Chemical Engineering MagazineChementator Briefs
New adsorbent Last month, BASF SE (Ludwigshafen, Germany; launched…
ConvalFocus on Valves
These versatile, modular valves enable quick installation More than 2,500…

Comment PDF Processing & Handling

Liquid-Liquid Extraction: Generating Equilibrium Data

By Don Glatz, Brendan Cross and Tom Lightfoot, Koch Modular Process Systems |

Equilibrium data and related information gathered from a liquid-liquid extraction laboratory “shake test” can provide information for process feasibility and column-type selection in the scaleup of liquid-liquid extraction processes

Most chemical engineers have had the experience of dealing with problematic separations, and most have a general understanding of distillation processes. When it comes to liquid-liquid extraction (LLE) processes (Figure 1), however, the details of how these processes work are often less clear. Most academic chemical engineering degree programs do not heavily emphasize liquid-liquid extraction, and most chemical engineering graduates did not receive more than a few days of instruction on generating equilibrium data for LLE in their degree programs.

Figure 1. Commercial-scale liquid-liquid extraction processes often transfer solutes from an aqueous phase to a solvent

Figure 1. Commercial-scale liquid-liquid extraction processes often transfer solutes from an aqueous phase to a solvent

This article is focused on going “behind the scenes” and revealing more about the earliest stages of generating LLE equilibrium data. It describes how to perform a series of laboratory “shake tests” to calculate LLE equilibrium data. Furthermore, the article explains how to use those data to determine the type of column that should be used in a given LLE application.

LLE process basics

LLE is a technique that exploits differences in the relative solubility of compounds of interest (the solute) in two immiscible liquids, most often an aqueous phase and an organic solvent. In an LLE process, a liquid stream that contains a compound of interest is fed into an extractor, where it will come into contact with a solvent. To allow for phase separation, the solvent and liquid stream are immiscible, or only slightly miscible, and have different densities. The two components are mixed to promote close contact between the two components, and to allow the transfer of the solute into the solvent phase.

The two main types of columns discussed here are rotating columns and reciprocating columns. There are other methods of separation, such as utilizing static (structured packing) columns, but those find limited use due to low efficiency and lack of flexibility.

Figure 2 shows a typical LLE operation. The feed containing the solute (B) is introduced into the extractor, where it transfers into the solvent. Solvent recovery plays a large role in the overall process economics. The raffinate is stripped to remove solvent in a separate operation, typically a stripping column. The solvent in the extract stream is typically recovered in a distillation column.

Figure 2.  A typical LLE process includes an extraction column, a raffinate stripping unit and a solvent-recovery system

Figure 2. A typical LLE process includes an extraction column, a raffinate stripping unit and a solvent-recovery system

Figure 3.  Laboratory shake tests generate liquid-liquid equilibrium data that can provide information for scaleup decisions

Figure 3. Laboratory shake tests generate liquid-liquid equilibrium data that can provide information for scaleup decisions

In cases where a separation can be accomplished economically with distillation, an LLE process would not generally be used, but in cases where distillation is not feasible, LLE is often the best type of process to use. Distillation may not be feasible because of high process complexity, heat-sensitive materials, low volatility or prohibitive energy requirements. LLE can be used to break azeotropes, and when a complex distillation sequence would be required.

LLE is used across many sectors of the chemical process industries (CPI), including in the chemical industry for extracting high-boiling organic materials from aqueous streams, washing acids, bases and polar compounds from organic chemicals, in the biotechnology industry to recover compounds from fermentation broths, in the petrochemicals industry to separate olefins from parafins, in the food industry to decaffeinate coffee and tea, and many other applications.


Shake test

Before setting up a LLE process, equilibrium data should be generated to define how the solute of interest behaves in the two immiscible phases. The data generated can then be used to construct a curve for the partitioning behavior of the solute (an equilibrium curve).

