Recovered Solvents and Mother Liquors: Recycling Rules

Every crystallisation is a sorting operation. The crystal lattice admits the target molecule and excludes most of what else is present, which is why crystallisation purifies. The corollary receives less attention: everything excluded is now in the mother liquor, at a concentration relative to the crop that the yield determines.

What a site does with that liquor, and with the solvent it recovers from it, is one of the strongest predictors of batch-to-batch behaviour. It is also invisible on any certificate.

The loop and what it accumulates

Solvent recovery is standard practice and good economics. The mother liquor is distilled, the solvent passes a specification and returns to the process, the residue leaves as waste. The engineering question is what the loop accumulates between distillations: water, high-boiling by-products, colour bodies, and any volatile impurity that co-distils with the solvent.

Each recycle enriches the loop. The impurity concentration rises batch by batch until it reaches a steady state, where the amount entering the loop each cycle equals the amount leaving through the purge and the product. Where the site runs the loop without a purge, the steady state is reached at a higher concentration, and the process's impurity profile quietly re-balances around it.

Control	What it prevents
Recycled solvent tested against the fresh specification	Enrichment that goes unnoticed
A defined purge fraction per cycle	Steady state at an uncontrolled level
Periodic identity check of the recycled stream	Cross-contamination from a previous campaign
Impurity profile generated with the loop running	A file that describes a process the plant does not run

Why the profile should be generated on the real process

An impurity profile measured on a campaign running fresh solvent describes a process that exists for one batch. Once the loop closes, the late-eluting and high-boiling impurities rise, because those are the ones the distillation leaves behind. A profile generated with the loop running is the honest one, and it is the one the customer's file should contain.

The route comparison matters here as well: a chemical route with chlorinated solvents has a different loop chemistry from an enzymatic route with alcohols and water, and the accumulation behaviour differs accordingly. The same logic applies to the residual solvent limits the ICH Q3C framework sets: a loop can concentrate a class 2 solvent to the point where the product's residual solvent result starts to drift upward, and the drift is a process property, not a batch anomaly.

Legitimate versus silent

Recovery is legitimate when three things are true. The recycled stream is held to the same specification as fresh. The accumulation is monitored, with a purge that actually runs. And the process description says the solvent is recovered, so that a customer's file is not describing virgin solvent to a plant running a closed loop.

The same three questions, asked in reverse, identify the silent version: no specification for the recycled stream, no purge record, and a process description that omits the loop. None of the three is proof of a problem. Together they mean nobody is watching the variable that most influences the impurity profile.

The European expectations for recovery are explicit in the GMP text: recovery requires documented justification, the recycled material must meet specifications, and the loop needs control. The EDQM general chapters define the solvent specifications themselves.

Second crops and the same logic

The mother liquor's other product is the second crop: the material that crystallises from the liquor as it cools further or is concentrated. It is legitimate yield, and it is predictably enriched in the excluded impurities. A site that combines second crops into the main batch has a blending question, covered separately; one that segregates them by a written rule has a process.

The certificate cannot tell the reader which happened. The batch record can, and the question "does the process anticipate second crops, and under what rule" is answerable in a sentence by a site that has thought about it.

What changes should be notified

Introducing a recycle loop where there was none, changing the purge fraction, or changing the solvent specification for the recycled stream are all process changes with potential effect on the impurity profile. They belong in the notification categories, because the customer's file describes a process and the loop is part of it.

A buyer evaluating a new supplier can compress the whole topic into one question: is solvent recovered on the step that makes this material, and is the recycled stream tested against the fresh specification? The answer, and the ease with which it comes, describes the site's process control more efficiently than any questionnaire.

Where this leaves a supply-chain partner

We say plainly whether a step runs a solvent recovery loop, hold the recycled stream to the fresh specification, and generate impurity profiles with the loop in its normal state rather than on fresh solvent. Where a customer's file was written around a fresh-solvent process, we flag the difference rather than letting the certificate imply it.

The economics of recovery are real and the control of it is cheap. The sites that do both are easy to identify: they can state their purge fraction without looking it up.

Reading the loop from the outside

A buyer cannot see the loop, but its effects are visible from three public surfaces. The residual solvent line, where a class 2 solvent that should sit at a fixed low level instead drifts upward across batches. The impurity profile, where the late-eluting peaks grow a little every month until a purge resets them. And the batch-to-batch spread, which is wider on a loop with no purge than on one with a controlled one, because the loop is a memory that every batch inherits.

