Polishing Dark Extracts:  Carbon Scrubbing & Diatomaceous Earth (DE) Filtering

These days everyone is hypnotized by light color extractions even though color alone does not determine quality.  Important quality considerations are actually clarity, cleanliness, aroma, terpene content, and potency.  It’s very possible to have a high quality concentrate that’s dark in color, or a low quality concentrate that’s light in color, so the deliberation shouldn’t be over what color the concentrate is, but why it’s that color.  This post is to help those that have concentrates, or may be making concentrates, that are dark for the wrong reasons and provide a reliable process to improve the product quality.

Everyone has had a wash or end product that came out darker, more green, or containing more undesirables than they hoped or intended.  Questions on how to deal with dark concentrates and extractions come up every day.  The easy answer is to take more care early in the extraction process so the undesirables are avoided altogether, but sometimes it’s too late to avoid and you just want to clean it up and carbon is the easiest answer.  It’s important to understand that when an extract is polished it loses desirable components along with the undesirable, inadvertently removing cannabinoids and terpenes.  Here lies the conundrum, do you accept the extraction as is, or do you polish it and sacrifice some of the desirables? To help crafters make that decision this post will explain a simple way to perform the polishing, using activated carbon powder for scrubbing to lighten the color and diatomaceous earth for very fine filtering, provide a side by side comparison of polished and unpolished results from the same extraction, and look at what was lost in the process.


I used decent quality trim and 190-proof food grade ethanol to make the wash.  Both the trim and ethanol had been in the freezer for a number of days and were between -2°F to -10°F.  To get the green wash I needed I did a 1-hour freezer wash to get some green pickup without going to a full FECO.



The 1-hour wash initial filtering went through fine mesh coffee filters for plant matter separation, then through the Buchner funnel using 3-micron filter paper.  The wash came out with the perfect amount of green and just over 1L in volume.




I separated two 300 ml samples from the 1L of filtered wash.  One sample would be processed as is and the other would go through the polishing process.  I will refer to the sample that will be processed as is as the ‘control sample’ and the other as the ‘polish sample’.  The sample size of 300ml was used to match the capacity of the Source Turbo used to make the oil and recover the ethanol.




The carbon is essentially charcoal and functions in the same way as a charcoal filter commonly used to purify water.  Carbon powder is really good at removing contaminates from liquids, but unfortunately it does this in a very indiscriminate manner and is difficult to filter out of the wash due to the very fine particle size.



The carbon scrubbing process itself is nearly as easy as mixing the wash and carbon together and shaking. The process is most effective if the wash is warmed to about 80F before adding the carbon.  The recommended amount of carbon to use is about 5% of the estimated volume of oil that is contained in the wash.  For 300ml wash, I estimated about 3g of oil so the recommended amount of carbon to be used was 0.15g.  I made and error here and added quite a bit more, about 1.5g, but I proceeded and pretended not to notice.  I originally planned to use a stir plate for agitation, but a friend with much experience advised that shaking provides much better results, so I put the wash and carbon in a small jar and gave it a good shake instead. That’s all there was to it and the carbon scrubbing was complete, but the resulting wash was utterly black with carbon and required very fine filtering.


Wikipedia defines diatomaceous earth: “Diatomaceous earth ( /ˌdaɪ.ətəˌmeɪʃəs ˈɜːrθ/) – also known as D.E.diatomite, or kieselgur/kieselguhr – is a naturally occurring, soft, siliceous sedimentary rock that is easily crumbled into a fine white to off-white powder.”



The DE is a very fine powder, almost like flour, and when mixed with ethanol and properly filtered with a vacuum assisted Buchner funnel will form a filtering cake that makes for very fine filter media.  I layered a Buchner funnel with a 3 micron filter paper, then a coffee filter on top of it to help hold the DE in place and form the filter cake.

To make the DE filter cake I first stirred together the DE and ethanol to make a slurry, that resembled a mud puddle.  When the slurry is poured into the Buchner funnel, with the vacuum assist running, the ethanol sucked through into the beaker and the DE was left behind in a solid cake form.


The more you pour in the thicker the cake will form.  I don’t know, and can’t find information on, what thickness of the filter cake is ideal, so I made it about a 1/4-1/2 inch thick. I let the cake settle into a solid form then began adding the carbon wash for filtering before the DE cake dried out.

Even with the vacuum assist from a large pump the progress was quite slow, it took about 5-minutes to filter 300ml.  I think this would have worked better if I had used one of the larger Buchner funnels.  The slow progress was no real problem, but the 1L size would have provided more filter surface area and better flow.  That was a good lesson learned here and a point to keep in mind.  Another thing to keep in mind is that the Source Turbo vacuum port can be useful for normal vacuum assisted filtering, but for this kind of heavier work it’s probably not be sufficient.


The polish sample worked its way through the filtering layers and emerged in the beaker cleaned up and glorious.  When the control sample and polish sample are set side by side the improvement in color is obvious.



