Ketamine synthesis work

Hi everyone, I've posted a while ago about some problems I was having in the synthesis of ketamine. I'm happy to announce that I sorted out everything with your help, and i concluded my work. It's quite long, but it includes the synthesis of some reagents, 2 synthesis of 2-chlorophenyl cyclopentenyl ketone (+ some tips for the grignard!), the synthesis of ketamine via thermal rearrangement, autoxidation ring expansion and finally chiral resolution. Sorry for the long post, but it's also full of comments and my experience, which some might find useful. There's still something to fix, but all aorund i'm proud of it.
Also, I know that drug synthesis is not very welcomed here, but the real motivator behind this work was the science of it, and i believe it might be useful for this community too. If this is in anyway against the rule I'll understand. Thanks!

KETAMINE SYNTHESIS, COMPREHENSIVE GUIDE AND DISCUSSION OF VARIOUS PROCEDURES, by arclightshroom

This work serves as an investigation in the synthesis of ketamine. No new discoveries are made, but ketamine somewhat lacks love in chemistry forums as there are very few procedures around. In this project the classic route is explored with various techniques plus one new and interesting procedure Ive found in a recent paper that no one seems to have tried and reported. With this I hope to make the process smoother and easier for future bees that intend to explore this compound, cheers!

Part 1, making the reagents

This section illustrates the synthesis of all the major reagents used, apart from those deemed to be too hazardous or easily acquired. For an easier reading of this work I suggest starting from section 2, and only consult this section to see how certain chemicals were made.

1) Cyclopentene and cyclopentyl bromide:
Adipic acid-->cyclopentanone: Cyclopentanone was bought directly, as its cheaper to acquire than to make in large amounts. A small scale reaction was carried out as a proof of concept, following this procedure:[ftp] https://orgsyn.org/demo.aspx?prep=CV1P0192 [/ftp]
NOTE: Temperature control is critical for obtaining high yields in this procedure, as above 300C adipic acid rapidly distills over clogging everything and ruining the yield. Keeping the temperature at 280C will give you room for error while still carrying out the reaction at a satisfactory rate.

cyclopentanone-->cyclopentanol: 24g (0.5 molar equivalents) of NaBH4 were dissolved in 400mL of ice cold 2M NaOH and the flask was placed in an ice bath. With strong stirring, 100g of cyclopentanone were added dropwise over the course of an hour, taking care to replenish the ice and keep the reaction mixture at around 20C. After everything was added the reaction was left undisturbed for 10 minutes, and then gently refluxed for half an hour to destroy remaining hydride. The aqueous layer was extracted 3 times with 20mL DCM, and the extract added to the previously separated organic phase. A bit of extra DCM was added, then everything was washed with brine, dried over sodium sulfate and after the solvent was stripped everything was distilled. The fraction boiling at 140-144C was collected, final yield 85%.
NOTE: Temperature control is not crucial, but the heat generated in the reaction is enough to boil everything, which is not ideal. Also, cyclopentanol is quite fragile and can dehydrate easily, so care must be taken if the hydride is quenched with acid.
The reduction of the ketone in plain water with NaBH4 is far superior than the one with aluminum isopropoxide, as while being a less OTC reagent its very commercially available, produces less waste and can be scaled better. Reduction with isopropoxide has a messy workup and introduces the problem of mercury waste, while in this reaction the product floats to the top as soon as its finished, leaving water as waste.

cyclopentanol-->cyclopentene: 80g of cyclopentanol and 32mL of 85% phosphoric acid were added to a flask and set up for distillation. A mixture of cyclopentene suspended in water distills over at 65-100C, and after everything has passed through the receiving flask is stoppered and left undisturbed until 2 layers separate. The top organic layer is recovered and washed first with a sodium carbonate solution and then with brine. To ensure the anhydrous conditions necessary for the next reaction cyclopentene was distilled over sodium sulfate at 50C. Final yield 88%
NOTE: Sulfuric acid can be used in this reaction, but it doesn't work nearly as well. The reaction with sulfuric acid goes MUCH faster but produces a lot of tar, the yield with sulfuric acid was less than 30%.

