GMP hash lab: zones, flows, and batch release
HomeFacilityGMP hash lab: zones, flows, and batch release
Facility · GMP

GMP hash lab: zones, flows, and batch release

This paper gives the GMP requirements for a hash facility: the cleanroom grade zones, the flows of product and personnel, and the batch release. A batch must go through seven groups of tests and three release gates to get a signed Certificate of Analysis.

Facility13 diagrams9 sources~18 min to read
Start here

Purpose and scope

Good Manufacturing Practice (GMP) is a system with written records and audits. The system shows that a product agrees with its label, and that no dangerous material went into the product during production. A GMP hash lab makes purified resin concentrates (bubble hash, rosin, live resin, distillate) from cannabis biomass, with full control of contamination.

Extraction can increase the concentration of a contaminant, decrease it, or change where it is. The effect is different for each analyte, process, and mass yield[9]. A flower result that is in the limits does not show that the finished concentrate agrees with the requirements. Do a test of the incoming material and of the finished batch. Then use the two sets of results to calculate the transfer for each process. Material goes to the next step only when the release data show that it is safe and that its label is correct.

Two sets of regulations give most of the requirements for the facility. EU-GMP Annex 1 gives the cleanroom classifications and the contamination control strategy. The U.S. cGMP regulations in 21 CFR 210/211 give the production controls, the records, and the authority of the quality unit to release product[1]. This paper is a reference model. It is not legal advice. The limits and grades are different in each jurisdiction.

The one basic requirement of GMP

‘If there is no record, the task did not occur.’ GMP is a system with documents and validation, and an inspector can examine it. An auditor does not accept a very clean room that has no records. An auditor accepts a basic room that has full, signed records.

The four primary numbers for this designPrimary values: 5 grades, 7 mandatory test groups, 100% traceability of batches, zero open critical deviations.02855821105Cleanliness grades7Release test groups100Genealogy coverage0Critical deviations at release
Figure 1. Four numbers are the primary values for the facility. There are five cleanliness grades. There are seven mandatory groups of release tests. The genealogy coverage is full. At the time of release, the number of open critical deviations must be zero.
The standards and the part of the building that each standard controls.
StandardContentArea of the facility that it controls
EU-GMP Annex 1Classification of cleanrooms and the contamination control strategyEach room that has a cleanroom class
cGMP 21 CFR 210/211Production controls, records, and the authority of the QC unit to release productAll the plant, and QA
ICH Q7 / Q9 / Q10Quality system, risk management, and lifecycleQuality management system
ISO 14644-1Particle-count classes of cleanrooms (ISO 5/7/8)Classification of air
GACPGood Agricultural and Collection Practice for biomassGoods-in and intake
NFPA 30 / C1D1Flammable liquids, and classified electrical areasSolvent extraction room
Terms

Definitions

These terms occur many times in this paper. The paper gives the full information about each term where the term first occurs.

Cleanroom gradeA symbol (CNC, D, C, B, A) that shows how clean the air of a room is. Each symbol agrees with an ISO 14644 particle class. Grade C is ISO 7, and Grade A is ISO 5.
Batch / lotOne specified production run with one lot ID. Personnel record each event of the batch with that ID. The record is the genealogy of the batch.
DeviationEach difference from the approved process. Personnel record each deviation and do an investigation. The deviation must have the status Closed before the batch with the deviation can have the status Released.
CAPACorrective And Preventive Action. It is the loop that corrects the problem (corrective) and makes sure that the problem does not occur again (preventive).
Quarantine / Released / RejectedThe three statuses of a material. Each material has only one of these three statuses at one time. No material is between two statuses. The status of a material is always one of these three.
CCP (Critical Control Point)A step in which a measured limit prevents a hazard (HACCP term). An example is wash water at less than 4 °C.
ALCOA+The properties that a good record must have: Attributable, Legible, Contemporaneous, Original, Accurate, plus Complete, Consistent, Enduring, Available.
Water activity (Aw)The free water that microbes can use, on a scale of 0 to 1. A low value prevents the growth of mold and bacteria. CoA is Certificate of Analysis. C1D1 is a classified room for flammable material.