The processing plan was simple.  First, use the Source Turbo (extractcraft.com) to separate the oil and reclaim the ethanol, then pop the oil samples into the vac oven to finish.  I experienced a small spillage mishap of the polish sample so I evened the samples to 250ml each for processing.  Each sample ran separately in its own Source Turbo for about 2 hours, then in the vac oven overnight at 92F.


The difference in the resulting oils was literally night and day, light versus dark.  Both oils were very clear and clean, the darker oil was obviously carried over the green tinge from the wash, while the blond oil was defiantly pleasing to the eye.



Out of the vac oven the completely purged end products finished as expected with the lighter sample as blond as could be and the darker turning out much better than expected.

Directly from the vacuum oven
Shatter after folding and condensing samples into thicker form

I knew polishing would improve the concentrate color but I didn’t really know what to expect with regard to the how much scrubbing would effect the yield or quality of the final product.  For me, determining exactly how much oil was lost in the process was the most interesting part of the whole exercise.  The control sample was obviously going to be heavier than the polished sample, but by how much?  The control sample was 2.75g compared to the polished sample coming in at 2.26g.  The scrubbing and filtering reduced the yield of the final product by 0.49g or 17.8%.

All of the colored, undesirable material was removed from the polish sample along with a small amount of terpene character.  The darker extract ended up being both, more rich with terpenes and robust, but the green from the wash was identifiable by a faint plant taste. The terpene profile was a little weaker, not by an enormous amount but by enough to note the difference, but the psychotropic effects were notably more potent than the unpolished extract.


That was a lot of fun!  I’ve been wanting to do this project for a while to figure out if carbon scrubbing and the necessary advanced filtering is worth hassle and yield sacrifice.  It is obvious by the results that it works very well to lighten the color and increase potency. So, when are those benefits of better color and increased potency worth sacrificing the 17.8% in yield?

Let’s start with when I would not consider polishing useful or necessary. For edibles, tinctures, or topicals it would not be useful, and I would instead opt for the benefit of maximizing yield over color, taste, and potency.  The color and taste will be diluted by whatever carrier is employed and the potency is easily adjusted by the ratio of oil used.  I would also not consider polishing for a smokable if it were a singular concern over color when all other considerations of clarity, cleanliness, aroma, terpene content, and potency are all acceptable.  We know from by blog post ‘Color and Cannabis Concentrate Quality’ that dark concentrates can be as good or better as lighter concentrates.

I will only recommend polishing if my wash inadvertently picked up green when I didn’t want it to, or I have an already made concentrate that has an unpleasant flavor.  What I would highly recommend is avoiding the problem from the beginning and use extra caution in careful wash execution, like in my blog post ‘Cannabis Oil QWET Extraction Battle of the Wash: Dry Ice vs. Freezer’.

I hope this post provided some useful insight and  enjoyment.  If you appreciate my work please share  and spread the word!  Stay lifted and blessed.



33 thoughts on “Polishing Dark Extracts:  Carbon Scrubbing & Diatomaceous Earth (DE) Filtering

  1. I need some expert advice:

    I cannot find high concentration THC to import into my country to help treat cancer. However, I can purchase without problem shatter/wax that contain at least 80% THC and I can buy locally 95% pure ethanol from a distillery.

    The plan is to dilute the shatter into the ethanol and separate the solvent and impurities, such as butane, etc… from the THC while keeping the terpenes and any other beneficial components together…

    This will be small scale production, mainly for my own use and it makes no sense to buy a full scale short path distillation set up. I can circulate refrigerated water through the condenser part and I have a heating source: what else do I need to make it all work?

    Any suggestions??…

    1. I’m not understanding why you would breakdown the shatter? What do you suspect is wrong with the shatter? Also, even low THC starting material will concentrate into high percentage concentrate….. What is your intended use?

    2. One quick question
      When you warm the wash to about 80F snd add the AC and agitate it, do you cool it back down with dry ice or cool it down with anything before pouring it into the funnel filtering through the DE cake?

  2. What is the mechanism for the filtering for the green leak, because the rate that you described is the flow rate for the filter media, which Whatman says that it flows at a 64 ml/min. Your rate was 5 minutes for 300 ml, in line with the Whatman specification, which gave 60 ml/min. So the controlling filter media appears to be the 3 micron filter since it is flowing per the specified rate. So what caused the chlorophyll to bind to the carbon and diatomaceous earth since it wasn’t acting as a filter media. It has to be a chemical bonding I would think instead of filtering….

      1. If the filter media is flowing at the stated rate for the filter media, about 60 ml/min for the 3 micron filter then if the diatomaceous earth was acting as a filter it has to be longer than the 5 minutes it took to filter the wash. 300 ml/60 ml is five minutes, the stated rate of the filter by most manufacturers of slow filters, which a 3 micron would be. So the flow rate was the same pre as well as post scrubbing. So I suspect that the process is a chemical process and not a filter process. Since it should take longer, and the filter media originally passed the green leak, what stopped it after scrubbing, since the filter was acting as it should and passing the wash at the same rate. Something is askew…

        Have you tried any hydrophobic filters to filter out green leak…?!?

