Synlett 2020; 31(19): 1888-1893
DOI: 10.1055/s-0039-1690884
cluster
© Georg Thieme Verlag Stuttgart · New York

Continuous-Flow Synthesis of Tramadol from Cyclohexanone

Timothy M. Monos
a   Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA   Email: tfj@mit.edu
,
Jonathan N. Jaworski
a   Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA   Email: tfj@mit.edu
,
John C. Stephens
b   Department of Chemistry, Maynooth University, Maynooth, Co. Kildare, Ireland
c   The Kathleen Lonsdale Institute of Human Health Research, Maynooth University, Maynooth, Co. Kildare, Ireland
,
Timothy F. Jamison  ∗
a   Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA   Email: tfj@mit.edu
› Author AffiliationsThis work was supported by the DARPA Make-It program (contract ARO W911NF-16-2-0023).
Further Information

Publication History

Received: 14 February 2020

Accepted after revision: 19 March 2020

Publication Date:
09 April 2020 (online)

    Permissions and Reprints All articles of this category


Published as part of the Cluster Integrated Synthesis Using Continuous-Flow Technologies

Abstract

A multioperation, continuous-flow platform for the synthesis of tramadol, ranging from gram to decagram quantities, is described. The platform is segmented into two halves allowing for a single operator to modulate between preparation of the intermediate by Mannich addition or complete the fully concatenated synthesis. All purification operations are incorporated in-line for the Mannich reaction. ‘Flash’ reactivity between meta-methoxyphenyl magnesium bromide and the Mannich product was controlled with a static helical mixer and tested with a combination of flow and batch-based and factorial evaluations. These efforts culminated in a rapid production rate of tramadol (13.7 g°h–1) sustained over 56 reactor volumes. A comparison of process metrics including E-Factor, production rate, and space-time yield are used to contextualize the developed platform with respect to established engineering and synthetic methods for making tramadol.

Key words

continuous processing - flow chemistry - E-factor - Grignard reaction - Mannich bases

Supporting Information

    Supporting information for this article is available online at https://doi.org/10.1055/s-0039-1690884.
  • Supporting Information
 