QA is the Quality Assurance function. QA makes the last release decision, and QA operates independently of production[1]. This difference between QA and production is the primary part of the system. The personnel who send product out of the facility must not give the approval for their work.

Statuses of a batch1Quarantinereceived, withno decision→2Releasedaccepted, forsale→3Rejectednot accepted, todispositionEach lot has only one of these three statuses at all times.
Figure 2. The three statuses. A material has only one status at a time. The location of the material and its status in the system must always agree.
Is this step a Critical Control Point?an easy HACCP decisionIs there a hazardhere?Can a subsequentstep control it?CCP: must control andmonitor hereNot a CCP, control itin a subsequent stepyesnoyes
Diagram. The HACCP test to find if a step is a Critical Control Point. You must monitor each Critical Control Point.
Water activity (aw): the mold limitaw is the free water that microbes can use. Cure to approximately 0.55-0.65: stable, good to use, below the mold limit.dry / stablecure targetriskmold growth0.30.60.9
Diagram. Water activity gives the limit for safety from microbes. It also gives the limit for a stable product.
Quality of the information

Evidence and limitations

Three types of information

We want each paper to be correct. Thus we examine each paper for information that is not solid. This information comes from the view of one grower, from a small number of sources, or from grower methods that have no controlled test. We identify this information, and we do not show it as solid.

Frequently there is no paper for the decision that you must make. Then we use the reports of other growers and the methods that work in the rooms that we know. This information can help you, but it is not the result of a laboratory test. Use the methods that work for your plants, for your room and for your meters. If a table does not agree with your crop, use the data from your crop and record the difference.

Solid
Tests of plants, specifications or many sources that agree show that the information is correct
  • A concentrate can have a higher concentration of residues than the input biomass. The difference changes with the extraction yield and the partitioning.
  • Systems that you can examine, with records, CAPA and genealogy, are the primary part of GMP.
Grower method
Many growers use this method in their rooms. Start with it, then change it for your room.
  • Procedures for room flow, for gowning and for batch records, in facilities that a regulator controls.
Weak
The view of one grower, a small number of sources, one test, or a method that works in one room
  • ICH values for residual solvents as the limits of your cannabis license. The limits are different in each jurisdiction.
  • EU Annex 1 Grade A or B for all hash. The grade changes with the product and with the license.
  • A list of seven mandatory groups of tests that ignores the panel of your regulator.

If you find an error, make a report. We will correct it. Make the report on GitHub, with the name of the paper and the information that is incorrect. If you have a source, include it: Make a report of an accuracy problem. The regulations, labels and licenses of your area always override each procedure in these papers. The label weak in a paper shows the points where the risk of an error is highest.

Basic information

Zones and cleanroom grades

The building is a set of shells of cleanliness. Each shell is in the next shell. The dirty tasks are intake, milling, and waste. They are at the outer edge. The clean tasks, with the highest cleanliness, are the collection, the fill, and the packaging of open product. They are in the inner zone.

The flow requirement is easy, and you must always obey it: Product moves to the zones with more cleanliness, to the inner zone. Personnel and waste move to the dirty edge. Do not let the two flows go through the same area without control. Between each pair of shells there is an airlock, and the grade changes by one step.

In EU-GMP Annex 1, Grade C agrees with ISO 7 and Grade A agrees with ISO 5[8]. Most hash facilities for recreational cannabis and for medicinal cannabis use grades D to C. They use Grade C for the fill point, with local protection. A full Grade A or B is necessary only for sterile dose forms or pharmaceutical dose forms.

