Optimal Dilution of Polysorbate (Tween) 20 for the Determination of Enterobacteria and Total Yeast and Mold on Cannabis Concentrate


In this study, Tween 20 amounts were increased in 1% increments while monitoring two different factors.

The microbial plating of cannabis and hemp derived concentrate has been historically challenging because of issues with consistency. Tween 20, an emulsifying agent, is often added to diluent to aid in extracting the concentrate. In this study, Tween 20 amounts were increased in 1% increments while monitoring two different factors. The first factor was the turbidity of the solution, which determines the saturation point at which no more concentrate can be extracted, given the Tween 20 method. The second factor was the microbial survival rate, measured in colony forming units (CFU) for E. coli and C. albicans. Per the state of Nevada regulations (NCCR effective November 5, 2020), enterobacteria and total yeast and mold are plated for concentrates, thus these two microbes were chosen to be representative. Following analysis, it was determined that a solution of 3% percent Tween 20 was optimal for survival of target microbes and emulsification of concentrate.

The state of Nevada Cannabis Compliance Board (CCB) defines concentrated cannabis as “the extracted or separated resin, whether crude or purified, containing tetrahydrocannabinol (THC) or cannabidiol CBD” (1). Through this extraction process, cannabinoids become highly concentrated while simultaneously making it challenging to extract microorganisms from matrices. Due to this innate difficulty, most manufacturer methods offer a patented emulsifying agent that aids in the breakdown.For plating applications, the use of an emulsifying agent named Tween 20 has been widely adopted. Tween 20 is a non-ionic detergent that aids in emulsification for the preparation of stable oil-in water (2). Ideally, adding the solution in small amounts serves to aid in breaking up the concentrate. However, a caveat of the process is ensuring that a suitable environment for the target microorganisms remains after application of Tween 20.

Though cannabis and cannabis-containing products are illegal on the Federal level in the US, certain states have legalized them to allow for medicinal or recreational products. As a result of this patchwork legalization, the testing (and more specifically the targeted microbiological contaminants) are determined state by state. In Nevada and several other states, concentrated cannabis products need to be tested for enterobacteria and total yeast and mold (1,3). Detection methods are structured via a threshold system wherein microorganism may be detected, but as long as they are below the target threshold then they are considered safe for human consumption or use. In large quantities, it has been found that these categories of microorganisms can lead to human harm if not properly tested for (4). Taking this into account, researchers hypothesized the following: what was the optimal amount of Tween 20 solution that could be added until the environment became unfavorable.(measured in colony forming units [CFU]) for the plated target microorganisms?

Experimental Methods

Study Design

For the study of enterobacteria, E. coli 0157:H7 NCTC 12900 was the representative organism. For the study of total yeast and mold, C.albicans ATCC10231 was the organism used. The procedure used to create viable spikes prior to their dilution into Butterfield’s phosphate buffer and Tween 20 is property of RSR Analytical Laboratories. Incrementally increasing concentration of Tween 20 was diluted appropriately as described in Table I. While aseptically creating the dilutions into sterile 150 mL dilution bottles, 100 mL total was created. Of this solution, 9 mL of each respective dilution was added to 1 g of concentrate distillate. This solution was then vortexed for 1 min and then added to a sterile turbidity vial. The vial was allowed to sit, followed by the turbidity procedure. After turbidity was done, either aerobic plate count or yeast and mold method was conducted, dependent on the spike utilized in the preparation process. All remaining milliters of diluent were plated on their respective plates, and resultant CFUs were counted. CFUs were then averaged for each percent of Tween 20 added and analyzed with survival of the target organism as the outcome of interest.


Turbidity of the diluent was measured using a LaMotte TC-3000e turbidity meter. To measure turbidity, the meter was calibrated utilizing known nephelometric turbidity unit (NTU) standards (5).The various concentrations of the Tween 20 solution were then added to the 1 g of cannabis distillate concentrate. The sample was then vortexed for 1 min to completely allow the emulsifying agent, cannabis distillate, and microorganisms to mix completely. The solution was transferred to a sterile turbidity tube and allowed to rest on the bench for 30 s. The turbidity of the sample was taken and recorded. If the turbidity of the liquid was too numerous to calculate, it was subsequently diluted to be in a readable range. Overall the point of this method was to gain a general understanding of how much concentrate was eluted into the buffer.

