Introduction
This guidance provides an example of the documentation for validation of a solvent recovery process. Solvent recovery validation is needed in some situations such as where the recovered solvent is intended for general site wide reuse into suitable manufacturing processes, including other API manufacturing processes than those from which the used solvent originated.
This guidance provides an example of the contents that may be found in solvent recovery validation documentation, including appropriate acceptance criteria for solvent recovery validation. The processing addressed by this type of validation includes the recovery of used solvent to provide solvent that is acceptable for use.
This guidance applies only to solvent recovery validation and is not intended to describe validation practices for API or drug product manufacturing.
Recommendations & Rationale
Information on solvent recovery for API manufacturing may be found in another guidance that describes solvent recovery practices and includes considerations for setting impurity limits for recovered solvent. Internal impurity limits for used solvent intended for recovery are established based on knowledge of the capability of the intended recovery process and the intended use of the recovered solvent.
In general, to begin the work process for validation of a solvent recovery process, a documented description of the recovery process should be provided. This should include explanation of the controls that ensure that the recovery process can reliably provide acceptable quality recovered solvent that is suitable for its intended uses. Justification should be provided for processing parameters that are critical for control of recovered solvent quality. A validation protocol with acceptance criteria for demonstrating consistent control of the recovery process then provides the plan for evaluating validation batches, much like protocols for validation of manufacturing processes.
Process consistency:
If the potential for significant variability in the feedstock of used solvent exists, it is recommended that validation batches be prepared over a significant period of time to encompass the likeliness for variability in the used solvent feedstock. A good practice that has been used is to include validation batches made with an intervening time interval between each, as shown in the example provided later in this guidance (see acceptance criteria section). The increased risk introduced by potential variability in feedstock is therefore addressed by evaluating more validation batches over a longer time period.
Concurrent validation;
Concurrent validation is typically appropriate for solvent recovery validation. Validation may typically be done over an extended period of time rather than necessarily including consecutive batches, and releasing individual solvent batches that meet validation acceptance criteria pose no appreciable risk to customers of our API and drug products (i.e. solvent is a raw material rather than a final product).
Process controls:
A computer simulation is sometimes used to predict volumes where fractional cuts in the distillate should be made. The simulation typically uses analytical data from analysis of the used solvent feed batch and known analytical requirements established for the start of each cut. When a simulation is used, it is recommended that the effectiveness of the computer simulation to predict process control points be shown.
Alternatively, in-process gas chromatograph (GC) solvent, Karl Fisher (KF) water, or other analytical methods for evaluation of distillate samples may be used to determine cut points, if a computer simulation is unavailable. Either in-line or off-line testing may be used. It may be helpful to perform a lab-scale feasibility run to predict where to make distillate cuts. Where a continuous recovery process is used, other operating parameters than cut points (e.g., feed rates, reflux rate, and/or column temperature profile) may be important process controls that should be defined to insure control of the quality of recovered solvent.
Validation of any process is dependent on the particular systems used for the process, and validation of solvent recovery operations is not different in this regard. Controls and decision criteria may need modification if the distillation is run with a different column, for instance. When validation of the solvent recovery process is needed, an impact assessment of the systems used for a solvent recovery process should be performed to determine if they could directly impact the quality of an API.
Acceptability:
Validation batches should conform to release specifications established for the recovered solvent. These specifications should include tests for volatile and non-volatile impurities. A concluding validation report should evaluate how each of the acceptance criteria was met.
Validation of the use of recovered solvent should be done as part of validating the process in which it is used, if this is considered necessary.
Example: THF recovery process
The remainder of this guidaqnce provides one example of validation of a batch solvent recovery process that of the recovery of tetrahydrofuran (THF) from a used THF/toluene mixed solvent . The used THF may be derived from multiple process streams and the recovered THF will be available for mult-process use. While this example includes the use of computer simulation to predict when to make cuts for collection of distillate fractions, validation of the process may also be achieved with appropriate in-process monitoring.
Control limits for impurities permitted in the used THF stream and for the in-process stream acceptable for final distillation are established at different points in the solvent recovery process based on where impurity controls are established. See Appendix I for a diagram of this THF recovery process and Appendix II for impurity control limits for used THF to be recovered.
Volatile impurities of used THF of greatest concern for this recovery process include alcoholic impurities, water, and toluene. Key steps in the THF purification scheme address these impurities. Non-volatile impurities are of a lesser concern to the design of the purification scheme because they are expected to separate from THF during the distillation.
Validation of this THF recovery process is focused on control of the distillation step. Other processing steps may be performed to remove some volatile impurities (such as water and alcohols) to provide THF suitable for use in the distillation, but these steps are not included in this validation example.
Process Overview and Description of a THF recovery process
The solvent recovery scheme for THF from THF/toluene mixtures is shown in Appendix I. Preliminary steps provide controls for particular volatile components (alcohols) and water. A batch distillation process is the final purification of THF from the used stream. The distillation column for this process is set on total reflux with the initial steam setting at about 1600 kg/hr for a minimum of 30 minutes to stabilize the column. Four slop cuts (preliminary fractions richer in more volatile impurities) are collected at a reflux ratio of 16:1 until reaching a specified volume determined by the computer simulation (or by in-process testing). The first slop cut is composed of THF, methylene chloride, and other solvents.
