Triphenylphosphine Oxide- Waste Not, Want Not

Love it or hate it, triphenylphosphine oxide (TPPO, Figure 1) is something we all encounter at some point during our chemistry careers. Most of the time it’s just a by-product from well-established and widely used process such as the Mitsunobou, Wittig, Staudinger, Appel and Corey-Fuchs reactions.1 I remember having a vial full of the white, crystalline material on the shelf during my PhD. I’d isolated it from a Wittig reaction, having assumed at the time that it must be my product, not a stochiometric by-product. I then had to go on a tedious hunt for my olefin in various aqueous and organic mixtures. I kept the TPPO as a reminder never to assume anything (and hoping I might be able to use it for some fancy new reaction in the future). It was also a reminder not to brag to my lab-mates about how easy the work-up was- just a simple filtration…..

Figure 1: Triphenylphosphine oxide (TPPO)

Removal of TPPO, particularly on industrial scale, can involve tedious processing and reworking of crude reaction mixtures. Generation of stochiometric quantities of the by-product makes this task especially difficult. To add insult to injury it often ends up in industrial waste streams and we end up paying someone to dispose of it for us. In fact, in Europe alone, thousands of tonnes of TPPO end up in waste streams every year. Waste-water treatment plants don’t deal with it effectively and incineration results in the formation of fine aerosols containing phosphorus. These can cause blockages in dust filters, drastically reducing the lifetime of the catalysts used for the removal of nitrogen oxides.2 Not surprisingly, efforts to find a method to recycle the phosphorous remain an active area of research. Linked to this is obviously development of improved methods for removal of TPPO from processing streams. Structurally modifying the phosphine, for example introduction of a solubilizing group(s), renders the corresponding phosphine oxide more soluble, thus providing a handle to ease the work-up burden. A good example is tris(4-carboxyphenyl)phosphine (CAS:  807-19-2). I’ve used this successfully in my past to cleave disulfides to thiols and never had problems with phosphine oxide removal.3

Use of catalytic phosphine and turning over the phosphine oxide during the reaction- relying on P(V)/P(III) oxidation state cycling, has been demonstrated for a number of the synthetic transformations listed above. This approach minimises the  phosphine oxide waste burden. Most recently Denton et al  developed elegant a catalytic Mitsunobou reaction that locks a phosphine oxide catalyst in the P(V) oxidation state, mitigating formation of stochiometric amounts of oxide waste (Figure 2).4a A polymer supported version of the catalyst has also been developed.4a Catalytic Witting reactions have also been developed by O’Brien et al and rely on strained phosphine oxides that are easily reduced in situ using stochiometric silanes (Figure 2).4b The Radasovich group have also published a number of interesting papers on P(III)/P(V) redox cycling processes, again using a ring-strained phosphine oxide (Figure 2).4c In the examples the phosphine oxide acts as a pre-catalyst.

Figure 2: Phosphine oxide catalysts

Reducing the P=O bond and freeing up the phosphine is not an easy thing to do. The high oxophilicty of phosphorous and the very strong P-O bond (134 Kcal/mol) provides a thermodynamic driving force for the well-used process listed earlier. As such, once the phosphrous atom is locked into the P(V) oxide, it’s essentially taken out of the phosphorous life-cycle. You might be surprised to learn that there is a limited global reserve of phosphorous and that by 2050 those reserves will be somewhat depleted.5,23

It’s not all doom and gloom however. TPPO does have some useful synthetic applications in its own right. I’ll get to these later.  Before I do, let’s look at the methods available to directly remove the TPPO by-product using both physical techniques and methods that can chemically reduce/recycle it back to the more useful parent phosphine.

The easiest approach is to simply slurry the reaction mixture in a suitable solvent system. A recent paper by Merwade et al in ACS Omega describes separation of TPPO from crude reaction mixtures by optimization of solvent and temperature.6 The paper describes this approach applied successfully on multi-kilo scale.