Data for liquid-liquid equilibrium are generated using the shake test, a method that allows for the calculation of solvent-to-feed ratio, versus the number of stages the process will require. Liquid-liquid extraction “shake tests” were historically done in a separatory funnel, which was hand-shaken. This is where the term ”shake test” comes from. The older method of hand-shaking was complicated by variations of human practice, including varying agitation speed, time of tests and so on. It is important to note that current shake tests are much more standardized, with the same agitation speed, time and characteristics used.

The selection of which solvent to use with a particular feed depends on a number of different factors, and often requires a great deal of expertise to achieve success. Among the considerations is the relative volatility of the solvent, if it needs to be recovered (this is typical in commercial-scale systems). The ultimate decision of which solvent to use comes after evaluating the feed and testing multiple solvents. The correct solvent choice is usually the result of an experienced team.


Shake test steps and equipment

The following paragraphs describe how to perform a laboratory shake test for collecting LLE equilibrium data. In the shake test method, a jacketed, round-bottom flask with a standard half-moon impellor and a bottom outlet is used. Tempered water flows through the jacket to control the temperature during the test. The feed and solvent are carefully weighed and added to the round-bottom flask at the desired solvent-to-feed ratio. Next, the agitator is turned on slowly and the contents are heated to the desired temperature. Typically, a mercury thermometer is used for temperature measurement, however other methods, such as using a thermocouple, can also be used.

Once the desired temperature is reached, the agitator speed is increased. For most extraction applications, the agitator is run at high speed (greater than 1,000 rpm) for exactly two minutes, at which time the agitation is shut off and the phases are allowed to separate. The authors have learned that there are two types of systems that may require more than two minutes of mixing in order to reach equilibrium. The first is when one or both of the liquids are viscous and thus, mass transfer is slow. The second type of system would be one that tends to emulsify very easily, usually due to low interfacial tension and low density difference between the phases.

For systems that emulsify, low agitation speed will be required to prevent emulsification and thus a longer mixing time is required. Upon phase separation, the time for the phases to separate should be timed to determine the separation time to reach a relatively sharp interface. The phases are then drained and weighed to determine the material balance. Samples from each phase are taken for analyses of all specified components.

After the first mix-decant sequence, the raffinate phase is returned to the flask and fresh solvent is added at the same solvent-to-feed ratio (S/F) and the procedure is repeated. This is repeated as many times as necessary to produce raffinate with the desired solute concentration (usually 4–6 times).

During the phase separations, qualitative observations are made, including what happens when the liquid phases separate. This is where a long history of experience in LLE scaleup is critical. Experience with many types of feed and solvent systems in the laboratory allows engineers to observe the performance in the laboratory shake test and use the results, together with observations, to select an extraction column technology for pilot-plant testing and scaleup. Some of the key observations that should be made are the following:

  • Does the entire system emulsify, thus resulting in very slow separation or no phase separation at all?
  • How quickly do the phases separate, and after they do, is a clear, sharp interface obtained?
  • Is entrainment observed in either of the phases where the entrained amount does not separate with the bulk of the liquid from that phase?
  • Is there an emulsion band present that forms at the interface, and if so, how long does the band take to break?
  • Are there solids that build up at the interface between the two phases (this is known as a rag layer)?

Reporting results

The shake typically returns log sheets for each of the experiments performed. If the shake tests are performed by a team of experts in LLE, you should expect a full report of the experimental results, including the raw log sheets from each of the mix-decant experiments, as well as a summary of the log sheets and analytical data. This should be presented in a table that shows full mass balances. In addition, the report should include recommendations, based on the shake test data, on the type of column that would be best suited to the separation project under discussion.


Advantages of shake test

There are several key benefits of the shake test methodology for the collection of information for LLE process scaleup. First, it is relatively inexpensive and can be completed relatively quickly. Two or three days is typically sufficient, depending on the analytical setup that is required. Shake tests allow a great deal of information to be learned quickly and without great expense. Although some solutions can move directly into pilot testing because the LLE equilibrium data are already available, having the qualitative shake test observations before pilot testing can be very advantageous to the ultimate success of the scaleup and the ultimate process (Figures 4 and 5).