cyclopentanol-->cyclopentyl bromide: On a salted ice bath phosphorus tribromide, 96 grams, was added dropwise to 86 grams of cyclopentanol at such rate so that the temperature remained between 0 to 5C. After all the PBr3 was added the mixture was stirred overnight at room temperature, then around 150mL of water were added and the mixture steam distilled. The lower fraction was collected and washed with a sodium carbonate solution. In total 122g of bromide, corresponding to a yield of 82% , were obtained after drying. The compound is kept over anhydrous calcium chloride.
NOTE: Cyclopentyl bromide can be made in a more OTC fashion by reacting the alcohol with concentrated HBr, but i found that this method performed poorly due to the formation of cyclopentene, the dehydration byproduct. I haven't been able to achieve more than 53% with it.

2) 2-Chlorobenzoyl chloride and 2-Chlorobenzonitrile:
Both these reagents were more convenient to buy than make. These can be made from 2-Chlorobenzoic acid, obtained from anthranilic acid through Sandmeyer reaction. The acid can then be converted to the nitrile following Zealots work or to the acyl chloride with thionyl chloride, phosphorus trichloride or phosphorus pentachloride.

3) Copper (II) bromide:
To 25g copper sulfate pentahydrate in 100mL water a saturated solution of 10g NaOH is added, causing precipitation of copper hydroxide. The solution is then heated for 10 minutes until all the hydroxide decomposes to the black CuO, which is then filtered and dried in a oven at 200C, yielding 7.5g in an almost quantitative yield. All the copper oxide is added to 23mL of 48% HBr with stirring, forming an oily purple solution which crystallizes from the top in a vacuum chamber with a desiccant, yielding about 20g of product.
NOTE: To isolate solid CuBr2 from solution vacuum desiccation is the easiest solution, as it decomposes to elemental bromine when heated. A commonly used desiccant for this purpose is phosphorus pentoxide, but I found NaOH to be sufficient, although slower. After 2 days in the chamber with NaOH i was able to recover a huge cluster of crystals at the top of the beaker. In about a week everything was solidified. The solid is highly hygroscopic and has to be stored airtight.


3) Copper (I) bromide:
To a stirring solution of 30g copper sulfate pentahydrate in 100mL water 17.5g of NaBr are added. After all has dissolved 7.6g of sodium metabisulfite are added, everything is stirred for 1 hour, filtered and the resulting CuBr dried in a vacuum chamber. CuBr was obtained in a 87% yield.
NOTE: Once its removed from water CuBr quickly oxidizes to its Cu(II) form, so its crucial to dry it under vacuum without any oxygen. Some oxidation still occurs, and the white solid turns slightly green, but thats not an issue for the use as a grignard catalyst.

Part 2, making the ketone precursor

Every synthesis of ketamine starts from the same precursor, which is 2-Chlorophenyl cyclopentyl ketone, of which only 2 viable synthesis exist as today, the grignard and the modified friedel-crafts.

Friedel-crafts reaction:

To 43.75g AlCl3 in a dry rbf is added 50g of 2-chlorobenzoyl chloride in 150mL cold DCM under a stream of nitrogen, the flask is then put in an ice bath and stirring is commenced. Once the solution gets to -5C, 21.4g of cyclopentene is added dropwise over the course of 1 hour taking care that temperature stays around 0C. After all the cyclopentene was added the solution was stirred for 1 hour while the ice bath melted, then another 43.75g AlCl3 was added with 75mL cyclopentane and 15mL more DCM. After the addition nitrogen was interrupted and the solution refluxed for 2 hours. Once cool, everything was added dropwise, slowly, into a large beaker containing ice, 600mL of water and 200mL DCM. The mixture is filtered over a buchner funnel under vacuum, and the precipitate washed with DCM until white. The organic layer is collected and washed with 25% NaOH, dried with sodium sulfate and desolventized. The black residue is distilled under vacuum at 150C (10 mmHg) yielding 40g (66% yield) of light yellow liquid.