Shells of cleanliness (outer = dirty, inner = clean)Product flows to the clean inner zone. Personnel and waste flow to the dirty edge.CNC intakeGrade DGrade CInner Grade A024
Figure 3. Each zone is in the next zone, from the CNC outer shell through Grade D and Grade C to the clean inner zone. Product goes to the clean inner zone. Personnel and waste go to the dirty edge.
Air changes each hour increase with cleanlinessHigher grades change the air more frequently. Recommended rate for ISO 7: ~20 to 40 ACH.0681352022705 ACHCNC15 ACHGrade D30 ACHGrade C240 ACHGrade A
Figure 4. The air-change rate increases with the grade. ISO 7 rooms (Grade C) have approximately 20 to 40 changes each hour[8]. A Grade A fill point with unidirectional airflow has a much higher rate.
Each room with its grade, ISO class, air-change rate, and primary task.
RoomGradeISO classAir changes each hour (ACH)Task
Goods-in and quarantineCNC, 4 to 6Personnel receive and hold the biomass.
Milling and dispensingGrade DISO 810 to 20Personnel make the biomass smaller and weigh it.
Wash and extractionGrade CISO 720 to 40Personnel remove the trichomes from the plant material.
Freeze-dry and pressGrade CISO 720 to 40Personnel dry the material and press it to make rosin.
Solvent recoveryGrade D (C1D1)ISO 810 to 20Personnel recover the solvent and do an LEL purge.
Open-product fillGrade AISO 5unidirectionalPersonnel fill open product (pharmaceutical use).
Cleanroom grades: more clean near the productthe number of particles in the air is lower near open productGrade DGrade CGrade BGrade AEach inner zone has a smaller number of airborne particles. Grade A (or ISO 5) is for open product. The outer grades are a buffer.
Diagram. Each cleanroom grade is in the grade before it. The air is more clean nearer to open product.
Basic information

Pressure cascade, HVAC, and gowning

The primary medium for the movement of contamination in a cleanroom is air. Air flows from a high pressure to a low pressure, and not in the opposite direction. A positive-pressure cascade uses this effect.

The clean rooms have a higher pressure than the adjacent dirtier spaces. Thus air always flows out from clean to dirty. When you open a door, clean air flows out. Dirty air cannot flow back to the product.

The pressure increases in steps, from shell to shell. EU-GMP Annex 1 recommends that the difference between adjacent classified zones is approximately 10 to 15 Pa[8]. This difference gives these values: 0 Pa (CNC), +15 (D), +30 (C), +45 (B), and +60 (A). The solvent room is different: its pressure is negative, approximately -15 Pa. As a result, the flammable vapor stays in the room and flows to the LEL exhaust. It does not flow into the building.

Pressure cascade: air flows from clean rooms to dirty rooms1CNC 0 Paouter shell→2Grade D +15milling→3Grade C +30wash / dry→4Grade A +60fill zone→5Solvent -15negative:contains vaporPositive pressure keeps product clean. The solvent room is negative to contain flammable vapor.
Figure 5. Positive pressure that increases from CNC to the Grade A inner zone keeps the airflow from clean to dirty. The C1D1 solvent room has a negative pressure to contain vapor.

Personnel are the largest source of particles and microbes in a cleanroom[5]. Thus entry is a one-way sequence of airlocks, and the gowning increases when the grade increases.

Filtration also changes with the grade. The milling and dispensing rooms have F9 pre-filters. The wash and dry rooms have H13 HEPA filters. The fill point has H14 HEPA filters. An H14 HEPA filter stops a minimum of 99.995% of the particles at the most penetrating particle size[4]. Thus this filter is at the point with the highest cleanliness.

Gowning airlock cascade (entry is one-way)1CNC entrystreet clothesoff→2D airlocklab coat and oneglove→3C airlockcoverall and twogloves→4B/A airlocksterileone-piece andtwo gloves→5Cleanroomwork, removegown at theother exitGowning increases with the grade. Personnel remove the gown in a different corridor at exit.
Figure 6. A one-way sequence of gowning from CNC entry, through the airlocks of each grade, to the work zone. A different exit, with a dashed line, is for the removal of the gown. Thus entry and exit do not use the same door.
Health check at the door

Do not let a person with an open wound, a respiratory illness, or gastrointestinal symptoms go into the cleanroom. Do a health check at the entry, and record each person that you stop. Each airlock has an interlock. Thus the two doors cannot open at the same time.

