pilot reactor ( few hundred liters) to industrial reactors ( a few m3 to 40/50m3 ) .
For pilots and industrial reactors quite often 2 condensers , one supplied with cooling water ( main) and a second supplied with chilled water (safety condenser). Most of the time those reactors are multipurpose meaning additional surface of condensers for safety purpose.
an SB latex recipe is about 20 ingredients .
My experience
Pierre

 

[Latexman]

He was simplifing the math, folks, for discussion reasons. An EA/water latex is predominately EA and water with 1-5% of other stuff, like surfactants, initiators, reducers, buffers, defoamers, etc. 20 ingredients is about right, but > 95% is EA and water. The solubility of EA in water is 1.5 g/100 mL @ 20^o C (Wikipedia). I know it forms a heterogeneous, minimum-boiling-point azeotrope at atm pressure, but this virus has me working at home, and my data on the azeotrope is in the office.

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[pierreick]

Hi latexman,
you remind me the use of dean stark apparatus to control the quality of FG !
Pierre

 

[TiCl4]

Yes, the EA/water example was just to illustrate the confirm my basic understanding of LLVE. We don't actually make EA homopolymers, but it simplified the math for the example. Due to the solubility of monomer in water and visa versa, the actual vapor pressure developed in a pure EA/water mixture is lower than would be predicted by purely immiscible phases.

However, based on experimental data collected during calorimetric testing, the actual developed pressure was much, much less than would be predicted by compensating for the miscibilities of the monomers.

I thought there had to be an additional effect present, and thought that the emulsifying agents were the prime candidate due to the nature of their interaction with both phases of the LLVE. Thus this post.

I think Latexman has answered my question quite sufficiently - emulsifying agents tend to depress the activity of the monomers. However, gathering the required thermodynamic data for all possible combinations of components present would be a monumental undertaking, and using simplifying assumptions to pare the list of required data gathering would result in large enough errors that the model would not be useful or predictive.

I have not done much with modeling in the past, but is there any functional group contribution model (like UNIFAC) that can handle such systems ex-ante, with no need to gather interaction data?

 

[Latexman]

Yes, we have a few of those in our labs. Mostly on our SBR glass RXs.

Good Luck,
Latexman
Pats' Pub's Proprietor

 

[PaoloPemi]

TiCl4
in DDB or Dechema you can find both VLE and LLE data for many binary and ternary mixtures including those which interest you,
some values are free

http://www.ddbst.com/en/EED/VLE/VLE Ethyl acetate;Water.php

yes, there are also predictive models for liquid-liquid phase equilibria but these are considered less accurate than the equivalent for VLE (UNIFAC etc.)
the reason is very simple, with liquid-liquid you have similar volumes and fugacities where with vapor-liquid you see (in general) quite large differences,
this makes things worse since small differences can originate very different results,
which explains the importance to adopt accurate models.

 

[georgeverghese]

There is also some new reading material in the 7th edition of Perry Chem Engg Handbook on azeotropes and azeotropic distillation (in the chapter on distillation) that may be of interest.