Figure 4.  These LLE internals will be installed into a commercial column

Figure 4. These LLE internals will be installed into a commercial column

Figure 5.  Modular LLE units can be installed at user sites

Figure 5. Modular LLE units can be installed at user sites

At the pilot-plant-testing stage, the production-scale LLE column is designed based on the data generated at pilot-plant scale. Having robust and reliable shake-test data means the pilot testing will be more accurate right from the beginning. Results from the shake test help the engineer to select the best type of column for the process and provide a good starting point for pilot-plant testing, saving time and cost for process development.

The alternative would be to go directly into a full pilot-plant test, but doing so can introduce issues that complicate the project, such as lengthened timelines, and can drive up costs if the ideal type of column is not chosen.

The standardized shake test procedure provides a clear method for generating liquid-liquid equilibrium data. Ultimately, the liquid-liquid equilibrium data and the qualitative observations lead to a better overall solution — the best choice of LLE column for each application.


Column selection

Based on equilibrium data collected in the shake test, as well as the observed interaction between the two phases during mixing and separation, plus previous experience with LLE processes, a column type can be selected. The two main types of agitated extraction columns are rotating and reciprocating. A rotating column operates using impellers on a central shaft, plus baffling or plates (or both) to define the mixing pattern of the liquids and minimize axial mixing. The amount of shear will depend on the impeller type and the agitation speed. A reciprocating column, on the other hand, forms disperse phase droplets utilizing internals that reciprocate (up and down) at specified amplitude and frequency. Because of the uniform shear across the entire cross section of the column, this type of mixing is well suited for systems that emulsify easily.

Here are some general rules of thumb to follow when deciding which column type to use. A rotating column would be the first choice if the shake tests indicate a short separation time (on the order of seconds), and if the mixture of solvent and feed does not form an emulsion. Typically, these are systems with high density differences between the phases or high interfacial tension. Rotating columns can minimize capital costs. In cases where a slow-separating LLE system exists, rotating columns would not be ideal because of the high shear forces imparted by the tips of the impellers.

Reciprocating columns are used in cases where the shake tests indicate slow phase separation and if the system tends to emulsify, which typically happens with low density or low interfacial tension (or both). Reciprocating columns offer higher throughput than rotating columns due to higher open flow area and uniform shear mixing, which produces droplets in a narrow size distribution. The reciprocating column has also been proven to be the better choice when a significant amount of suspended solids is present in the system.

The case in the section below describes a specific example of how a shake test was conducted and how the data generated from it were used to make an informed decision about the scaleup to a commercial-scale LLE process.

A Liquid-Liquid Extraction Example Case:
LLE of Phenol from Phenolic Resin

The following represents an example of an actual procedure used by the authors’ employer to generate LLE equilibrium data for a process that was to be scaled up. The process in question required the removal of phenol from an aqueous stream produced in phenolic resin manufacturing. The aqueous feed contained about 7.5% phenol and the desired effluent concentration was less than 50 parts per million (ppm) phenol. Methyl isobutyl ketone (MIBK) was selected as the solvent. The first step in the process development was to generate liquid-liquid equilibrium data using the shake test procedure discussed in this article. A total of six mix-decant runs were performed at 30˚C in a 2-L flask at a solvent-to-feed ratio (S/F) of 0.2, on a weight basis. The actual procedure utilized for this phase of testing is summarized as follows:

