.

A 500mL three-necked round flask equipped with a reflux condenser, internal thermometer, pressure-equalised addition funnel and a large egg-shaped magnetic stirbar was charged with 25% sodium methoxide in methanol 31.55g (Aldrich; 146mmol) and methanol 100mL. The flask was placed on ice slush bath and after 15 min a solution of 1,3-acetonedicarboxylic acid dimethyl ester 25.00g (Acros; 143.55 mmol) in methanol 10mL was added within 15 min, the addition funnel was washed with methanol (2×20mL) and the washings were also added into the mix. The cooling bath was then removed and the flask was placed on a 65C oil bath and stirred for approximately 30 min. (The mixture gradually became homogeneous as the precipitated Na-enolate salt of the di-Me-acetonedicarboxylate re-dissolved with heating). When the internal temperature in the flask has stabilized, a mixture of 40% aqueous glyoxal 12.00g (Alfa; 82.7 mmol, 115% of the theoretic amount) with methanol 30mL was introduced dropwise from the addition funnel – very slowly – over a period of 1h45min, with a vigorous stirring on the 65C oil bath. After the complete addition the funnel was washed with methanol (10mL) and the washings were also added to the mix. The resulting cloudy reaction mixture was stirred for extra 15 min at 65C, then diluted with THF 200mL and the flask was removed from the heating bath. The mixture was stirred at RT overnight (12 hours). The precipitated intermediate (as a disodium salt hydrate) was collected by filtration using a large sintered-glass Buchner funnel. The collected solids were washed thoroughly with THF and then dried by suction for about 2 hours.

This intermediate salt (a cream-colored heavy powder, 27.92g; 90%Y) was dissolved in water 400mL in a 1L flask. 37% concentrated HCl 46 mL was added dropwise with a vigorous stirring (as to limit the formation of dumplings) and the resulting heterogeneous mixture was placed on a 100C oil bath. The mixture was stirred at reflux at 100-120C for 1 hour and at 120C for additional 2 hours – during this time the mixture became homogeneous as the gummy deposits gradually dissolved. The flask was then removed from the heating bath, a large spoon of activated charcoal was added into the stirred mix, the charcoal was removed by filtration while warm (the charcoal was washed with additional water) and sodium chloride 100g was added to the combined filtrates. The mixture was stirred on ambient bath until the complete salt dissolution  (5 min). This mixture was then extracted three times with dichloromethane (3×250mL), the organic extracts were washed with saturated aq. NaHCO3 200mL. The combined extracts were dried with magnesium sulfate and evaporated to dryness from ambient water bath. The obtained crystalline residue was dried on highvac for about 30 min.

Y=7.790g of a white crystalline solid, pure by NMR (78.5% overall from di-Me acetonedicarboxylate) .

1H(CDCl3, 400MHz): 3.048(m, 2H), 2.585(ddd, 19.5Hz, 8.7Hz, 1.8Hz, 4H), 2.156(dd, 19.5Hz, 5.2Hz, 4H)

Note 1: The product is also available commercially [Aldrich 5g/$400]

Note 2: A very slow addition of the glyoxal solution and a careful control of the reaction temperature (65C) by the oil bath during the first step is required for a good yield. The reaction is not very sensitive to moisture so a common-grade MeOH was used from a freshly-opened bottle. (The reflux in the first step was done under Ar but this may be unnecessary). The final product can be re-crystallized from MeOH;  in this preparation NMR-uniform material was obtained directly by evaporating the DCM extracts and drying the residue briefly in vacuo.

Note 3: This preparation was based on a large-scale (1.5 mol) procedure from OrgSyn (Vol 64, p.27, 1986). The medium-scale (140mmol) experiment described here was run in higher dilution, on a stirplate and with the oil bath inplace of a heating mantle. Also the hydrolysis step was simplified at this medium scale, etc – these modifications probably helped to improve the product yield and purity.