NOTE: Its necessary that the precipitate is washed using vacuum filtration, during gravity filtration DCM forms a muddy slurry with the precipitate that is very annoying to deal with. A fair bit of the product is usually stuck to the precipitated aluminum hydroxide in the form of a sticky yellow coating. DCM washings will recover the majority of the ketone.
NOTE 2: The slurry problem could have possibly been entirely avoided by dripping the reaction mixture in ice cold HCl, and then extracting the DCM, this way the aluminum hydroxide formed would have been immediately dissolved before clumping up. This theory was not tested as I was very fed up with this reaction, but if I were to do it again I'd probably try this out.

Grignard reaction:

Under a nitrogen atmosphere 40mL of dry ether and 5g magnesium turnings were placed in a flask, and a few mLs of a solution of 19.5g cyclopentyl bromide in 30mLs of ether were added at once with stirring. To initiate the reaction a few crystals of iodine were added, and the magnesium grinded a bit with a glass rod to expose fresh metal. After a few minutes of violent stirring the iodine color disappears and bubbles start to evolve, at this point the rest of the bromide is added dropwise to ensure a gentle but steady reflux. Complete addition took about 2-3 hours, after which the reaction was let stirring at room temperature for another hour. At this point the nitrogen intake was removed and substituted with a drying tube, and 9g of finely powdered 2-Chlorobenzonitrile mixed with 550mg copper(I) bromide (0.06 eq) were added to the flask, which caused the ether to boil. External cooling could be applied in case it gets too aggressive. The reaction was left refluxing for 3 to 4 hours, then the flask was put in an ice bath and 20mL of cold water dripped in slowly with stirring. Once all the water was added a solution of 20g citric acid in minimal water was introduced at once and everything stirred for 1 hour. On standing the ether layer separates and its collected, washed with water, brine, and dried over sodium sulfate. After filtering, the solvent was gently evaporated under vacuum, and the residue distilled under strong vacuum, the product distills over at 114.4C (2mmHg), yielding 9.65g of 2-chlorophenyl cyclopentyl ketone as a pale yellow liquid in 71% yield. A small prerun of what I assume is dirty starting material distills over at around 60-70C (2mmHg), weighing 1.22g.

NOTE: The use of CuBr catalyst and 2 molar equivalents cyclopentyl bromide to nitrile significantly sped up the reaction time, avoiding the need to do multiple day long reactions. The original reaction, 1:1 bromide to nitrile, uncatalyzed in THF showed unreacted nitrile on TLC even after 3 days, and the isolated product revealed heavily contaminated by impurities, leaving a 26% yield after distillation. For the workup, ammonium chloride and ammonia are in my opinion (i do not have analytical proof of this) insufficient for hydrolyzing the formed ketimine, and that might be part of the reason for the lower ketone yield. On the other hand HCl forms quite a bit of tar, so its best to use a weak acid like citric acid. Also, the use of ether as a solvent is preferred, as it can be removed more easily and in milder conditions, which is good since the impure ketone tends to darken significantly when heated in open air.


ANALYSIS: Eluent 1:10 EtOAc:n-hexane, visualized under 254nm UV

[list type=decimal]
[li]Plate for the reaction of 1:1 bromide to nitrile, uncatalyzed in THF after 3 days. First row is a sample of the impure ketone after solvent removal, second is pure 2-chlorobenzonitrile. Apart from the tar that stays at the bottom, 3 main spots are visible from top to bottom: target ketone, unknown compound (maybe unhydrolyzed ketimine?) and unreacted nitrile. [/li]
[li]Plate for the reaction of 2:1 bromide to nitrile, CuBr catalyst in THF after 3 hours, sample of the impure ketone after solvent removal. A faint nitrile spot is observed, but MUCH fainter than in the previous plate. One big ketone spot at the top and tar at the bottom.[/li]
[li]Plate for the reaction of 2:1 bromide to nitrile, CuBr catalyst in THF after 3 hours, after distillation. First row is the ketone fraction that came over at 114.4C (2mmHg), second row is the fraction coming over at 60-70C (2mmHg), and third row is pure 2-chlorobenzonitrile for reference. Rf value for ketone in this solvent system is 0.93[/li]