The HVAC control values for each grade: temperature, humidity, filtration, pressure, and air changes.
GradeTemperatureRelative humidity (RH)FiltrationPressure difference (ΔP)Air changes each hour (ACH)
CNCambientless than 70%F70 Pa4 to 6
Grade D18 to 24 °C45 to 60%F9+15 Pa10 to 20
Grade C18 to 22 °C45 to 55%H13 HEPA+30 Pa20 to 40
Grade A fill18 to 22 °C45 to 55%H14 HEPA+60 Paunidirectional
Solvent (C1D1)18 to 24 °Cless than 55%F9-15 Pa10 to 20
Basic information

Hash manufacturing process flow

Two processes start after the weigh-in. The solventless process removes the trichome heads (the resin glands) mechanically. The solvent extraction process dissolves the resin and then recovers the resin. The two processes make the same type of product, but their hazards are very different.

In the solventless process, personnel agitate fresh-frozen biomass in ice water. They sieve the resin through a stack of screens (220 to 25 micron). They freeze-dry the resin and press it to make rosin. The process has four critical control points: wash temperature (≤4 °C (39 °F)), water quality (RO, <10 CFU/mL), water activity (Aw ≤0.55), and press temperature (≤90 °C (194 °F)). A water activity of approximately 0.55 to 0.65 or less prevents the growth of microbes and fungi[3].

Solventless process with critical control points1Fresh-frozenweigh-in biomass→2Agitate ≤4°CCCP: temperature→3Sieve 220-25umRO water <10 CFU→4Freeze-dryCCP: Aw ≤0.55→5Press ≤90°CCCP: temperatureFour CCPs control the process: wash temperature, water quality, water activity, press temperature.
Figure 7. The solventless process: personnel agitate fresh-frozen biomass, sieve the resin, freeze-dry it, and press it. The figure shows the four critical control points.

Solvent extraction is a closed loop that uses butane, propane, ethanol, or CO₂. In this process, the primary hazards are flammability and residual solvent. Residual solvent is the trace of extraction solvent that stays in the product. The action limits for residual butane and propane are approximately 2000 to 5000 ppm, and the limit is different in each jurisdiction[2]. Ethanol is an ICH Q3C Class 3 solvent, and the usual limit for ethanol is approximately 5000 ppm[2]. Each batch must go through a headspace GC-MS test for residual solvent before the release of the batch.

Closed-loop solvent extraction with QC gate1Fill columnadd biomass→2Solvent flowdissolve resin→3Filter / winterizeremove fats→4Recover solventC1D1 + LEL purge→5Vacuum purgeGC-MS gatebefore releaseC1D1 safety interlocks operate at the same time. The residual solvent test is the release gate.
Figure 8. Personnel fill the column, apply the solvent, filter and winterize, recover the solvent, and do a vacuum purge. The QC gate for residual solvent is the last step. The C1D1 safety interlocks operate at the same time.
A hydrocarbon room must be a C1D1 room

Do butane and propane extraction only in a C1D1 room (NFPA classification). The room must have LEL (lower explosive limit) gas sensors with an automatic purge. It must have explosion-proof electrical equipment, a two-person rule, and pressure vessels with an ASME rating. This area is the most dangerous area in the building. One source of ignition can cause very large damage.

In the solventless process, water is an ingredient and not a utility. The water uses a loop that is only for water. The water goes from the mains to a carbon and sediment pre-filter.

Then it goes to RO/DI. Then it goes to a UV unit and a 0.2-micron last filter. Then it goes to a sanitized ice hopper, and then to the point of use. Personnel get samples at three points. Water that is out of specification goes to the quarantine area immediately.

Water treatment and ice loop (water is an ingredient)1Mains inS1 sample: feed→2Pre-filtercarbon andsediment→3RO / DIS2 sample: afterRO→4UV + 0.2umlast filter→5Ice hopperS3 sample: pointof useThree sampling points. A result out of specification puts the water in quarantine before it touches product.
Figure 9. The water treatment steps with their three sampling points and the quarantine procedure for water out of specification.
Procedure

Testing, sampling, and batch release

Personnel do tests at three stages: the incoming biomass, the process, and the release. The tests are at four sampling stations along the value stream. Personnel keep retained reference samples until 1 year after expiry. If a complaint occurs after the release, personnel can examine the kept material.