Note 4: This preparation provided 80% overall yield when run on twice as large scale (1L flask, 50.1g of di-Me-acetondicarboxylate, 300mL MeOH, 63.2g of 25% NaOMe, 24.25g of 40% glyoxal in 50mL of MeOH, 94mL of conc. HCl). Few minor changes: acetondicarboxylate was added neat by syringe, quite fast (over 10 min at 0C) as there is not much exotherm during the additon. In the second step, the intermediate salt (57.5g) was dissolved first in hot water (800mL) and the solution was placed on oil bath (120C) and conc. HCl (94mL) was added at approx 80C internal temperature with intense stirring, and the resulting emulsion was then stirred at reflux on 115-120C oil bath for additional 150 min. In this way the mix is easier to stir magnetically (as the formation of sticky dumplings is completely prevented).

Evaporating aqueous reaction mixture is a lamentable job –  by weight water has one of the highest evaporation enthalpy values – but with a good rotovap and enough persistence one can even take care of several liters of aqueous mix (if there is no better alternative). The one thing that can turn this into the most frustrating experience is foaming.

I was struggling today; a published homotropanone prep calls for freezing + lyophilizing the entire reaction mix. I scaled that thing by a factor of four and since I did not want to lyophilize a half-liter of the reaction mix, I just put it on the rotavap and suddenly the reasons for the recommended lyophilization became painfully clear…

Desperate people would add n-octanol or even couple of drops of silicone oil to their mix but I did not want to introduce non-volatile impurities into the product. I was dreaming about silanizing the flask glass surface instead (a rinse with Me2SiCl2 and tributylamine in dichloroethane does it) but in the end I just poured 1mL of of hexamethyl disilazane (TMS)2NH straight into my aqueous mixture  - and the foaming ceased like a miracle. It must have been the silicone film on the glass that produced this remarkable effect because when I later transferred the solution into another flask it started foaming crazy anew; and a little more (TMS)2NH and it was calm like a lamb again.

1-(2′-amino-1′-ethyl)-cyclohexene 17.53g (140 mmol, Alfa) was combined with water 60 mL and conc. 37% aq. HCl 11.5mL (140 mmol) was added, followed by additional water 40mL. The acidity of the mixture was adjusted approximately to pH=2 (on strip indicator paper) by adding few drops of 6M HCl. The mixture was placed on a 40C heating bath. A solution prepared by diluting 11.50g of 37% aqueous formaldehyde (140 mmol, Aldrich, stabilized with MeOH) with water 20mL was slowly added from the addition funnel with vigorous stirring, over a two-hour period. The addition funnel was washed with additional water (2×10mL) and the washings were also added to the mix. After the complete addition, the stirring at 40C was continued for additional 6 hours. Two small spoons of charcoal were then added, the reaction mixture was filtered into a 1L round flask (the charcoal was washed with additional water) and the combined filtrates were evaporated from a 40C bath at 15-20 Torr to dryness. The obtained solid residue was re-dissolved in methanol 80mL at reflux. Boiling acetone 450mL was added into the flask at once and the resulting slurry was allowed to rest at ambient temperature for 6 hours (overnight). The precipitated product was collected by filtration, washed with acetone and dried in vacuo. Y=23.722g (88.5%) of a white crystalline solid.

1H(D2O, 400MHz): 3.357(dd, 12.9Hz, 3.6Hz, 1H), 2.244(td, t:14.2Hz, d:4.0Hz, 1H), 3.159(m, 1H), 3.019(dd, 13.)Hz, 4.7Hz, 1H), 2.115(br m, 1H), 1.763(m, 2H), 1.672-1.258(m, 8H); 13C(D2O, 100MHz): 68.32, 44.75, 40.41(br s), 39.08, 25.96(br s, 2C), 23.45(br s, 2C), 22.26(br s)

Note: The reaction mix turns into emulsion with the formaldehyde addition but becomes homogeneous further on. It is important to add exact amount of formaldehyde (weighed out in a syringe) and use a solution of known concentration; old bottle of poorly-stabilized formaldehyde solution with paraformaldehyde deposits should be left for the biologists.