CONSIDERATIONS: the two procedures are both similarly yielding messy reactions, and one could be chosen over the other based on personal preference. The friedel-crafts seems to be a good route to analogues (differently substituted acyl chlorides are much easier to get/make than nitriles: DCK with benzoyl chloride, MXE with 3-Meo-benzoyl chloride, ..) and more reliable than the grignard (always pretty much the same yield). On the other hand the grignard is less messy (leaves behind glassware that requires minimal cleaning), the workup is easier and its less dangerous (dont have to deal with liters of HCl gas being produced during reaction AND workup). I personally prefer the grignard, as i dont like to clean glassware, hence why i investigated it more
The production of 2-Chlorophenyl cyclopentyl ketone is the most important step as an impure material will cause great yield losses or straight up hinder the subsequent reactions, so I highly recommend particular care in the purification of this product. The best way to assure maximal purity is column chromatography or careful vacuum distillation, with the latter being the best option for large scale. To guarantee purity in my ketone I fractionally distilled the residue under strong vacuum with gradual heating on an oil bath, collecting the fraction with the exact boiling point (with a boiling range tolerance of 3-4C) and discarding the rest, TLC analysis is a must.

Part 3, rearrangement route

This is the classic route to ketamine, which goes from ketone, to bromide, to imine and then to ketamine with a rearrangement. Various techniques of each step are investigated to determine the overall best. Note, yields are reported but are only indicative, as in all the middle steps little to no purification was done, so the yield counts the weight of some impurities. A more reliable yield will be the total one from ketone to ketamine, reported at the end.



Bromination reaction:
NOTE: The bromoketone is extremely unstable, especially so if the starting ketone material contains substantial impurities, and should be used immediately. Extensive purification is not possible, and solvent must be removed with vacuum or with a stream of dry inert gas. The products color ranges from yellow to brown, anything darker means its decomposed beyond use. Also, impurities in the bromoketone (most of them residual from the ketone) make the product degrade significantly faster.


1) Brominating with HBr and H2O2:
In a rbf 40g of the ketone was added to 48.55g of 48% HBr and hydrogen peroxide solution (30%, 32.64g) was dripped slowly into the mixture with strong stirring. The reaction was done in a waterbath adjusting the drop speed to keep the temperature below 50C, and after all the H2O2 was added the mixture was left reacting for an hour at 50C. Once the solution cooled to room temperature DCM (150 mL) was added with stirring alongside 200 mL of saturated sodium thiosulfate solution to quench the remaining bromine and hydrogen peroxide. After stirring for 10 mins at RT the water layer was extracted with 2x20mL DCM and the pooled organic phases washed 2x50mL sodium sulfite and 1x50mL water. After drying with sodium sulfate and filtering the solvent is removed under vacuum, yielding 47g of bromoketone with a 86% yield.
NOTE: Yields with this reaction are high but very inconsistent, ranging from 60 to 90%. The main reason for this in my opinion is temperature, as I was quite inconsistent with it in my various tests.

2) Brominating with CuBr2:
To 3g of the ketone in 30mL ethyl acetate 8.95g of CuBr2 were added and the mixture was refluxed for 3h. After cooling to room temperature the reaction mixture was filtered, washed with 3x15mL distilled water, 1x15mL brine, dried with sodium sulfate and desolventized under vacuum. 3.7g of brown oil was recovered with an almost quantitative yield.
NOTE: The bromide from this reaction resulted in darker brown than other methods. Reflux temperature is probably too much, if i were to do this again ill opt for milder conditions.

3) Brominating with elemental bromine:
To 3.7g of ketone in 12mL DCM is added 1mL of bromine in 9mL DCM on an ice bath. At the beginning a larger portion is added and stirred until a color change was visible, then the rest was added dropwise. 30 minutes after addition, 20mL of sodium thiosulfate solution is added with stirring, organic layer recovered and evaporated to give bromoketone with a 60% yield.

Imine formation reaction:
NOTE: Pure iminol is a brown solid, but its often obtained as a viscous liquid if impurities are present. The impure material could be rearranged into ketamine without further purification, but it can be recovered as a solid in most cases after A/B treatment. In my experience the iminol is quite stubborn to crystallize out, and if impurities are very abundant (impure ketone starting material for example) they will dissolve the product keeping it liquid no matter what. Ive heard reports of people obtaining whitish powders, Ive never came close to that, but I suspect that is because of column chromatography purified bromoketone, which I never used.