  1. To the flask, charge 1,000 g aqueous feed and 200 g MIBK – Note the solvent to feed ratio (S/F) is 0.2. Maintain this ratio throughout all of the tests.
  2. Set water bath at 31˚C. Agitate at very low speed and circulate hot water through jacket until the system is at 30˚C.
  3. When desired temperature is reached, then agitate at high speed for 2 minutes.
  4. Turn off agitation and allow phases to separate. Note the time for separation. Also make note of any emulsion formation, entrainment or rag layer (buildup of solids) that forms at the interface.
  5. Drain the bottom, aqueous phase precisely to the interface. Measure the total weight of the lower phase and take a sample (Raffinate #1). Drain the upper, organic phase completely, measure total weight and take a sample (Extract #1). Analyze raffinate sample for phenol, methanol and MIBK. Analyze extract sample for phenol, methanol and water. Record results.
  6. Weigh the remaining raffinate phase (feed phase after separation), and recharge to the flask. Add fresh solvent at a weight required to maintain the desired solvent to feed ratio (0.2).
  7. Repeat Steps 2 through 6 for a total of six shake tests. The final raffinate sample should be less than 50 ppm phenol. If the raffinate concentration is greater than 50 ppm, then repeat shake tests until desired result is achieved.

In this particular example, the phenol concentration was progressively reduced for each mix-decant run and required all six runs to arrive at a raffinate concentration below 50 ppm. The separation time for all runs was relatively fast (23–45 seconds). A slight emulsion band occurred at the interface for the first run and all subsequent runs produced a sharp, clear interface. Based upon the results from these tests, the authors’ company determined the optimal type of column to use for this application.

Subsequently, a pilot plant test was performed in a 3-in. diameter, glass-shell of the type determined from the data. The column performed perfectly for the process and the required data were generated for accurate scaleup to a production-size column. A complete extraction system was designed and built for the extraction step, followed by a distillation column for the extract phase and stream stripper for the raffinate phase. The entire system was installed and operated successfully.

Edited by Scott Jenkins


DonGlatz-300x273Donald Glatz is the manager of extraction technology at Koch Modular Process Systems, LLC (45 Eisenhower Drive, Suite 350, Paramus, NJ 07652; Phone: 1-201-267-8570; Email:; Website: His activities include evaluation and optimization of liquid-liquid extraction processes plus scaleup and design of extraction columns and systems. He has been working in this field for the past 25 years and has published a number of papers and articles covering this subject. Glatz is a frequent speaker at AIChE and international solvent extraction conferences. He also presents technical seminars covering liquid-liquid extraction technology. Prior experience includes 13 years in process development for GAF Chemical Corp. and General Foods Corp. Glatz holds a B.S.Ch.E. from Rensselaer Polytechnic Institute and an M.B.A. from Fairleigh Dickenson University.

Brendan CrossBrendan Cross is a senior process engineer for Koch Modular Process Systems, LLC (same address as above; Email: His responsibilities include the detailed design and commissioning of separation processes typically involving distillation, liquid-liquid extraction, and stripping. He has worked in separations process engineering for the past 10 years. Brendan holds a B.S.Ch.E. from Columbia University.


tomlightfootThomas D. Lightfoot, P.E., is the pilot plant manager for Koch Modular Process Systems (same address as above; Email: His responsibilities include supervising all pilot-plant testing performed at the Koch Modular Pilot Plant in Houston, Texas. He has been working in this capacity for the past eight years. Prior to that, he spent 22 years working in various positions in operating facilities within the CPI. Lightfoot holds a B.S.Ch.E. degree from the University of Pittsburgh and is a licensed professional engineer in the state of Pennsylvania.


Related Content
Liquid-Liquid Extraction
Liquid-liquid extraction is a common mass-transfer operation in which a target solute material is transferred from a feed phase into…

Chemical Engineering publishes FREE eletters that bring our original content to our readers in an easily accessible email format about once a week.
Subscribe Now
How can accurate pH measurement prevent power plant corrosion?
Five Reasons Why Chemical Companies Are Switching to Tunable Diode Laser Analyzer Technology
Simplify sensor handling and maintenance with ISM
Three reasons to measure pH in-line
ABB Ability™ technology to transform BASF rotating equipment into intelligent machinery and improve uptime and reliability

View More

Live chat by BoldChat