I have been doing some asymmetric hydrogenations recently. The pieces that I was making were carboxylic acids and since the highest optical purity that I could get from my system was about 90%ee and the product couldn’t be enriched by a simple recrystallization, I needed to make a salt  - preferably with some optically pure amine – to bring the material up to 98-99% ee.  I got quite lucky with chloramphenicol base: the salt enrichment in a single recrystallization proceeded with high recovery. 

I also tried few other chiral amines, norephedrine worked but it was not very efficient (with the salt being too soluble). And norephedrine is rather expensive and it is on the controlled precursor list – its delivery got halted for nearly a week by the bureaucracy and eventually we had to fax a statement to the supplier, declaring that we are not making a dope. 

This got me to realize that unlike with the chiral pool of acids, there is only a limited choice of inexpensive optically pure amines that one can buy both enantiomers of (a typical problem with many alkaloids and other natural-product derived amines). Chloramphenicol is a generic antibiotic made by old-fashioned racemic synthesis and resolution hence both enantiomers are available. And not every amine is likely to provide a nice crystalline salt; what probably makes chloramphenicol base effective is the combination of para-subst nitrophenyl group and the two hydroxy groups that contribute to ordered interactions in the crystal structure. 

Chloramphenicol base has been used as a resolving agent in industry: Roussel Uclaf, the company that commercionalized a large-scale production of chloramphenicol many years ago, is using the base as a resolving agent for racemic trans chrysantemic acid – a precursor for making synthetic pyrethroid insecticides such as Deltamethrin. The resolution takes place very early in the sequence. Given that pyrethroids are made on ton scale, the availability of the optically pure amine needs to be pretty good.

The actual resolution of racemic chloramphenicol base is noteworthy: Chloramphenicol base is among the rare (1-2%) compounds that crystallize as a conglomerate of optically pure crystals. This means that it can be resolved by selective crystallization based on seeding with one and then the other enantiomer. In reality the process is somewhat tricky – one has to work under carefully controlled conditions and use a massive quantity of the enantiomerically-pure seeding material to have a fast growth, and stop the crystallization very early to ensure that the other enantiomer keeps in the super-saturated solution. But this sort of process is doable in a chemical plant and “resolving stuff with nothing” helps to produce the optically pure material cheaply .

I would like to direct the readers here to the excellent project-and-career story from Bruce Maryanoff in the most recent J. Med. Chem ASAP. It is very illuminating on how the drug discovery and development works, and it describes in some detail what a bright chemist can hope to achieve in this profession -with the necessary motivation and a decent employer (and tremendous amounts of luck).  

It is also an illuminating story on how the process does not work. For example, the currently most popular, target-driven rational-design-based approach can be pretty futile in CNS drug projects . The author also suggests that the management mantra about focusing on the discovery of the next “blockbuster drug” actually bankrupts the industry – financially and scientifically; his drug Topiramate (which has been making 2 billions a year for the company) would have not been discovered or developed under the management methods currently prevalent in the industry. Few things stand out: 1) It seems that having a blind luck and testing the compounds in a realistic animal model is more important than having a correct mechanistic understanding how the drug candidate actually works.  2) Few independent-minded individuals in their pharmacology and chemistry have made a good use of their lucky break. They stubbornly kept the research program going – even as their managers were lukewarm and would not support the compound development for a long time. It also goes to the credit of the management that allowed their researchers to pursue this as their hobby. The story shows that the progress in pharma research does not really happen by imposing some management-theory-derived reporting structure on the research department, by drafting the flowcharts and aligning the teams. For medchem research to succeed, the projects should be allowed to self-organize around the bright individuals rather than being planed out from top down, with red tape and micromanagement.

In this context it is entertaining to read rather disingenuous remarks made by the Merck chief strategy officer Merv Turner at the pharma management conference. He explained that they are currently sacking lots of people in research because  “Seventy-five cents of every dollar we spend on R&D goes to fund failure” and “the future results must come at a lower cost”.  