Precipitation of the solid by co-crystallisation with a suitable crystal co-former such as benzidine or diphenylurea is possible but involves adding one impurity to chase out another (swallowing the spider to catch the fly as it were).7 TPPO has been used as an additive to aid co-crystallization of organic compounds. This is particularly useful if you have a poorly crystalline compound, or not very much compound to work with.  Large, well-defined crystal complexes with proton donors have been described by Etter et al, originating through hydrogen bonding between an acidic hydrogen and the phosphoryl oxygen.8 Co-crystallization of TPPO with H2DIAD (the reduced DEAD by-product from the Mitsunobu reaction) has been described on industrial scale by Anderson et al. Following a multi-kilo scale Mitsunobu reaction, 85% of the triphenylphosphine oxide and H2DIAD present in the crude reaction mixture was routinely removed by filtration of this complex. Single crystal X-ray analysis confirmed a 1:1 complex of triphenylphosphine oxide and diisopropyl hydrazinedicarboxylate (Figure 3).9

Figure 3: Triphenylphosphine oxide−diisopropyl hydrazinedicarboxylate complex

Lipshutz has demonstrated the use of scavenger resins such as Merrifield or immobilized dichlorotrazine as alkylative traps.10 High loading chloromethylated polystyrene has also been used.

An interesting approach reported by Gilheany et al, that strictly speaking falls under chemical recycling methods, is treatment of a crude reaction mixture from a Wittig or Appel process with oxalyl chloride and converting TPPO to the insoluble phenylphosphonium salt.  The salt can then be removed by filtration. Reduction of the salt with LiAlH4 generates the phosphine (if required).11 The same reduction chlorination/reduction sequence reported by Jolly et al uses hexachlorodisilane as a reducing agent.12

These approaches are similar to the 2-step industrial TPPO recycling process developed by BASF. In this process the phosphine oxide reacts with phosgene to form tripheylphosphine dichloride. This intermediate is subsequently reduced with aluminium powder at 130°C (Figure 4). A BASF patent for the process descries a variety of chlorinating agents in combination with Mg, Al or Fe as reducing agents.13 The BASF site in Ludwigshafen, Germany specialises in phosphine intermediates.

Figure 4: BASF industrial TPPO reduction

Other chemical reduction methods include silicon hydrides (the reducing agent of choice for many of the catalytic P(V)/P(III) processes described above), boranes and aluminium hydrides. Reduction using borane generates the stable phosphine-borane complex.14 These various approaches have been nicely reviewed by Herault and Buono.15 Reduction of phosphine oxides in constrained rings is generally much faster and easier than TPPO because of relief of ring-strain. This principle is used successfully in catalytic versions of the Wittig and Staudinger reactions (vide supra) and elegant chemistry by Radosevich et al.4c

There is limited data on the toxicity of TPPO. The EMA’s analysis, based on scant animal data, concludes there is insufficient data to reach a meaningful conclusion.16 Historically BASF carried out chronic in vitro animal studies, presumably because of the possibility that residual TPPO may be present in vitamin A acetate, hundreds of tons of which are prepared synthetically via a late-stage Wittig coupling.17 Complex phosphine oxides have been used in drug discovery but remain underrepresented. A recent review by Gnamm et al from Boehringer Ingelheim gives a good overview of the subject.18

Another approach used to remove TPPO from reaction mixtures is formation of insoluble Lewis acid−TPPO adducts. MgCl2 and ZnCl2 form TPPO complexes that are insoluble in a variety of solvents, particularly toluene, ethyl acetate or mixtures of the two.19a A drawback is that extraction of TPPO from THF using Mg or Zn complexation is ineffective. Magnesium gives very poor recovery and zinc either gives an oil or does nothing at all. Ethereal solvents such as THF are frequently used for reactions generating TPPO by-product (Mitsunobu, Wittig etc), so direct removal of TPPO from these solvents is impractical. Tedious and time-consuming solvent exchange to toluene or ethyl acetate mixtures are required to precipitate the complexes if  THF is used as a reaction solvent.19a An older paper by Frazer et al describes formation of non-transition metal complexes of metal/metalloid halides.19b

Figure 5: Removal of TPPO using metal salts in various solvents

A recent single-author paper by Hergueta addresses this limitation using anhydrous calcium bromide. Removal of TPPO from a THF solution using CaBr2 is very efficient (95-98% removed).20 The TPPO-calcium complex is also insoluble in 2-MeTHF and MTBE (99% TPPO removed) making it an attractive method that expands the scope of the Lewis acid complexation/filtration work-up . Other salt complexes were less efficient (Figures 5+6). Full solid-state characterisation and detailed analysis/optimisation of the complex formation are described in the paper. Judging by the article views on the ACS website it’s struck a chord with many people.