1) Aqueous methylamine:
26.5g of ketone bromide was treated with 400 mL of 15% aqueous methylamine solution and the mixture stirred for 48 hours. A suspended solid was expected, but instead a brown oil separated out, so the reaction mixture was extracted with DCM, the organic phase washed with brine, dried with sodium sulfate and evaporated. After analyzing the product with TLC the oil was deemed as unreacted bromoketone, tar, and a tiny bit of iminol.
NOTE: Ive done and redone this reaction multiple times with the same outcome, which leads me to believe the reaction just doesnt work, considering TLC and the fact that the same bromide batch used with it worked with other methods of imine formation.

2) Methylamine in methanol:
10g of ketone bromide were dissolved in 30mL 40% methylamine in methanol and stirred at -20C for 2 hours, then overnight at room temperature. After evaporating the solvent under vacuum there is added 30mL n-hexane and some anhydrous sodium carbonate, the suspension is then stirred vigorously for an hour, filtered, and stirred again with fresh carbonate for half an hour. After removing the hexane 6.4g of viscous brown liquid is obtained in 77% yield.
NOTE: Even when the starting material is very pure the iminol can be a viscous liquid. In this case the solid is obtained by crashing out the HCl salt out of diethyl ether as a tarry goop, removing the ether, dissolving the goop in minimal water and basifying with NaOH. This way brown solid flakes precipitate in water. As I mentioned above however, the liquid can be rearranged without further purification, which is more convenient in my opinion.


3) Liquid methylamine:
To 3.5mL of liquid methylamine 2g of bromoketone were carefully added, and everything was stirred at -20C for an hour. After that the flask was connected to a HCl trap, and the ice bath was allowed to return to room temperature. The residue was diluted in n-hexane, stirred with anhydrous sodium carbonate for a while, filtered and desolventized, yielding 1.38g of iminol with a 83% yield. The dark imine HCl precipitates slowly from diethyl ether after gassing as needle-like crystals, this is done directly in the flask where the thermal rearrangement will be done.
NOTE: Methylamine is easily liquefied with an eutectic ice bath without recurring to dry ice, which could be handy though for a larger scale. To liquefy the gas coming from a MeNH3+Cl-/NaOH generator a NaCl/ice bath (1:3 ratio) reaching -20C would suffice, the better option is a CaCl2x6H2O/ice (0.8:1) inside a Dewar, which are able to reach -40C and will comfortably stay at very low temperatures for hours. A thermos could be used instead of a Dewar vase.


Thermal rearrangement:
24g iminol were added in a flask and neutralized with 2M HCl in diethyl ether and evaporated to dryness. 200mL ethyl benzoate was added alongside 770mg of AlCl3 catalyst and the mixture was stirred at 125C for 5 hours. Once cooled to room temperature 200mL hexane and 50mL 3M HCl are added with stirring, the mixture filtered through cotton and the organic layer washed with 3x50mL 3M HCl. Pooled aqueous extracts were washed with 3x25mL DCM. The extract was basified with concentrated NaOH solution and filtered to yield 7g of dirty ketamine freebase after drying, which is precipitated as the HCl salt in diethyl ether. Ketamine HCl is then suspended in 70mL acetone and refluxed for 30 mins, after which 10mL of water was added until all the solids dissolved, the solution is then slowly cooled to 0C. 7.4g of ketamine HCl crystals are recovered, with a freebase yield of 27%.
NOTE: After adding the HCl a lot of tar separates out, which makes the two phases impossible to separate. This is easily resolved by filtering everything through cotton before proceeding, using 2 funnels or changing the cotton piece halfway will speed things up significantly as the tar is very slimy and will clog the funnel.