Four sampling stations along the value stream1S1 incomingmicrobials,pesticides→2S2 in-processmoisture, visual→3S3 bulk concentratepotency,residual solvent→4S4 finished goodsfull releasepanelEach station has the tests for its stage. S4 is the full legal release panel.
Figure 10. The stations go from incoming biomass (S1), through in-process (S2) and bulk concentrate (S3), to finished goods (S4). The figure shows the tests of each station.

The release panel for finished goods is the legal gate to the market. The panel has seven groups of tests: potency, residual solvents, pesticides, microbials, heavy metals, mycotoxins, and water activity and moisture[3]. Personnel measure the heavy metals (lead, cadmium, arsenic, mercury) with ICP-MS, the standard method[3]. The mycotoxins in the regulations are aflatoxins B1/B2/G1/G2 and ochratoxin A. They are carcinogens, and their limits are in parts per billion.

Test the finished concentrate

Extraction changes the distribution of the contaminants[9]. Thus a raw flower result in the limits does not make sure that the finished concentrate is in the limits. Make the release decision on the finished concentrate. Use the input results and the output results of the same batch to set the carryover factors for each process.

The seven mandatory groups of release tests. Pathogens such as Salmonella and E. coli must not be in the product.
GroupAnalytesMethodUsed for
PotencyTHC, CBD, total cannabinoidsHPLC-DADLabel accuracy
Residual solventsButane, propane, ethanolHeadspace GC-MSSolvent safety
PesticidesState pesticide listLC-MS/MS, GC-MS/MSChemical safety
MicrobialsTYMC, TAMC, E. coli, Salmonella, AspergillusPlate or qPCRPathogen control
Heavy metalsPb, Cd, As, HgICP-MSLimits for toxic metals
MycotoxinsAflatoxins, ochratoxin ALC-MS/MSCarcinogen control
Water activityAw, moistureAw meter or KFMold prevention

Then QA, and not production, operates three gates one after the other. If the result of a gate is ‘no’, the batch goes to remediation. Only a batch that all three gates accept gets the release. The release has a QP/QA signature and a Certificate of Analysis that QA gives[1].

  1. 1
    Gate 1: Records
    Make sure that the batch record is full and signed from start to end (ALCOA+). If it is not, the batch cannot go to the next gate.
  2. 2
    Gate 2: Results
    Make sure that the results of all seven groups of tests are in the specification, with no pathogens. Each result out of specification (OOS) goes to an investigation.
  3. 3
    Gate 3: Deviations
    Make sure that all deviations of the batch have the status Closed, with a CAPA. A critical deviation with the status Open stops the release.
  4. 4
    Release
    When all three gates accept the batch, QA writes its signature and QA gives the CoA. Then the status changes to Released.
Batch release decision tree (QA-owned)1Records signed?yes: next. no:hold→2Results in limits?yes: next. no:OOS→3Deviations closed?yes: next. no:remediation→4QA RELEASEsign, give CoAThree gates in sequence, each with yes or no. A no goes to hold, OOS, or remediation.
Figure 11. The three gates in sequence are records, results, and deviations. A batch goes to the QA release, or to hold, to OOS, or to remediation.
Quality conditions of a batch1Quarantineproduct made orreceived→2Testingin test, noresult→3Releasedsatisfactory→4Rejectedunsatisfactory,isolatedDo not use a batch before its release. Keep each unsatisfactory batch in quarantine. Record it and do not discard it.
Diagram. A batch starts in quarantine. Then the laboratory does a test on the batch. After the test, you release or reject the batch. You make a full record at each step.
Frequent problems

Deviations, CAPA, and troubleshooting

When the process is different from the approved process, the deviation system finds the difference. ICH Q10 gives the pharmaceutical quality system, with CAPA and change control, and this loop is part of it[6]. Each deviation has a CAPA loop of five stages. The deviation does not get the status Closed until the last effectiveness check gives a correct result.