The actual drug discovery cost makes only few percent of the final drug development cost. By far the most expensive part is the clinical trials and namelly the late-stage clinical trials. What the Merck management poseurs do not tell in public is that it was the Merck top management decisions that cemented their company’s commitment to these “the next blockbuster” projects –  which eventually led to a string of stunningly expensive late-stage failures. When the top executives receive massive stock option bonuses, they become mercenaries of the stock prices. Their wishful thinking baloney percolates from top down through the management layers, etc.

There are many parallels between the state of pharma industry and the recent financial sector collapse, and it is always the executives who run their companies to the ground that are rewarding themselves most obscenely. Remember this whenever the pharma companies claim that the freedom to price their drugs is essential for the innovation.

_________________________________________________________

Update:  Here are additional two great articles from Bruce Maryanoff on the subject

A solution of 4-bromo-2-fluoronitrobenzene 1.300 g (5.909 mmol) and 1,3-propanediol 5.0mL (69mmol) in anhydrous dioxane 20mL in a 100mL 14/20 joint flask was placed on ambient water bath and a solution of sodium tert-butoxide 1.00 g (10.8 mmol) in anh dioxane 40mL was added dropwise from an addition funnel over 15 min with vigorous stirring. The resulting deep-orange reaction mix was stirred under Ar for additional 2 hours. A solution of potassium carbonate 2.80 g (20mmol) in water 30 mL was added into the flask and the mixture was deoxygenated by argon sparge (long needle through the septa) on ultrasonic bath for 15 min. A solid mix of pinacolate ester of 4-pyrazolylboronic acid 1.50 g (7.73mmol) and Pd(PPh3)4 600mg (0.519mmol; 8.8mol%) was added in one portion and the mixture was sparge-deoxygentaed with Ar for additional 10 min on ultrasonic bath, then vigorously stirred under Ar on a 95C oil bath for 150 minutes. The reaction mix was cooled and portioned between saturated aqueous ammonium chloride 200mL and ethyl acetate 200mL. The aqueos phase was re-extracted with EtOAc (200mL). The organic extracts were washed with additional sat ammonium chloride (200mL) and then combined, dried (MgSO4) and evaporated. The obtained solid yellow residue was suspended in benzene 20mL, heated briefly to reflux and then allowed to sit at RT for 1 hour. The precipitated product (1.248g) was collected by filtration, washed with additional benzene (2×10mL) and dried on highvac. Concentrating the supernatants and re-crystallizing the oily residue from benzene (5mL) overnight provided additional small fraction of the product (47mg, 95% pure). The combined yield was 1.295g (83% th) of a pale yellow solid, >98% by HPLC.

1H(d6-DMSO, 400MHz): 13.165(br s, 1H), 8.455(s, 1H), 8.124(s, 1H), 7.891(d, 8.5Hz, 1H), 7.529(d, 1.5Hz, 1H), 7.344(dd, 8.5Hz, 1.5Hz, 1H), 4.577(t, 5.1Hz, 1H), 4.301(t, 6.1Hz, 2H), 3.594(q, 5.4Hz, 2H), 1.899(quint, 6.2Hz, 2H); LC/MS(+ESI): 264(M+1)

The aldehyde 800mg (2.565 mmol) slurry in dichloromethane (not anhydrous, 6mL) was placed on ambient water bath, Deoxyfluor 0.80mL (1.7 equiv.) was added in one portion and the mix was stirred under Ar for 150 min, then quenched by addition of few spoons of crushed ice.  The reaction mix was portioned between ether 120mL and water 120mL, separated and the aqueous phase was re-extracted with ether 120mL. The organic extracts were washed with aqueous sat. NaHCO3 120mL, combined, dried (MgSO4) and evaporated. The crude product was purified on a column of silica (40g) in hexane-dichloromethane 1:1 mix, isocratic. Y=672mg of pearl-shiny white soft flakes. (78.5% th)

1H(d6-DMSO, 400MHz): 8.464 (s, 1H), 8.386(s, 1H), 7.072 (t, 53.5Hz, 1H); 19F(d6-DMSO, 376.5Mhz): -115.74 (d, 53.7 Hz, 1F)

The starting aldehyde preparation was described here on July14 2008.