ANALYSIS: Eluent 1:10 EtOAc:n-hexane, visualized under 254nm UV

[list type=decimal]
[li]2-Chlorophenyl cyclopentyl ketone and the bromoketone compared. This solvent system is not the best for spotting the bromoketone, as it basically travels with the solvent up to the very top. This way its very difficult, if not impossible, to tell how much ketone there is still present. The only indication is the amount of tar present, this sample is fairly clean, but with dirtier bromoketone a brown tar spot is visible at the bottom even with the naked eye.[/li]
[li]Iminol plate, comparing one that Im sure its the right substance (first row) and the one from the problematic imination in water. The sample from the 15% methylamine in water is bright neon under UV for some reason, and shows a huge spot at the bottom and at the top, tar and bromoketone possibly. Both plates showed 2 spots, one upper one with Rf = 0.79 and one lower with Rf = 0.17. The lower spot is much more visible in the aqueous run, and Im not sure which is the right one, but I suspect the upper one, as it has a value closer to the one of ketamine. In any case, this shows a major contaminant present in every imine formation reaction that followed the product through the workup, which might be the cause to the brown color, the difficult crystallization and the tar in the rearrangement. The two spots are very well separated on TLC though, which tells me column chromatography would be very effective at purifying the compound.[/li]
[li]Ketamine freebase before and after extensive purification in acetone. Rf is 0.77[/li]

CONSIDERATIONS: The most appealing aspect of this route is how incredibly OTC everything is; with a little bit of patience the whole thing can be done from scratch from hardware store chemicals, and the overall process is very cheap. For bromination, the best way seems to be only one I didnt try, which is NBS, as mild bromination conditions are necessary for yield and purity of the fragile bromoketone. Second to that Im unsure between CuBr2 and HBr/H2O2, the first one should be the most selective out of the bunch, yielding 99% to quantitative in literature, and the second being the cheapest and easiest way. With the HBr/H2O2 system one wouldnt need to make the expensive CuBr2. For the imine formation, 15% methylamine in water was sadly inconclusive in my experiments, which would have been the best candidate if it worked as promised. The actual best is methylamine gas dissolved in an appropriate solvent (methanol in my case, benzene is also used) as working with anhydrous liquid methylamine was unpleasant enough in the 2g scale, I dont want to imagine the horrors of scaling up. For the thermal rearrangement I opted for a milder temperature alternative vs the classic 200C reflux, which is accomplished with a lewis acid catalyst like AlCl3 in ethyl benzoate. Ethyl benzoate is a very good solvent as it is inert, high boiling and able to dissolve the tar. This way the reaction proceeds smoothly without clogging, its also easily made from ethanol and benzoic acid. I originally wanted to report the classic reaction too, which is 30 mins at 175-195C in o-dichlorobenzene or undecane but by the end of the route I was so done with it that I decided against it. The thermal rearrangement at 125C produced enough tar to color the solvent pitch black by the end, I read reports on hyperlab that mention solid carbon precipitating out at 195C. Part of the reason for the tar and the low yield (27% best run) is in my opinion cumulative impurities in all the previous steps, particularly the imine formation reaction which produced brown solids no matter how much I tried to clean them. This could not only mess with the reaction conditions, but inflate the yields in the previous steps. What if that 80% imine yield is in reality 40% pure imine? Maybe the lower yielding reaction is that, and not the rearrangement. For anyone which decides this route is the most suited for them, I would invest in some column chromatography.
I mentioned how cheap and OTC everything is, but the other side of the coin is how messy and fragile each step is. I personally really dislike the thermal rearrangement route, maybe Im biased because out of the year it took me to do all this work 80% of the time was figuring out these cursed 3 reactions, but still. The bromoketone is so fragile that purification is very very limited: distilling it is impossible, heating too much the solvent its contained in will cause it to darken and adding sodium bicarbonate to quench the HBr caused it to darken. The iminol freebase is quite fragile too, i tried to evaporate the hexane it was in at 60C atmospheric pressure and it turned from dark yellow to black. The whole reaction is generally unpleasant due to the massive excess of methylamine used, which smell never really goes away from the product. Both of these reactions are a surprise, you never really know how well it worked until the rearrangement itself.

Part 4, autoxidation ring expansion route

THANK YOU FOR YOUR HELP, Hope to do something useful for the community

This route is a fairly new publication that piqued my interest due to the high yields achieved and simplicity.