The consistency question and the loop question are therefore the same question asked at different altitudes. A site with a tight batch history either has no loop, or has one that is purged and monitored. A site whose results wander has a process with a state that carries over, and the loop is the most common carrier.

The second crop sits in the same family. A site that recovers second crops under a written rule produces a batch record that says so; one that recovers them when yield is short produces a process whose inputs vary with the quarter. The certificate is identical in both cases, which is why the question belongs in the supplier conversation rather than in the document review.

What changes deserve notification

Three loop-related events change what the customer's file describes, and each belongs in the notification categories. Introducing a recovery loop where none existed. Changing the purge fraction or the frequency of testing the recycled stream. And changing the solvent specification itself, including the acceptance criteria the recycled stream must meet.

The last one is the quietest and the most consequential, because the recycled solvent's specification is effectively a second specification for every product the loop touches. A loosened solvent limit propagates into the residual solvent results of every subsequent batch, and the propagation is invisible unless somebody compares the two documents.

The regulatory anchors are public and consistent: the ICH Q3C framework for solvent limits, the recovery expectations in the European GMP text, and the solvent specifications themselves from the EDQM general chapters.

One question compresses the topic: what is the purge fraction of the solvent loop on this step, and how often is the recycled stream tested against the fresh specification? A supplier with two numbers has a loop under control. A supplier with a description of their commitment to sustainability has answered a different question.

Water in the loop

A solvent recovery loop is also a water balance. Reaction and work-up steps generate water, extraction steps remove some of it, and distillation carries the rest into the recovered solvent as an azeotrope or as dissolved moisture. A loop with no water control reaches a steady state at whatever the incoming water and the purge allow, and that steady state sets the water content of every batch that uses the recovered solvent.

For a process where water shifts an equilibrium or hydrolyses an intermediate, this is a process variable hiding inside a utility. The symptom is a gradual change in yield or in a specific impurity, correlated with nothing obvious, and the cause is a solvent that has been carrying slightly more water for six months.

The control is a specification on the recovered solvent that includes water, tested on a defined frequency, plus a purge that is actually executed. Both belong to the solvent's own specification rather than to the product's, which is why they are easy to leave out of a customer conversation and easy to find in an audit.

The question that opens the subject

What is the water specification on your recovered solvent, and when was it last tested? Two numbers, one date. A site with the answers is monitoring a variable it knows about. A site without them is running a loop whose composition is an outcome rather than a decision.

How the loop is monitored analytically

The recovered solvent needs its own test set, and it is shorter than a product's: identity by the appropriate technique, assay or purity against the fresh specification, water, and a residue-on-evaporation figure that catches high-boiling carryover. Gas chromatography covers most of it in one run, and the residue test catches what the chromatography does not elute.

The frequency matters more than the test. A loop tested once a year is a loop that reaches steady state unobserved; one tested per campaign or per defined number of cycles has a trend. The trend is the control, because steady state is the target and the question is whether the loop is sitting where it was designed to sit.

Frequently asked questions

Is recovered solvent acceptable for pharmaceutical use?

Yes, where the recovery process is validated, the recycled solvent meets the same specification as fresh solvent, and the accumulated impurities are controlled. The European GMP text treats recovery as a normal operation with documentation duties, not as an exception.

Why is the mother liquor the interesting material?

Because crystallisation works by excluding impurities from the crystal lattice. Whatever did not fit is in the liquor, concentrated relative to the crop. Handling that liquor determines whether impurities leave the process or recirculate through it.

What accumulates in a solvent recycling loop?

Everything less volatile than the solvent or formed during the process: high-boiling by-products, water, colour bodies and any solvent carried over from an earlier step. Each pass enriches the loop until the impurity reaches steady state or is purged.

How is the accumulation controlled?

By a purge and by monitoring: a defined fraction of the loop leaves the process each cycle, and the recycled solvent is tested against the fresh solvent specification at a defined frequency. Steady state, not zero impurity, is the engineering target.

Does using recovered solvent change the impurity profile?

It can. The profile of a process running a recycle loop differs from one on fresh solvent, particularly for the late-eluting or high-boiling impurities. That is why the profile should be generated on the process as it actually runs.

What should a buyer ask?

Whether solvent is recovered on the relevant step, whether the recycled stream is tested against the fresh specification, and whether the impurity profile in the file was generated with the loop running. Three questions, all factual.