Figure 6: Removal of TPPO via formation of a CaBr2 complex

TPPO complexation has been used to remove residual ruthenium from ring closing metathesis (RCM) reactions. Pre-treatment of crude reaction mixtures with TPPO generates polar metal complexes that are easily separated by silica-gel filtration.21 The disadvantage, which might explain why it’s not proved very popular, is that 50 eqv are required to reduce the ruthenium levels to manageable levels. Solving one problem and creating another.

An interesting paper by Han et al describes reduction of TPPO with sodium metal.22 Depending on reaction conditions, C-P, C-H or O-P bonds can be cleaved generating sodium diphenylphosphinite, sodium 5H-benzo [b]phosphindol-5-olate or sodium benzo[b]phosphindol-5-ide, all of which are used industrially (Figure 7).

Figure 7: Reduction of TPPO with sodium

One of my favourite papers published in 2020 by Sevov et al describes the electrochemical reduction of TPPO.23 As with many developments in synthetic organic electrochemistry the inspiration came from battery technology (Figure 8).

Figure 8: Electrochemical reduction of TPPO

The anode is made of aluminium and in combination with a suitable electrolyte (TMEDA/chloride source) anodic oxidation continuously generates a Lewis acid (AlCl3) that facilitates reduction of the phosphine oxide to the corresponding phosphine at the cathode. The authors demonstrate direct one-pot electro-reduction of the TPPO by-product present in a crude Wittig reaction mixture without reduction of the alkene (Figure 8).

In a similar vein, an older paper by Shibasaki et al describes taking crude reaction mixture after a Wittig process and, without work-up, using the TPPO by-product to facilitate a subsequent Sm(III)-catalyzed asymmetric epoxidation reaction (Figure 9).24 The paper describes unusual looking unsaturated pyrrole amides as substrates- a useful synthetic handle for further elaboration I guess.

Figure 9: TPPO mediated Sm(III)-catalyzed asymmetric epoxidation reaction

Chiral phosphine oxides have been used as a catalyst in a broad range of catalytic asymmetric reactions.25 Phosphine oxides are also known to play a role in several aryl cross-coupling chemistries. A good example is a recent paper by Amgen describing mechanistic studies on a key Suzuki reaction used in the manufacture of Sotorasib (Figure 10), an oncology drug that targets the KRAS G12C mutation.26 A review by Hong et al discusses some early examples of phosphine oxides as ligands for transition-metal catalysed cross coupling reactions.27

Figure 10: Phosphine oxide mediated Suzuki reaction towards Sotorasib (Amgen)

To close- a somewhat unusual application of TPPO. The FBI have trained sniffer- dogs to detect electronic storage devices (digital media, cell phones and microSD cards) by the smell of trace quantities of residual TPPO.28 The latter is found universally in electronic storage devices. So, if you find yourself at the airport after working up that Wittig reaction, the affectionate little dog sniffing round your feet at the gate might have an ulterior motive….

See you next time.