Rapid Aerobic Plate Count

Rapid aerobic plate count petrifilm by 3M is a self-contained dehydrated media that facilitates the growth of bacteria aerobically for the purpose of enumeration. Following the measurement of turbidity at least 8 replicates, plates of 1 mL of sample were pipetted to the center of the plate and pushed with a 3M flat spreader. These plates were incubated at 35 ±1 °C for 24 ±2 h. This was done in accordance with the instructions given by the manufacturer and AOAC recommendations (6). Following incubation, CFUs were counted as a measure of microbial survival. Though plated on aerobic plate count (APC), this served as a target organism for enterobacteria.

Rapid Yeast and Mold

Rapid yeast and mold petrifilm by 3M is a simple ready to use plate system with specialized antibiotics to facilitate the growth of only yeast and mold on the plate in a rapid manner. The plates of 1 mL of sample were pipetted to the center of the plate and pushed with a 3M flat spreader. These plates were incubated at 25 ±1 °C for 60–72 h (7–9).This was done in accordance with the instructions given by the manufacturer and AOAC recommendations. Following incubation, CFUs were counted as a measure of microbial survival.

Statistical Methods

Data consisted of roughly 90 samples of CFUs from rapid APC and yeast and mold (Y/M) petrifilms taken at 10 levels of Tween 20 (1% to 10%, by increments of 1%), with minimal missing data. To assess whether there was a value of Tween 20 at which the recovery rate of APC and Y/M significantly changed, two threshold regression models with hinge threshold effects were fit with Tween 20 as a continuous predictor, and APC and Y/M as separate outcomes (10).A hinge threshold effect was used; we assumed that the recovery rate would be constant (flat) until the change point value, at which point it would begin decreasing, rather than using a step threshold effect that assumed two constant recovery rates that were different between the threshold value. Alpha for a significant difference in recovery rates above and below the threshold Tween 20 value was set at 0.05. All analysis was conducted using R version 4.0.1.

Results and Discussion

The results of the turbidity experiment are shown in Figure 5. Following the procedure outlined above, the results were plotted along their respective Tween 20 concentrations. The data appears to follow a logarithmic pattern; no matter how much Tween 20 is added, there is a limit of how much cannabis concentrate can be extracted through this method. Following 7% the growth appears to level off, rendering any additional Tween 20 useless (Table II).

Following the proper incubation and CFU counting, statistical analysis was conducted to assess any differences between plates. The results of the threshold regressions with a hinge effect can be seen in Table II; in the APC model, a changepoint was identified at Tween 20 of 6% (95% CI 4, 9), with an increase in a unit of Tween 20 values larger than 6% associated with a decrease in APC recovery rate of 3.98 (95% CI: -14.87, -2.17), while Tween 20 less than or equal to 6% had a constant recovery rate of 63.81 (62.01, 66.41). However, this difference in recovery rate beyond the threshold value of 6% is not significant (p-value=0.22), indicating that there is not sufficient evidence to suggest that 6% is a threshold value associated with a significantly different recovery rate.

In the Y/M model, a changepoint was identified at Tween 20 of 3% (95% CI: 2, 4), with an increase in a unit of Tween 20 values larger than 3% associated with a decrease in Y/M recovery of 5.32 (95% CI: -7.18, -4.23), which was significant (p-value<0.01). Tween 20 values less than or equal to 3% had a constant Y/M recovery rate of 208.69 (95% CI: 203.29, 212.91). So, there was evidence suggesting that Y/M recovery rate significantly decreased at Tween 20 values larger than the 3% threshold, while Tween 20 less than or equal to the threshold value had a constant recovery rate. Figures 1 and 2 demonstrate the APC and Y/M log likelihood plots of each threshold value, and frequency plots of each threshold value from 1000 bootstrappings of the analysis, respectively.

