The target constituents in cannabis that we covet, are aromatic hydrocarbons soluble in the simple Alkanes n-Butane and n-Propane, as well as their isomers Isobutane and Cyclopropane.  BHO extraction involves passing one of those solvents, or a mixture through the plant material and dissolving the trichomes containing the target elements.

The LPG is then boiled away in the collection tank, leaving the raw oleoresin extracted from the plant material behind.

Butane and propane are both non polar, so not prone to pick up excessive water or chlorophyll, but does extract other undesirable C-30 molecules like anthocyanin plant pigments and plant waxes.

Whole plant extractors won’t find that objectionable, but in other products those components are undesirable, so we avoid them as much as possible by extracting at subzero temperatures, typically between -30C and -50C.  That reduces the solubility of all the constituents, and the ability of the solvent to hold solute, but it slows down dissolution of the C-30 molecules more than the C-10 through C-22 targeted elements.

It also freezes any water present, which locks in both the water and any water solubles, so that they aren’t exposed to the solvent.

An alternate to that procedure is to extract at ambient temperature and then inline dewax by dropping the LPG cannabis essential oil solution to -30/-50C, so that they precipitate out of solution, so that the plant waxes can be subsequently removed by filtration.  I will return to that subject at the end of the discussion on subzero extraction and processing.

Plant waxes can also be subsequently removed by winterization, which involves dissolving the raw BHO oleoresin in a polar solvent like ethanol, and placing in a freezer long enough for the plant waxes to precipitate out so they can be removed by filtration, to make an Absolute. 

For some products that works well, but for those relying on retaining high levels of monoterpenes it doesn’t, because many of the monoterpenes exit with the alcohol when it is distilled off.

We typically either extract at subzero temperatures so as to not extract the plant waxes in the first place, or winterize with ethanol afterwards for those products where an Absolute works well.

Freezing plant material ruptures the cells, exposing the contents to the solvent, so best done one time and kept that way until extraction is complete.  We prefer to pack our columns and then freeze them, to minimize any condensation pickup from the atmosphere that might occur packing frozen material, which can form ice over the trichomes that we are attempting to harvest. 

It also minimizes opportunities for the material to thaw, but if done in a walk in freezer, that wouldn’t be an issue.

Another clever way I’ve seen to deal with large amounts of material and limited columns, is to make tubes out of bedsheet material, that can be prepacked and frozen, before inserting into a column.

Have everything prepared when you pull the column out of the freezer, so that in can be quickly bolted in place, evacuated, and flooded with subzero solvent before it has a time to defrost.  You can increase the time envelope using removal insulating sleeves on the columns.

The next issue is to chill the LPG to -30C or below.  At those temperatures, the viscosity of the LPG increases, so pushing the solvent through the column requires more pressure.  Just the push pull from the Haskel recovery pump is typically enough for a properly packed column, but below around -30C we switch to a Butane/Propane mixture, typically either 70/30% or 50/50%.

Injecting LPG from a tank cooled to those temperatures can be problematic due to the increase in viscosity, so we keep our LPG supply and recovery tank around 0C/32F and use counter flow heat exchangers to chill it after it leaves the tank and is on its way to the column.  Using liquid nitrogen (LN2) for the coolant on the jacket side, allows us to operate down to about -50C.

Another method commonly used to achieve the same end is to use a 50′ or longer coil of stainless tubing in a antifreeze/dry ice bath as the heat exchanger.

That brings us to pressure.  Even a 50/50% mix of Butane/Propane at 0C/32F has only about 7.6 psi tank pressure, so below about -30C we use a pressure boost during injection, to augment the push pull from the pump.

There are a couple techniques to achieve that end.  We use an auxiliary tank of the same LPG mixture sitting in up to a 110F hot water bath to provide additional head pressure, and connect the hot and cold tank via their vapor ports.  A 50/50% mixture at 110F is 122 psi.

The advantage to using a hot vapor tank versus a inert gas to pressurize the system, is that no subsequent burping is required to remove it.

That brings us to an alternative to the hot vapor tank, and that is inert gas, and N2 is commonly used for the purpose.  A pressure bottle of N2 is connected to the vapor port of the supply and used to add head pressure for injection.

That technique does require subsequent burping before the recovery pumps can recover the LPG, because non condensable gases in the system slow recovery down to a crawl.  Burping always requires venting some LPG with the non condensable gases, so recovery is less and operating costs higher.