Note: This kind of reactions is often initiated by one drop of ethanol but here common-grade DCM serves equally well. Deoxyfluor is a more stable alternative of the popular DAST reagent –  and it is just as nasty (HF burns). Gloves, sash down,  larger-scale experiments should be quenched by pouring the mix on ice as the quench is very  exothermic.  The aqueous washings contain HF so one may want to use gloves during the work-up as well.

When I was in high school, I got free run in a chemistry lab that belonged to a youth center. I was trying to synthesize papaverine, and this was completely above my ability (and the lab resources) but I was very persistent. The key building blocks for papaverine are homoveratric acid and homoveratryl amine, and I set out to make them by myself: I had several bottles of catechol to start from - my problem was how to methylate it, I could not just buy stuff – I had to go by with what was in the stockroom.

First I made lots of methyl iodide, from red P + iodine and methanol – refluxing and distilling it on the bench so I know how methyl iodide smells - but the methylation was messy. Next I tried to make dimethyl sulfate from sulfuryl chloride and methoxide in situ, and it worked to some degree – the vanilla smell of guaiacol was everywhere – but again I could not isolate anything from the mess. So my next idea was to use diazomethane made from nitrosomethyl urea.

So I was cooking and then distilling AcNH2 on a grand scale, from AcOH and urea - and this went quite well (apart from the all-pervasive mice urine-like smell of acetamide) and then I was to carry out the Hoffman degradation.  When one uses concentrated aq KOH and bromine, the in-situ generated MeNCO reacts with a second molecule of acetamide to produce MeNHCONHAc, an intermediate for preparing nitroso methylurea; I needed it for a large-scale diazomethane reaction so I did it on a mol scale on the first run.

The procedure called for Br2 to be added into AcNH2 + 30% aq KOH mix in a 1L flask, then gently heating the mix until a rapid gas evolution commenced. Since I scaled up the preparation by a factor of twenty on the first run, and I did not have a 20L flask, I used the biggest flask I could find, a 4L Erlenmeyer, and loaded the stuff up; it all fit in there. But then, the mix did not wait to be gently heated and instead jumped out at me all at once.

I usually did my work without glasses, on the bench - but this one time I put goggles on and it was well worth it. The hot KOH + KOBr solution rained all over the place and bleached my hair blond; also my T-shirt and jeans ended up with white vertical stripes. A colleague stood nearby and saw the whole thing and dragged me into shower.

Eventually I did make some nitrosomethyl urea and diazomethane but never finished the papaverine project. No explosion, poisoning or other injuriy happened during all these crazy experiments. But there was another like-minded highschool kid, repeatedly working on some chemistry involving acetone cyanohydrine, and he was making it from acetone and NaCN, in our little fume hood with a lousy fan-driven exhaust. I think he never finished his project either but I remember once we were going down the stairway and he was saying “I don’t know if this exhaust really works ’cause I was smelling hydrogen cyanide this time a lot” and then we got outside there were two dead pigeons on the grass and we looked up and the exhaust from our fume hood was looming right above us…

Ψ*Ψ has a new post on undergrad lab disasters, I would like to add few more of my own making, from two decades ago:

In the junk-room of our chemistry department I found an ancient belt-driven vacuum pump that operated on 380V three-phase AC. I brought it from the basement into our freshly-renovated lab, put it on the bench, plugged it in to see if it worked – and it did. Unfortunately the 380V AC wiring must have been wrong in our lab (we never used 380V plug in there before) or the phase order in the pump itself was switched. At any rate, the motor started spinning backward and the pump pumped out its oil from the inlet hole in eye-blink -  a gallon of black muck that hasn’t been changed for eons. The intlet had a short piece of rubber hose on and it worked like a nozzle - directing the stream of goodness at the high ceiling, right in the middle of the room whence it rained down all over the place. This mishap actually shut down our lab for two weeks as the oil-soaked plaster had to be knocked off down to the brick and concrete in order to make the new plaster stick.   