  1. Mitsunobu: S. Fletcher Chem. Front., 2015, 2, 739-752;Wittig: S. Marsden Nat. Chem. 2009, 1, 685-687; Appel: H. R. Bjorsvik et al, React. Chem. Eng. 2022, 7, 1650-1659; Staudinger: Y. G. Golobobov et al, Tetrahedon 1981, 37, 437-472; Corey-Fuchs: O. kolodiazhnyi
  2. Extent of sonochemical degradation and change of toxicity of a pharmaceutical precursor (triphenylphosphine oxide) in water as a function of treatment conditions: R. Emery et al, Int. 2005, 31 , 207-211
  3. Selective reduction of disulfides by tris(2-carboxyethyl)phosphine: G. Whitesides et al, Org. Chem.1991, 56, 2648–2650
  4. a) Redox-neutral organocatalytic Mitsunobu reactions: R. Denton et al, Science 2019, 365, 910-914; The sun never sets on classic synthetic methodology- redox neutral organocatalytic Mitsunobu reactions: J. Studley blog post 2019,;Development of a robust immobilized organocatalyst for the redox-neutral mitsunobu reaction: M. Pericas et al, Green Chem. 2021, 23, 8859-8864; b) Phosphetane oxides as redox cycling catalysts in the catalytic Wittig reaction at room temperature: T. Werner et al, ACS Catal. 2019, 9, 9237–9244; Catalytic Wittig and aza-Wittig reactions: Toy et al, Beilstein J. Org. Chem.  2016, 12, 2577-2587; c) anti-1,2,2,3,4,4-Hexamethylphosphetane 1-oxide: A. Radosevich et al, Org. Synth. 2019, 96, 418-435
  5. Let’s make white phosphorous obsolete: C. Cummings et al, ACS Cent. Sci. 2020, 6, 848-860
  6. Triphenylphosphine oxide removal from reactions: the role of solvent and temperature: A. Merwade et al, ACS Omega 2021, 6, 13940-13945
  7. Predicting co-crystallization based on heterodimer energies: the case of N,N′-diphenylureas and triphenylphosphine oxide: M. Solomos et al, growth des. 2015, 15, 5068−5074 and references therein
  8. Triphenylphosphine oxide as a crystallization aid: M. Etter et al, J. Am. Chem. Soc. 1988, 110, 639-640
  9. Sulfonation with inversion by Mitsunobu reaction: an improvement on the original conditions: N. Anderson et al, J. Org. Chem. 1996, 61, 7955-7958
  10. Efficient scavenging of PPh3 and Ph3P=O with high loading Merrifield resin: B. Lipshutz et al, Lett.2001, 3, 1869–1871
  11. A convenient chromatography-free method for the purification of alkenes produced in the Wittig reaction: D. Gilheany et al, Org. Biomol. Chem. 2012, 10, 3531-3537
  12. A mild one-pot reduction of phosphine(v) oxides affording phosphines(III) and their metal catalysts: I Jolly et al, Organometallics 2021, 40, 693-701
  13. Preparation of triphenylphosphine: US5527966A (BASF); Process for the preparation of tertiary phosphines: EP0548682B1 (BASF)
  14. Reduction of functionalized tertiary phosphine oxides with BH3: M. Pietrusiewicz et al, J. Org. Chem.2015, 80, 1672–1688
  15. Reduction of secondary and tertiary phosphine oxides to phosphines: D. Herault & G. Buono et al, Chem. Soc. Rev. 2015, 44, 2508-2528
  16. Provisional peer reviewed toxicity values for triphenylphosphine oxide (CASRN 791-28-6): EPA/690/R-07/038F (2007)
  17. The Wittig reaction in industrial practice: H. Pommer, Chem. Int. Ed. 1977, 16, 423-429
  18. Phosphine oxides from a medicinal chemist’s perspective: physicochemical and in vitro parameters relevant for drug discovery: C. Gnamm et al, J. Med. Chem. 2020, 63, 7081-7107
  19. a) Removal of triphenylphosphine oxide by precipitation with zinc chloride in polar solvents: D. Weix et al, J. Org. Chem. 2017, 82, 9931-9936; b) Triphenylphospine oxide complexes of non-transition metal halides: M. Frazer et al, J. Inorg. Nucl. Chem. 1963, 25, 637-640
  20. Easy removal of triphenylphosphine oxide from reaction mixtures by precipitation with CaBr2: A. R. Hergueta Org. Process Res. Dev. 2022, 26, 1845-1853
  21. A convenient method for the efficient removal of ruthenium byproducts generated during olefin metathesis reactions: G. Georg et al, Org. Lett. 2001, 3, 1411-1413
  22. Conversion of triphenylphosphine oxide to organophosphorus via selective cleavage of C-P, O-P, and C-H bonds with sodium: L-B Han et al, Commun. Chem. 2020, 3, Article 1
  23. Direct and scalable electroreduction of triphenylphosphine oxide to triphenylphosphine: C. Sevov et al, J. Am. Chem. Soc. 2020, 142, 3024-3031
  24. Catalytic asymmetric epoxidation of α,β-unsaturated N-acylpyrroles as monodentate and activated ester equivalent acceptors: M. Shibasaki et al, Tetrahedron 2006, 62, 6630-6639
  25. Enantioselective and chemoselective phosphine oxide-catalyzed aldol reactions of n-unprotected cyclic carboxyimides: S. Kotani et al, Eur. J. 2022, e202203506
  26. Kinetic and mechanistic investigations to enable a key Suzuki coupling for Sotorasib manufacture- what a difference a base makes: J. Murray et al, Process Res. Dev. 2023, 27, 198–205
  27. Secondary phosphine oxides: Versatile ligands in transition metal-catalyzed cross-coupling reactions: F-E Hong et al, Coordination Chem. Rev. 2012, 256, 771-803