  • References and Notes

  • 1 Present address: MilliporeSigma, 400 Summit Drive, Burlington, MA 01803, USA.
  • 2 Present address: Bristol Myers Squibb, 1 Squibb Drive, New Brunswick, NJ 08901, USA.
  • 3 Cyclohexanone (neat), formaldehyde (37 wt% in H2O) and dimethylammonium chloride (9.7 M in water) were pumped using ASIA syringe pumps at 172, 92, and 86 μL/min, respectively, into an IDEX cross mixer (P-722, 0.02” thru hole). The resulting combined stream proceeded at 350 μL/min through peek tubing (0.03” I.D., 454 cm, 5.26 mL internal volume) heated between 105–113°C, amounting to a residence time of 15 min. Streams of H2O (239 μL/min) and cyclopentyl methyl ether (244 μL/min) met the output stream of the heated coil at another IDEX cross mixer (P-722). The biphasic stream passed through PFA tubing (0.02” ID, 60” length, 0.309 mL internal volume) at a total flow rate of 834 μL/min and into a Zaiput back pressure regulator pre-set to a pressure of 50 psi. Following depressurization, the reaction stream swiftly proceeded through PFA tubing (0.02” I.D., 12.6” length, 0.065 μL) to separator 1. At steady state, the aqueous layer, containing ca. 1.7 M 2·HCl was removed at a constant 491 μL/min by an ASIA syringe pump. The 2·HCl product stream was neutralized by NaOH (7 M in H2O, 5 equiv, 617 μL/min) and extracted into pentane (634 μL/min). Each of these streams intersected at an IDEX P-722 cross mixer directly connected to PFA tubing (0.08” I.D., 3.0” length, 0.243 mL internal volume) containing one 17 mm long static mixer, followed by PFA tubing (0.02” I.D., 60” length, 0.309 mL) to equilibrate the neutralization. Lastly, the product stream of 2 was collected in separator 2, with an active withdraw of the aqueous layer at 1.108 mL/min using an ASIA Syrris pump. Passive pumping of 2 in pentane was collected over 5.73 h, amounting to 39.1 g of 2 (97% yield). Spectral data of 2 agreed with prior characterization.28 2-((dimethylamino)methyl)cyclohexan-1-one (2)1H NMR (400 MHz, CDCl3): δ = 2.65 (dd, J = 12.5, 6.0 Hz, 1 H), 2.52–2.42 (m, 1 H), 2.41–2.23 (m, 2 H), 2.17 (s, 6 H), 2.06–1.94 (m, 1 H), 1.83 (tt, J = 7.9, 4.7 Hz, 1 H), 1.75–1.56 (m, 2 H), 1.37 (dtd, J = 14.9, 11.1, 3.7 Hz, 1 H). 13C NMR (75 MHz, CDCl3): δ = 212.6, 59.0, 49.0, 45.8, 42.0, 32.5, 28.0, 24.6. IR (neat): 2933.3, 2859.6, 2818.8, 1707.6, 1448.6, 1338.8, 1332.6, 1263.3, 1218.5, 1178.6, 1155.8, 1125.9, 1098.1, 1027.8, 963.1, 943.2, 876.2, 848.9, 828.5, 809.5, 749.8, 668.4 cm–1. Prior to executing the Grignard reaction, a solvent switch from pentane to freshly distilled THF was undertaken. NMR analysis ensuring <10% of the colorless product solution from step 1 was pentane. A solution consisting of 11.64 g of 2 diluted with 40.7 g of THF, all amounting to 50 mL in total volume was prepared. To prepare the reactor, MeOH followed by THF was pumped through the reactor volume with an inline IR spectrophotometer just before the 40 psi BPR to monitor the presence of water in the reaction. At the point in which the reactor was appropriately flushed with THF and no OH stretch remained the 2 and 3 were pumped to begin the Grignard addition. A solution of 2 (0.92 M, THF) and 3 (0.92 M, THF) were pumped at 1.20 mL/min and 2.39 mL/min, respectively, through PFA tubing (0.03” I.D.). These streams met at a Y-assembly (IDEX P-512), combining to a total flow rate of 3.60 mL/min. The combined stream was passed through PFA tubing (0.03” I.D., 187” length, 2.171 mL internal volume) submerged in an oil bath heated to 40 °C, followed by PFA tubing (0.03” I.D., 26.2” length, 0.303 mL internal volume) exposed to ambient temperature. The stream then fed into a Mettler-Toledo FlowIR, followed by PFA tubing (0.03” I.D., 13.8: length, 0.160 mL) feeding directly into a stirring solution of 28 wt% NH4Cl in water. After equilibration, the reaction material was directed to a quench solution consisting of 15.9 mL of a 7 M NH4Cl diluted with 69 mL of deionized water. The temperature of the quench solution did not exceed 40 °C over the course of the collection. The collection lasted 34.4 min (56 reactor volumes). The resulting heterogeneous solution was acidified with 127 mL of 4 M HCl. The mixture was poured into a 500 mL separator funnel and extracted with 127 mL of heptane. The aqueous phase was returned to the quench flask and neutralized with 53 mL of a 5 N aqueous solution of ammonium hydroxide. The aqueous solution with a pH of 11 was extracted 3 times with 127 mL of heptane each time. The organic phase was combined and dried over 50 g of magnesium sulfate. Concentration yielded 7.86 g of a yellow oil (79% yield, 4:1 mixture of diastereomers). The spectral data agreed with published literature.14 It is known to not obtain full mass recovery using liquid–liquid extraction of tramadol.21a The trans-diastereomer (tramadol) was selectively crystallized from a mixture of 4a and 4b using a mixture of 3:1 mixture of 2-propanol to 2-propanol–HCl (5 M) at 0 °C. This step yielded 5.7088 g of tramadol as dense, colorless crystals (85% yield). 