Some months later I was working as a guest student in another lab (in another building, at another school) – and they had a gas-powered water heaters installed above the sinks because their building lacked central hot water. It was a strange and dangerous thing to have in the lab (right next to the  organic solvent bottles); and the last person in the lab always made sure to turn off the pilot lights before leaving.  One early morning I was washing my hands and water was coming out freezing cold - the  pilot light was off -  so I grabbed matches and without turning off the running water (and the stream of gas), l lit the pilot.  A yellow fireball shot up and the casing flew off from the heater infront of me, with an impressive bang. When my ears stopped ringing  I could hear a calm voice from the opposite corner of the lab where a colleague sat her desk: “Mr. Borivoj, I promise that the next time I’ll pour my coffee over you.”    

The same colleague a week later decided to clean up and inventorize all her glassware - she emptied her drawers and put it all on a long bench. Meanwhile I was distilling 0.5L of old and nasty-looking  N-methylmorpholine to which I added lots of calcium hydride - and when I finished I had CaH2-rich leftover sludge in the distillation flask. I was asking around if it was OK to quench it with ethanol (I had quenched NaH and BuLi before) and a faculty dude said I should go right ahead; I did not realize they probably did not work much with CaH2 in that lab.  So I was carefully adding ethanol with cooling, the bubbles were coming out, calm and nice.  One hour later (back from lunch) I added some more ethanol - no bubbling anymore – so I was certain I could pour that into the waste, and I went to the sink to dilute the sludge with water a bit so that I could pour it out. Suddenly it became clear that ethanol does not really quench CaH2, and water does. The  mud volcano in my hands erupted away, spewing the hot lime and fishy amine on the nearest bench - all over that clean and inventoried glassware. The owner again took it  calmly - she just muttered ”We have to assign you a working space in the hallway”…    

 A mixture of anh. acetonitrile 12mL and isopropyl alcohol 1.4mL (18mmol) in a 25mL flask was cooled to 0C and N,N,N’,N’-tetramethylguanidine 1.9 mL (15mmol) was added under Ar, followed by azidotrimethyl silane 2.0 mL (15mmol), dropwise over 5 min – an exothermic reaction. The resulting solution of TMG-azide was warmed up to RT and then transferred by syringe to 2-chloro-4-iodo-5-trifluoromethylpyridine 3.075g (10 mmol) solid. The mixture was stirred under Ar on a 40C bath for 6 hours, then cooled to ambient temperature and portioned between half-saturated NaHCO3 solution 70mL and ether 80mL. The aqueous phase was re-extracted with ether 80mL. The organic extracts were washed with additional half-saturated bicarbonate 70mL, combined, dried with magnesium sulfate and filtered. The obtained solution of crude azidopyridine was placed on ambient water bath, water 1mL was added followed by 1M trimethylphosphine solution in toluene 11mL (11 mmol; exothermic reaction). The mixture was stirred at RT for 30 min, then washed twice with water (2×250mL) to remove Me3PO, the aqueous phases were re-extracted with ether (250mL). The combined extracts were dried (MgSO4) and evaporated. The residue (2.0g) was re-crystallized from cyclohexane 120mL to provide 1.517g of a 95%-pure product as white feathers (77% overal).

1H(d6-DMSO, 400MHz): 8.170(s, 1H), 6.932(br s, 2H), 6.758(s, 1H); 19F(d6-DMSO, 376.5Mhz): -61.93 (s, 3F);

Note: The neat crude monoazido intermediate can be isolated as a colorless oil (and separated from a crystalline side-product, posibly derived from bis-azido pyridine) but 4-pyridylazides are rather unstable and reducing them in diluted state without further purification is a safer alternative. TMS-N3 is skin and gloves-permeable and causes unpleasant acute azide poisoning (= nasty headaches). PMe3 has a pungent, obnoxious odor.   

This procedure uses organic-soluble azide salt. TMG.HN3 can be isolated and stored; it is an extremely-hygroscopic crystalline solid (Org Prep Daily September 25, 2006). The preparation of the starting iodo pyridine was also discussed here, in Strange bits from Schlosser, on June 7, 2008.