Here is a chart showing the relationship to the different mixtures of Butane and Propane and various temperatures, compliments of The Engineering Toolbox at:

Vapor Pressure in PSIG
 MixturePropane %1007050300
Butane %0305070100
      Temperature      (F)-44F00000

 Here is Isobutane pressure at temperature:

How and how much solvent a system passes through the plant material depends on equipment design and process, but as a for instance, a Mk IVC Terpenator has a 12″ X 12″ recovery pot and 4″ X 36″ column, which we operate as follows:

We flood from the bottom and typically pass two (2) volumes of solvent through the column from the bottom and overflowing to the recovery tank from the top, and then open the lower column dump valve and rinse one (1) column volume the top.

We monitor the color of the discharge from the top of the column using a sight glass, and may shorten the bottom flood if the color suggests it, so three volumes is only a rule of thumb.

We leave the column dump and vent valve open during recovery, until we reach about -12/15″ hg vacuum, at which point we close both valves and open another valve plumbed directly to the filter drier, bypassing the collection tank. 

That permits us to inject 150F water into the jacket surrounding our column, which drives out the residual LPG and recovers, without affecting the temperature in the recovery tank.

One reason for not wanting the hot LPG in the lower collection tank at that point, is that it negatively affects some fragile processes like Cotton Candy, which is my preferred way to recover.  Cotton Candy is cannabis essential oil foam, produced by inflating the oil under vacuum and drying it in that condition.

We accomplish that by holding our collection tank bath temperature around 21C/70F during recovery and using a sight glass in the collection pot lid to determine when the visible LPG is gone, but bubbles still blurping from the puddle.  At that point we turn off the Haskel recovery pump and turn on the Vaccon venturi vacuum to evacuate the collection tank to -29.5 Hg vacuum levels.

The discharge from the Vaccon vacuum is plumbed to the extraction system and it is pneumatic and meets NEMA 7, Class I, Div I requirements, so no vapors go into the extraction booth.

The timing of now is where experience comes into play, but if you are watching the inflated foam through the sight glass, you will notice it starting to lose its glisten as it dries in place and can tell when its dry.

The next step in this dance is inert backfill, and you can use it to periodically check to see if you are dry enough.  We backfill our systems after extraction with N2, so when we open them the contents are below 10% of LEL.

If you meter the N2 in slowly and watch the foam as the pressure rises, you can if it is starting to slump or not.  If it starts to slump, close the back fill valve and turn on the high vacuum again.  Do this carefully, because the slump can’t be recovered now that all the solvent that inflated it is gone.

That brings us to plant material quality.  We typically yield 20/25% from prime bud and sugar leaves, and about half that for just trim.  Some strains are low oil producers, and we’ve seen as low as 5.7% from a landrace high CBD strain.

Some also inflate into Cotton Candy better than others, and the THC and CBD both need to be in carboxylic acid form to be stiff enough to dry as shatter.  Older decarboxylated material may inflate but when the pressure is raised, it will slump to the bottom as oil.

Instead of cotton candy, you can also simply recover the LPG and scrape out the raw oleoresin that is left behind.  It will end up with more residual LPG to remove later in finishing, but the system can be backfilled with N2 so as to be inert when opened.

Lastly, one process calls for opening up the system before the LPG is fully recovered and pouring off the last bit into a thin film on parchment or PTFE film.  In that case, the pot is not inert when opened and poured, so not one we employ.  The fire hazard might be addressed by opening in an exhaust hood designed for the purpose, but I’ve never personally gotten such a system certified and permitted.

Here is where we return to the subject of inline dewaxing, which is gaining popularity, but though I actually did design the Mk VII and Mk VIII around that process, I never built one for testing.

 Mk VIII conceptual

The way inline dewaxing works, is that there is a chamber under the column large enough to hold the volume of LPG you pass through the column, where you can hold the LPG essential oil mixture and drop the temperature using an coolant and dry ice in a jacket around it.  This allows the plant waxes to precipitate out of solution, so they can next be drawn through a filter to remove them.

The Mk VIII above uses a shell and tube heat exchanger for the hold and chill tank, with a 4″ X 8″ one (1) micron sock filter for wax removal.  When the valve is opened, the recovery pump connected to the lower collection tank, draws the solution through the sock filter.

An alternative to the sock filter, is filter plates and face filters, but they are limited in surface area and pores, so blind easily, especially in sizes as small as one micron. 

Leave a Reply