1H NMR and 13C NMR spectroscopy agreed with published characterization.29 Tramadol•HCl (4•HCl) 1H NMR (600 MHz, DMSO-d 6): δ = 10.08 (br s, 1 H), 7.26 (t, J = 7.9 Hz, 1 H), 7.09–7.07 (m, 2 H), 6.80 (dd, J = 8.0, 2.5 Hz, 1 H), 5.11 (s, 1 H), 3.76 (s, 3 H), 2.82 (g, J = 11.6 Hz, 1 H), 2.56 (m, 3 H), 2.42 (m, 4 H), 2.23 (t, J = 10.7 Hz, 1 H), 2.09 (m, 1 H), 1.81–1.39 (m, 7 H). 13C NMR (150 MHz, DMSO-d 6): δ = 159.2, 149.9, 129.1, 117.2, 111.6, 111.1, 73.9, 59.4, 54.9, 44.8, 40.8, 40.4, 40.2, 26.1, 24.5, 21.2.
  • 4 Scott LJ, Perry CM. Drugs 2000; 60: 139
    CrossrefPubMedSearch in Google Scholar
  • 5 Critical Review Report: Tramadol, Expert Committee on Drug Dependence, Geneva, Switzerland, November 12-16, 2018; World Health Organization: Geneva, 2018
  • 6 World Health Organization Model List of Essential Medicines, 21st List, 2019. World Health Organization, Geneva, 2019. Licence: CCBY-NC-SA3.0IGO.
  • 9 May SA. J. Flow Chem. 2017; 7: 137
    CrossrefPubMedSearch in Google Scholar
  • 10 Bogdan AR, Dombrowski AW. J. Med. Chem. 2019; 62: 6422
    CrossrefPubMedSearch in Google Scholar
  • 11 Shukla CA, Kulkarni AA. Beilstein J. Org. Chem. 2017; 13: 960
    CrossrefPubMedSearch in Google Scholar
  • 13 Riva E, Gagliardi S, Martinelli M, Passarella D, Vigo D, Rencurosi A. Tetrahedron 2010; 66: 3242
    CrossrefSearch in Google Scholar
  • 14 Fitzpatrick DE, Maujean T, Evans AC, Ley SV. Angew. Chem. Int. Ed. 2018; 57: 15128
    CrossrefPubMedSearch in Google Scholar
    • 15a See Supporting Information Figure S2.
      Crossref
    • 15b see Supporting Information, optimization data, experiment 2.
      Crossref
  • 16 Adamo A, Heider PL, Weeranoppanant N, Jensen KF. Ind. Eng. Chem. 2013; 52: 10802
    CrossrefPubMedSearch in Google Scholar
  • 18 MilliporeSigma. Greener Solvent Alternatives: Supporting the Advancement of Chemistry Through Sound Environmental, Social & Fiscal Responsibilities. https://www.sigmaaldrich.com/content/dam/sigma-aldrich/docs/Sigma/Brochure/greener_solvent_alternatives.pdf (accessed May 24, 2019).
    CrossrefPubMed
  • 19 See Figure S31–S43 for the comparison of Et2O to CPME as an extraction solvent.
  • 20 Pedersen MJ, Skovby T, Mealy MM, Dam-Johansen K, Kiil S. Org. Process Res. Dev. 2016; 20: 2043
    CrossrefSearch in Google Scholar
    • 21a Jarvi ET, Grayson NA, Halvachs RE. US 6399829 B1, 2002
      PubMed
    • 21b Sohani SV, Khochikar PG. A, Tamboli AA. H, Kulkarni RS. WO 2010032254 A1, 2010
      PubMed
  • 22 Carlson R, Carlson JE. Design and Optimization in Organic Synthesis, 2nd ed. Elsevier; Amsterdam: 2005
    Search in Google Scholar
  • 23 Molar flow rate (mmol/min) is directly related to residence time. Molar flow rate can represent residence time on two levels (fast/slow), whereas with the changing reactant stoichiometry, the residence times of the different experiments were conducted on 4-levels.
  • 24 Yoshida J.-I, Nagaki A, Yamada T. Chem. Eur. J. 2008; 14: 7450
    Thieme ConnectPubMedSearch in Google Scholar
  • 25 Le NA, Jones M, Bickelhaupt F, De Wolf WH. J. Am. Chem. Soc. 1989; 111: 8491
    CrossrefSearch in Google Scholar
  • 27 At the scale of our experiments, we did not observe poly-THF formation: Rivera NR, Kassim B, Grigorov P, Wang H, Armenante M, Bu X, Lekhal A, Variankaval N. Org. Process Res. Dev. 2019; 23: 2556
    CrossrefSearch in Google Scholar
    • 28a Roversi E, Scopelliti R, Solari E, Estoppey R, Vogel P, Braña P, Menéndez B, Sordo JA. Chem. Eur. J. 2002; 1336
      PubMed
    • 28b Sielemann D, Keuper R, Risch N. J. Prakt. Chem. 1999; 341: 487
      CrossrefSearch in Google Scholar
  • 29 Smyj R, Wang XP, Han F. Profiles Drug Subst. Excip. Relat. Methodol. 2013; 38: 463
    PubMedSearch in Google Scholar