Figures 3 and 4 present the fitted lines for the threshold regressions, with a flat pre-threshold value slope, and the post-threshold regression line. While both post-threshold lines have negative slopes, the APC line was not significantly different than the APC pre-threshold slope, whereas the Y/M post-threshold slope was significantly different than the pre-threshold slope.


Innovations in cannabis science allow researchers to push forward in finding new ways of conducting science accurately. Concentrated cannabis has exceptionally difficult matrices to accurately test for microbes because of an oftentimes hard and sticky consistency. While sometimes the amount of extraction, or turbidity, can indicate success, microbiologists must also take into account the environmental conditions of the organisms targeted for analysis. In this study, E. coli 0157:H7 served as the model for enterobacteria and C.albicans served as the model for total yeast and mold. It was determined that, when utilizing Tween 20 to aid in the emulsification of concentrated cannabis, no more than 3% of Tween 20 to dilution should be added (Figures 3 and 4). The more Tween 20 that is added, the more the turbidity increases, which can contribute to better emulsification of concentrate. However, if it leads to an unfavorable microbial environment, the target organisms might begin to perish and, subsequently, results will be underreported (Figures 3–5).

The goal for cannabis scientists is to keep all consumers, medicinal and recreational, safe. By doing specific research on target matrices, it is possible to incorporate food and environmental testing methods to match cannabis testing needs. Minor changes to the method could be beneficial in testing, and potentially lead to more precision and quality. Many scientists in the cannabis field strive to develop the science; working with state-to-state regulations provides a unique opportunity to aid the entire field. By continuing to publish and grow the grassroots of testing, any increases in testing quality will hopefully permeate across state regulations to the benefit of all cannabis users and producers.


(1) NRS: Chapter 453 - controlled substances. (n.d.). https://www.leg.state.nv.us/NRS/NRS-453.html#NRS453Sec042.

(2) "TWEEN® 20 for molecular biology, viscous liquid,"Sigma-Aldrich | 9005-64-5. (n.d.). https://www.sigmaaldrich.com/US/en/product/sigma/p9416.

(3) S. Dailey, "Cannabis microbial testing regulations by state," Medicinal Genomics. https://www.medicinalgenomics.com/cannabis-microbial-testing-regulations-by-state/ (2021).

(4) R. Upton, L. Craker, M. ElSohly, A. Romm, E. Russo, and M. Sexton, "Cannabis inflorescence: Cannabis spp. ; standards of identity, analysis, and quality controlm," American Herbal Pharmacopoeia (Scotts Valley, California, 2014).

(5) "LaMotte TC-3000e Manuals," Manuals Library. (n.d.). https://www.manualslib.com/products/Lamotte-Tc-3000e-11491053.html.

(6) AOAC Official Method 2015.13 Enumeration of Aerobic Bacteria in Food 3M™ Petrifilm™ Rapid Aerobic Count Plate First Action 2015

(7) 3M Petrifilm Rapid Yeast and Mold Count Plate Interpretation Guide. 3M, https://multimedia.3m.com/mws/media/902046O/3m-petrifilm-rapid-ym-count-plate-interpretation-guide.pdf. Accessed 2 February 2021.

(8) P. Bird, J. Flannery, E.Crowley, J. Agin, D. Goins, and R. Jechorek, "Evaluation of the 3M™ Petrifilm™ Rapid Yeast and Mold Count Plate for the Enumeration of Yeast and Mold in Food: Collaborative Study, First Action," Journal of AOAC INTERNATIONAL, 98(3), 767–783 (2015).

(9) A. Repay, Cannabis Science and Technology 4(3), 32-34 (2021).

(10) Y. Fong, Y. Huang, P.B. Gilbert, and S.R. Permar, BMC Bioinformatics. 18(1), 454 doi:10.1186/s12859-017-1863-x (2017).

About the Authors

Anthony J. Repay, M.S. is the Director of Microbiology at RSR Analytical Laboratories. Wyatt J. Tarter, M.S. is a Research Instructor Biostatistician for the Colorado School of Public Health’s Center for Innovative Design and Analysis. Direct correspondence to: repayanthony@gmail.com