      2. There wasn’t anything askew as far as I know. The 3 micron under vac is pretty fast, it obviously the DE that slowed the flow. What kind of hydrophobic filter? Molecular sieve?

  3. Nice post. I see you had 1L after filtering. How much plant matter was required to produce that much wash? I’ve got a ounce of nice flower and a couple ounces of trim. The trim has been sitting in the freezer until I fired up the Source again. The flower has been in a black glass storage container with a couple small Boveda packets.

    Would a food dehydrater work as a cheaper option to the oven you use? I can get a good waxy result if I leave it on the mat for a couple days, but not a hard shatter.

    Instead of a stir plate, I’ve used a ultrasonic jewelry cleaner to mix dissimilar liquids with great results. Cheap on Amazon.

    1. Hi Ron,

      I ended up with about 1.2L. I’m not sure exactly how much plant matter was used in terms of weight. The photo in the blog is exactly what I used, the 2L jar with very fluffy trim, I’d guess about 2 oz or so. My focus was on comparable wash samples so didn’t bother with the weight of starting material because it wasn’t of interest.

      The end result of what you have on the mat will depend on the strain itself, quality of filtering, quality of winterization, state of material, humidity, and such. Another problem I often see is people lay a think glob or line of concentrate on mat and expect it to turn to shatter and unfortunately it won’t, it needs to be very thin if using natural purging process.

      With the carbon its not a matter of mixing to incorporate, but for scrubbing so the ultrasonic wouldn’t work well.

      The dehydrator might be a stroke of genius, I’ll have to try it and think about it a bit.

      1. We discussed a food dehydrater when a vacumn oven is not available. I haven’t been able to retry it.

      2. the comment right above this asks using a dehydrator instead of your vacuum oven to purge where you then reply this may be a stroke of genius and you may try doing that

    2. Hi ron, I like the idea of a ultrasonic jewelry cleaner to mix things. Do you just place the glass? container in side the cleaner and let it run? I just read that you shouldn’t let the ultrasonic jewelry cleaner run dry for more than 30 seconds? curious?:)

  4. Guys, I am keeping up with this – as I use extractors for food (vanilla extract and the like), if I want to dehydrate, don’t use heat – use a freeze drier from https://harvestright.com/ It will give you far, far better results, and you also end up with a good vacuum pump too (-;

    (freeze drying removes all water using a vacuum at a low temperature – so more ‘goodness’ is preserved (-;


  5. i just tried this, didnt really see any difference so far so . Both of the pics i took of the wash before and after look the same (dark brown/amber) how long did you soak the wash with the carbon? i used about 1.5g for 5oo ml as well…im gonna winterize now and then filter again time will tell thanks for this though..

    1. For this I used about 2g or so of activated carbon for about 300 ml. With ethanol warmed to around 80F, and shook quite well for about 5 minutes? The effectiveness will depend on a number of factors; agitation, temp, how dirty the wash is, time exposed, etc.

  6. I noticed that it looks like you didn’t wash the filter cakes afterwards. You had 300ml of solution before filtering and about 250ml after. Couldn’t you have increased the yield by about 16% if you washed that lost solution out of the filter cake with clean ethanol? I’m also curious to see what the difference in color etc. would be using 0.15g of carbon like you intended since it would barely affect the yield.

    1. I don’t believe washing the filter cakes would have resulted in anything consequential. I believe I mentioned the 50ml loss was due to a spill accident, if not I’ll have to go back and clarify. Using less carbon would have defiantly made a difference. Using too little also has the opposite effect, so if you are doing it for top performance you have to work with the same material and same wash a bit to find the perfect balance for each batch, its defiantly not an exact ratio for every wash. It’s going to range from 3%-10% and probably best around 5%-7% depending.

  7. Where did you get the carbon powder? I want to make sure I dont get too fine of powder. Im using a 3 micron funnel paper and the DE cake you described

  8. I used this method and it worked wonderfully a few times.

    Right up until this last batch. I put too much carbon in it and it robbed me of most of my medicine. I started at 1g of concentrate and ended up with like 0.2g. I was careless with the carbon and too much got into it. Won’t do that again. I thought I’d share that piece of info with everyone else so they won’t make the same mistake.

    1. Can use iso, I prefer ethanol for the food grade aspect. I was going to use Celite 545 but was told that for ethanol DE works better. Something to do with the polar component of the solvent, but not sure. If anyone knows I’d like to get schooled.

  9. After you mix in the DE how long do you wait before filtering through the cake? Also, I used. 0.5g in 2000ml, at 75 deg. F. it cleaned it but not as much as unexpected when compared to the other jar not scrubbed yet. Is that way too little? I estimated may e 10 grams of oil so used half a gram. Does volume of wash come into play with amount if AC used? I may add 1g and scrub it again

Leave a Reply