Five-stage CAPA loop1Findrecord thedeviation→2Containquarantine,scope→3Examineroot cause:5-why / fishbone→4Correctrepair theproduct→5Prevention, checkSOP, training,effectivenesscheckRoot-cause tools (5-why, Ishikawa) are for stage 3. The loop closes only when the check is correct.
Figure 12. The loop has five stages: find, contain, examine, correct, and prevention with a check. A feedback arrow closes the loop back to the process[6].

The severity of the deviation sets the time limit and the person who gives the approval. A critical deviation has a risk to the safety of patients or a risk of a recall. It goes to the QA director, and containment must occur in less than 24 hours. A major deviation goes to QA in a specified time. A minor deviation goes to a supervisor, and the supervisor examines the trend. Root-cause tools, for example 5-why and Ishikawa (fishbone) diagrams, are the standard methods to find the cause of the problem[6].

The three frequent errors
  • You use the biomass CoA for the concentrate. You think that the raw flower result shows that the concentrate is safe. You do not do a test of the concentrate. The important result is the result for the concentrate.
  • You release a batch with a result for residual solvent near the limit. Do not release the batch. Do the test again, do the purge again, or reject the batch. Record the decision for the batch.
  • The status of the material and the status in the system are not the same. Keep quarantine stock in a locked cage or on a controlled rack. The system and the shelf must always show the same status.
Waste and reject disposition1Rejected, usedinto locked cage→2Quarantine cagecontrolled,recorded→3Make unusabledenature /destroy→4Licensed disposalmanifest,witnessSolvent waste is a different stream of dangerous material. Cannabis waste becomes unusable, then goes with a manifest and a witness.
Figure 13. Rejected product and used biomass go into a locked quarantine cage. Personnel make them unusable. Then licensed disposal occurs with a manifest with a witness. Solvent waste is a different stream of dangerous material.
The limits

Expected results and limitations

Compliance is a continuous process that personnel measure with data. It is not a task that you do one time. Management review monitors a small number of KPIs.

The target for the right-first-time release rate is a minimum of 98%. The target for the deviation rate for each batch is less than 5%. The target for the median time to close a CAPA is a maximum of 30 days. The target for the results of environmental monitoring (EM) in the limits is a minimum of 95%. The target for the retrieval time in a mock recall is less than 24 hours.

Quality KPI targets for the systemFirst three: high values are the target. Deviation rate: a low value is the target.0285582110target98Right-first-time %95EM in-limit %90CAPA closed <=30d5Deviation rate %
Figure 14. The primary KPI targets with their thresholds. The system is in good condition when the right-first-time rate stays high and the deviation rate stays low.

Equipment must have a qualification before it makes product for release. The qualification follows the validation V-model: DQ (design), IQ (installation), OQ (operation), PQ (performance). Each of these stages makes sure that the stage on the opposite side of the V-model is correct.

The qualification follows the IQ, OQ, PQ sequence[7]. Only then does the process validation start. Usually, it is three batches, one after the other, that agree with the specification. They show that the process gives the same result each time[7].

V-model qualification, then process validation1URS / designfunction list→2Assemble, installIQ: installcheck→3OperateOQ: functioncheck→4MakePQ: check ofoutput→53 correct batchesprocessvalidationEach right-hand stage is the test of the related design stage. Process validation follows PQ.
Figure 15. The design leg goes from the URS at the top to the assembly at the bottom. The qualification leg (IQ, OQ, PQ) goes from the bottom to the top, and it makes sure that the design leg is correct. Process validation of three batches that agree with the specification follows PQ[7].

Cleaning validation also uses a calculated MACO (Maximum Allowable Carryover) limit. Personnel measure the carryover with a swab sample or a rinse sample, with TOC or HPLC, and they use microbial acceptance criteria. ‘Looks clean’ is not an acceptance criterion.

Select the gowning, the monitoring, and the grade for the process that you use. Grades D to C are usual for most hash operations. If Grade A or B is not necessary for the product, a facility with these grades is a waste of capital.