The second and third slop cuts are composed primarily of THF but also containing other solvents. The first three slop cuts may be sent to appropriate solvent recovery streams or sent to waste for disposal. The fourth slop cut, which is primarily THF but may not be of high enough purity to include in the product cut, may be returned to the used solvent collection bank. After the four slop cuts are completed, an additional cut (also returned to the collection system) is collected at about an 8:1 reflux ratio until a volume determined by the computer simulation (or by in-process testing) is accumulated. When the computer simulation is not available, collection of distillate for the THF product cuts is begun when in-process testing (such as GC) result shows that the distillate is providing acceptable quality recovered THF that meets the specifications (e.g., purity and volatile impurities).
The first THF product cut is collected in a suitable storage tank at about a 1:1 reflux ratio until a volume determined by the computer simulation (or by in-process testing) is accumulated. A second THF product cut is collected at about a 3:1 reflux ratio until the middle column temperature reaches 75 ºC, or until reflux is lost. After this, the still pot bottoms (composed of THF and toluene) are cooled and set aside, to be pooled with other used solvent for another recovery batch. Solvent in the column drain is charged to the next recovery batch, or if no further recovery batches will be run, it is sent to the used solvent collection bank.
The combined recovered solvent from the first and second product cuts is assayed to verify that it meets established specifications (see “Process Materials” section). When all assays are complete and it is shown that specifications are met, product in the storage tank is transferred to the released THF storage bank for future use in chemical processing steps.
Equipment requirements:
The following equipment is used for the production-scale THF recovery process:
- Used THF/toluene holding tank
- Still pot
- Distillation column, condenser, reflux drum
- Day-use storage tank
- THF holding tank
- Blow case
Process Materials:
Requirements for pooled solvent to be recovered for validation of this process should be provided or referenced in the solvent recovery validation protocol. As an example, specifications for the THF/toluene feed for the THF recovery process are summarized in the table included in Appendix II.
Critical Process Controls:
In preparation for the final distillation, information on the composition of each batch of used solvent to be distilled is needed. For THF recovery, the computer simulation does not account for isopropyl alcohol (IPA) in the stream, so it is critical that the amount of IPA is NMT 0.1 vol-%. Feed composition that is higher in IPA will negatively impact the quality of recovered THF by being unacceptably high in the total alcohols (C1-C4) specification. Used THF/toluene streams with higher amounts of IPA should be directed through the optional processing step for removal of alcohols (see Scheme I in Appendix I).
Slop cuts are described earlier in the process overview and are important to the success of the process but some tolerance for variation in where these cuts are started and ended will not be harmful to the quality of the recovered THF. However, the product cut start can negatively impact quality if begun too early by permitting capture of off-specification material. A late product start cut will not affect quality, only resulting in loss of yield. When the product cut is started is therefore a critical process control.
A Data Control System (DCS) or appropriate testing of samples may be used to control the various reflux ratios based on the recipe parameter setting by modulating the control valve as needed to achieve the specified reflux ratio. With or without DCS, it is important not to run with too low a reflux ratio (less than 2:1) for an extended period (i.e. running the distillation too fast), which could risk adding low-specification material to the product cut. Short-term deviations will only have a minor impact on quality. There is no adverse quality impact from running the reflux ratio higher than the prescribed 3:1. A high reflux ratio results in a slower distillation and slow product recovery, but this has no detrimental quality impact on the recovered solvent.
The end of the second product cut should occur when a column temperature of 75 ºC is reached. Ending the product cut earlier will result in loss of yield, but no detrimental quality impact. If the temperature exceeds 76 ºC and the product cut has not been completed, quality of the THF product cut could be diminished. This temperature is regarded as a critical process control.
To summarize, two critical process controls have been identified for the THF-COL5 recovery process:
| Process Control and Acceptable Range | Why Chosen |
| Start of first product cut: Slop cut volumes NLT simulation values or GC of distillate NLT 99.3 vol-% THF, NMT 0.25 vol-% ethanol, and NMT 0.03 vol-% water | If started too early, low specification material can be added to the product. |
| End of second product cut: Middle column temperature NMT 76 ºC | If allowed to continue further, low-specification material can be added to the product |
Validation Plan and Acceptance Criteria
To validate this solvent recovery process, THF will be recovered from three sets of at least three consecutive batches of used THF. The time period between the beginning of the last batch in a set and the first batch of the next set will be at least one month. The total number of batches will therefore be no less than nine.
Acceptance criteria:
No significant deviations from the master instructions for the THF recovery process are acceptable.
Critical process parameters must be controlled within the ranges specified in the table above.
The suitability of each feed batch for this recovery process will be examined. For each batch used for the validation, the computer simulation (or in-process control, where computer simulation is not used) must predict that acceptable THF can be produced.
Acceptability of the recovered THF will be shown by evaluating the results of the release testing on the representative sample for each batch. Each batch must meet these release specifications for THF:
| GC Tetrahydrofuran | NLT 99.0 area-% |
| Moisture (by KF) | NMT 0.03 vol-% |
| Non-volatile components | NMT 0.0035 % |
| APHA (color) | NMT 20 |
| Peroxides | NMT 150 ppm |
| Impurities (by GC): | |
| Methylene chloride | NMT 0.2 area-% |
| Alcohols (C1-C4) | NMT 0.15 area-% |
| Acetone | NMT 0.05 area-% |
| Total ketones | NMT 0.1 area-% |
| Other volatile impurities: | |
| Individual unidentified | NMT 0.2 area-% |
| Total unidentified | NMT 0.2 area-% |
Consistency of the recovery process will be shown by the successful completion of no less than three consecutive batches in each of the sets of recovery runs (or not less than nine batches total). All validation batches must meet the requirement described above.