Autoxidation ring expansion:
20g of 2-Chlorophenyl cyclopentyl ketone are placed with 23.92g triethyl phosphite in a flask inside a warm water bath. With strong stirring, a solution of 5.4g KOH in 200mL anhydrous ethanol is added, the flask is then purged with oxygen and left stirring at 30C with an oxygen balloon attached for 48 hours. The reaction mixture is then reduced under vacuum until the majority of the ethanol is removed, then 600mL of cold water are added and everything is stirred for 1h. After standing overnight in the refrigerator the precipitated yellow solid is collected by filtration, washed multiple times with cold water and dried in a vacuum desiccator. In total 16.2g of hexanone are collected, corresponding to a yield of 75%.

NOTE: The reaction can be done in a few hours by using a 1:1 mixture of DMSO:ethanol as the solvent, but in this case the reaction turns dark quickly and the DMSO is very difficult to remove completely from the product. Heating accelerates the reaction (at 50C is done in 24 hours), but milder conditions are higher yielding and produce a cleaner product.
The course of the reaction is easily followed even without TLC:

At about 12 hours the color switches from light yellow to orange, then it becomes cloudy after a day, and on day 2 solid starts to crash out. The oxygen balloon also shrinks visibly on the first day, and by day 2 the strong smell of triethyl phosphite is basically undetectable.

Mesyl chloride amination:
A flask inside an ice bath was charged with 16.2g of hexanone, 120mL THF and 25.6g triethylamine. A nitrogen balloon was attached, and a solution of 24.8g mesyl chloride in 20mL THF was added dropwise at such rate to maintain the temperature between -5C and 0C. After finishing the addition the mixture was stirred for an additional hour, after which 85mL of 40% methylamine in methanol was added at once, and the mixture stirred for 4 hours at room temperature. Most of the methanol and THF is removed under vacuum, and to the cloudy suspension is added 100mL of water + 100mL of DCM. The water layer is extracted with 2x20mL DCM and discarded, while the organic phases combined and washed several times with water. The solution is then extracted with 3x30mL 1M HCl, which is washed with 2x20mL DCM and basified with a concentrated NaOH solution until precipitates appears. The freebase is then filtered and dried under vacuum, yielding 9.91g of ketamine freebase, corresponding to a yield of 78%. Part of this ketamine is kept as freebase to resolve the racemic mixture later, the rest is dissolved in diethyl ether and the HCl salt precipitated by passing dry HCl gas into the solution.

NOTE: The addition of mesyl chloride is very exothermic and good temperature control is key to achieve high yields in this reaction. Letting the reaction heat too much in combination with no nitrogen atmosphere in another run basically halved the final yield.
Also, the paper used methylamine as an ethanolic solution, I opted for methanol as I had it anhydrous on hand. The yield loss (up to 90% in the paper) is probably due to this, since the rest of the procedure was followed exactly.

ANALYSIS: Eluent 1:10 EtOAc:n-hexane, visualized under 254nm UV

Row 1 corresponds to hexanone, row 2 is reaction mixture after mesyl chloride addition (sulfonate intermediate maybe?) and row 3 is ketamine freebase. The hexanone is basically immobile in this solvent system, I tried up to 1:1 ethyl acetate:n-hexane but it still wasnt moving much. For the analysis of the first reaction I just observed that the ketone starting material disappeared completely, but since the product isnt eluting properly I can't tell if there are significant impurities present. Ketamine freebase after filtration shows no contamination apart from a tiny bit of tar that stays at the bottom, which stays in the ether as a brown goop once the HCl salt is formed.
Rf for hexanone is 0-0.096 (very close to origin), Rf for ketamine freebase is 0.77