The set review intervals, from each batch to each year.
IntervalItems for reviewOwner
For each batchBatch record, release results, deviationsQA reviewer
Each weekEM trends, open deviations, OOS logQA lead
Each monthCAPA status, KPI dashboardQA manager
Each three monthsManagement review, supplier performanceQA director
Each yearProduct Quality Review (PQR), self-inspectionQuality and operations
Definition of 'compliant'

A compliant facility is not a building without faults. It is a building with evidence. In this building, each batch has traceability. For each limit, the result was in the limit, or the deviation had the status Closed.

An inspector can find all the events of the batch in the records only. The limits and grades are different in each jurisdiction. Before you make the facility, make sure that it agrees with the conditions of your license.

When the system operates, most faults start on the contamination side. Next, read the paper mold risk for that side of the facility.

Related papers

References

  1. U.S. Food and Drug Administration. 21 CFR Part 211, Current Good Manufacturing Practice for Finished Pharmaceuticals (esp. 211.22 Responsibilities of quality control unit; 211.165 Testing and release for distribution; 211.192 Production record review). Code of Federal Regulations, Title 21. (source not from a journal, for example from a manufacturer) https://www.ecfr.gov/current/title-21/chapter-I/subchapter-C/part-211
  2. International Council for Harmonisation. ICH Q3C(R9) Guideline for Residual Solvents (Step 5), reproduced by European Medicines Agency, 2024. EMA/CHMP/ICH/82260/2006. (source not from a journal, for example from a manufacturer) https://www.ema.europa.eu/en/documents/scientific-guideline/ich-q3c-r9-guideline-impurities-guideline-residual-solvents-step-5_en.pdf
  3. Seltenrich N. Cannabis Contaminants: Regulating Solvents, Microbes, and Metals in Legal Weed. Environmental Health Perspectives. 2019;127(8):082001. doi:10.1289/EHP5785. https://ehp.niehs.nih.gov/doi/10.1289/EHP5785
  4. Camfil. EN 1822 and ISO 29463 HEPA filter factory test (EN 1822-1:2019 filter classes; H14 minimum efficiency 99.995% at the Most Penetrating Particle Size, MPPS). (source not from a journal, for example from a manufacturer) https://www.camfil.com/en/insights/standard-and-regulations/en-1822-and-iso-29463-hepa-filter-factory-test
  5. Meng H, Shiue A, Wang C, Leggett G. Particle and bacterial colony emissions from garments and humans in pharmaceutical cleanrooms. Journal of Building Engineering, 2024;96:110...; ScienceDirect S2352710224023970. https://www.sciencedirect.com/science/article/abs/pii/S2352710224023970
  6. Enhancing Pharmaceutical Product Quality With a Comprehensive Corrective and Preventive Actions (CAPA) Framework: From Reactive to Proactive. Cureus, 2024. PMC11490658. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11490658/
  7. U.S. Food and Drug Administration, CDER/CBER/CVM. Guidance for Industry, Process Validation: General Principles and Practices. January 2011 (Revision 1). (source not from a journal, for example from a manufacturer) https://www.fda.gov/files/drugs/published/Process-Validation--General-Principles-and-Practices.pdf
  8. Pharmaceutical Engineering (ISPE). Pharmaceutical Cleanroom Design & ISO 14644-16, Sep/Oct 2021, air-change-rate optimization for classified cleanrooms. (source not from a journal, for example from a manufacturer) https://ispe.org/pharmaceutical-engineering/september-october-2021/pharmaceutical-cleanroom-design-iso-14644-16
  9. Luca SV, Roehrer S, Kleigrewe K, Minceva M (2020). Approach for simultaneous cannabidiol isolation and pesticide removal from hemp extracts with liquid-liquid chromatography. Industrial Crops and Products 155:112726. doi:10.1016/j.indcrop.2020.112726. https://doi.org/10.1016/j.indcrop.2020.112726

Each citation in the paper, for example [1], refers to an item in this list. Each item is a primary source or an instruction from an authority, unless the item shows a different type. The results of cannabis tissue culture change much with the genotype. Before you use the dilutions, hormone doses and local regulations, make sure that they agree with the primary sources.