CONSIDERATIONS: I was intrigued by this pathway the moment I saw it on paper, and the fact that no one ever tried it (and posted online, as far as I know) in a home lab setting made it even more appealing to me. The oxidative ring expansion consists, in a nutshell, in mixing the chemicals and waiting 2 days, which makes it reliably high yielding; if the reagents are pure, the oxygen is present and the heating is spot on nothing can really take an unexpected turn. The reaction is easily followed by observation and the workup is trivial, consisting in boiling off the solvent in vacuum and dumping everything in cold water. One thing to be aware of is that this reaction straight up doesnt work if the ketone is highly contaminated, nothing happens in 2 days and after adding water a milky suspension forms with nothing settling in. Attempts to extract this milky water with solvents resulted in dense polymer like goop, and after trying to crystallizing out solids from boiling hexane a one digit yield is obtained, suggesting that (speculations) large amounts of impurities may react with the reactive peroxide intermediate before it has a chance of being reduced by the triethyl phosphite. The substitution reaction with mesyl chloride is a bit more tricky, but once the right conditions are met everything goes smoothly and the product is recovered with a standard A/B extraction. The procedure from start to finish is generally higher yielding (58% overall yield from ketone, possibly 75% if matching the paper vs just 20% in the thermal rearrangement route) and MUCH cleaner, which is obvious comparing the picture of the obtained ketamine freebase after workup; the one from this route barely requires anymore purification, while the other needs to be refluxed in acetone to get rid of the brown color.
The only flaw this pathway has, which is a major one for many, is the exotic chemicals it uses and the worryingly high toxicity of mesyl chloride. Alternative reducing agents to triethyl phosphite such as sodium sulfite and sodium phosphite were considered in the paper, but those yielded only 34% and 25% respectively. Triethyl phosphite is not the biggest problem though, as it can be made (relatively) easily from PCl3 and ethanol with TEA catalyst, the real problem is mesyl chloride, considering its synthesized from methane gas, oleum and thionyl chloride which is a mixture no sane person would attempt in their house. P-toluenesulfonyl chloride was considered in the paper, but didnt work due to steric hindrance, and no other experiments were made. I wonder if the alcohol could be substituted by the amine by treatment with PCl3 or thionyl chloride, which should convert it to a more reactive halogenated intermediate. Mesyl chloride is extremely toxic, and it was handled with thick neoprene gloves over the nitrile ones, in a hood and transferred with a long needle syringe directly into the THF in the dropping funnel. Its not the worst, but definitely not something the inexperienced chemist should handle, but I would argue that nothing written here should be attempted by a beginner.

Part 5, chiral resolution and isomer recycle

This is a more optional part, but the work didnt feel complete without addressing the chirality part.

Separating the isomers:
To a flask there is added 4g of ketamine freebase, 1.1g L-tartaric acid, 40mL acetone and 2.7mL water, and the mixture was refluxed for 30 minutes until clear. The mixture is then slowly cooled to 0C and the (S)-ketamine tartrate precipitates isolated by filtration. The solid was treated with 1M NaOH solution, filtered, washed with water and recrystallized as the (S)-Ketamine HCl salt in diethyl ether. The (R) isomer is extracted from the acetone by reducing to dryness under vacuum, the HCl salt is formed as previously described, yield for the two isomers is basically quantitative.

Racemization of unwanted isomer:
If one isomer is of particular interest, the other can be racemized by rearrangement in ethyl benzoate. 1g of the (R) isomer is dissolved in 10mL ethyl benzoate and stirred with 100mg AlCl3 at 150C for 24 hours, after which the solid HCl is recovered by diluting with n-hexane, extracting with 3M HCl, basifying and forming the racemic HCl salt.
NOTE: The process is MUCH less tarry than the thermal rearrangement for iminol -> ketamine, which confirms my suspicions that the low yield on the thermal rearrangement has something to do with the purity of the iminol

Well done!
Very thorough; I haven't read it all yet though

Very good synthesis, I might have to try this after I get done with 4-MMC

@Raid If you are willing to discuss 4-MMC and related compounds (of course in a purely theoretical manner), you might be interested in this paper (https://pubs.acs.org/doi/10.1021/ja4096472) using acyl arenes and primary or secondary amines to form the a-aminoketones. And they used propiophenone with good results. It avoids the use of Br2 and the highest yielding substrates found, happened to be the cycloalkylamine analogues of the a- ketones; the 6-membered ring analogues are (for now) unrestricted in the US. Getting ahold of propionyl chloride for the substituted propiophenones however, is a different matter.



[Edited on 4-8-2023 by dettoo456]