Photolytic, radical-mediated hydrophosphination: a convenient post-polymerisation modification route to P-di(organosubstituted) polyphosphinoboranes [RR′PBH2]n

New, air-stable inorganic soft materials are accessible under mild conditions via TEMPO-mediated radical hydrophosphination of alkenes using polyphosphinoboranes.


Introduction
Polymers featuring elements other than carbon in the main chain are attracting widespread interest as functional so materials with an expanding range of applications. These macromolecules possess attributes that complement those of easily processed state of the art organic polymers by introducing additional features such as enhanced thermooxidative stability, low temperature elasticity, ame retardancy, tunable optoelectronic properties, and the ability to form ceramic lms and bers on pyrolysis. 1 Polyphosphinoboranes, [RR 0 PBH 2 ] n are formally isoelectronic with polyolens, and have recently emerged as a new class of inorganic polymers, 2 with potential uses as precursors to PB semiconductor-based ceramics, etch-resists, ameretardant materials, and as piezoelectrics. 3 The development of new and improved routes to high molar mass polyphosphinoboranes is therefore an expanding area of research. 4 It is now possible to access several derivatives of Pmonosubstituted polyphosphinoboranes [RHPBH 2 ] n where R is an alkyl or aryl substituent via catalytic dehydrogenation using Rh, Ir or Fe precatalysts or thermally-induced Lewis base elimination routes (Scheme 1a). In contrast, examples of Pdisubstituted polyphosphinoboranes (i.e. [RR 0 PBH 2 ] n , R and R 0 s H) are extremely scarce. Early work in the 1950s and 1960s claimed the formation of polymeric materials via thermally-induced dehydrocoupling of phosphine-borane adducts R 2 PH$BH 3 at ca. 200 C, oen in the presence of additives such as amines, which were suggested to prevent cyclisation. 5 However, the products were not unambiguously characterised and, where reported, yields and molar masses were very low. Attempts to apply current catalytic routes towards Pdisubstituted polyphosphinoborane targets by dehydrocoupling of secondary phosphine boranes, RR 0 PH-BH 3 , have been unsuccessful to date, yielding instead small rings or oligomeric materials. 2b,4a,c,e,6 High molar mass P-disubstituted polyphosphinoboranes would be devoid of P-H bonds and are likely to be the most thermally and environmentally robust and therefore the most realistically useful in applications. Strategies to access these materials are therefore of substantial interest.
Post-polymerisation modication (PPM), for example, by activation of main-chain E-X (X ¼ halogen, H) bonds of inorganic polymers such as polysiloxanes, 7 polyphosphazenes, 8 polysilanes, 9 and polyferrocenylsilanes, 10 is a well-known strategy for functionalising these polymers allowing the tuning of diverse physical and chemical properties. Indeed, the broad scope of PPM for polydihalophosphazenes is vital to the applications of polyphosphazene-based materials. 11 This methodology has also been used to synthesise bottlebrush polymers 12 and polyphosphazene gels which have interesting elastomeric properties. 11a,c We envisioned that a PPM approach involving conversion of preformed high-molar mass polyphosphinoborane [RHPBH 2 ] n to the target disubstituted [RR 0 PBH 2 ] n polymers would overcome the limitations associated with existing synthetic routes and give access to a more robust and tunable class of P-disubstituted polyphosphinoboranes. Herein, we report conditions under which a broad range of alkenes undergo insertion into the P-H bonds of poly(phenylphosphinoborane) to yield high-molar mass derivatives of [PhR 0 PBH 2 ] n (R 0 s H, Scheme 1b). In addition, we disclose the extension of this hydrophosphination approach to prepare crosslinked elastomers and water-soluble materials based on polyphosphinoborane backbones.

Results and discussion
Hydrophosphination of 1-octene using [PhHPBH 2 ] n The hydrophosphination of alkenes with primary and secondary phosphines is a well-studied reaction for which numerous catalytic and radical based protocols have been reported. 13 This addition is analogous to the ubiquitous thiol-ene addition reaction and has recently been exploited for the synthesis of phosphorus-containing network polymers. 14 Interestingly, the insertion of alkenes into P-H bonds of phosphineborane adducts (RR 0 HP-BH 3 ; R ¼ Ph, R 0 ¼ Ph or Me) has also been reported by Gaumont and coworkers, 15 providing a model for the putative addition of alkenes to P-monosubstituted poly(phenylphosphinoborane) [PhHPBH 2 ] n (1).
For all of our investigations, 1 was synthesised via previously reported iron-catalysed dehydrocoupling of phenylphosphineborane (PhH 2 P-BH 3 ). 4a PhH 2 P-BH 3 was heated to 100 C in toluene for 20 h in the presence of 1 mol% [FeCp(CO) 2 OTf], yielding polymer as a pale yellow solid with a molar mass of around 68 000 Da and a PDI of 1.5. The discolouration of this polymer is reported to come from residual iron species remaining despite repeated precipitation from DCM into cold pentane (À78 C). 4h Initial studies showed that, unlike for the aforementioned phosphine-borane adducts studied by Gaumont, heating 1 (0.2 mmol) with 1-octene (0.2 mmol) in THF (0.5 ml) at 60 C for 24 h did not result in detectable insertion of the alkene into the P-H bonds of the polymer based on 31 P NMR analysis. However, when the reaction mixture was irradiated under UV light for 20 h at 20 C (Table 1, entry 1), a single peak emerged in its 31 P NMR spectrum at d ¼ À23.5 ppm with no apparent 1 J PH coupling (cf. d ¼ À48.9 ppm, 1 J PH ¼ 349 Hz for 1). The 1 H NMR spectrum of the reaction mixture showed a signicant reduction in the intensity of the P-H resonances and emergence of a number of broad peaks in the 0.8-1.3 ppm region corresponding to new aliphatic protons. These spectroscopic data are consistent with insertion of 1-octene into the P-H bond of 1. Analogous to Gaumont's work with phosphine-borane adducts, 15 the emergence of a single peak in the 31 P NMR spectrum suggests that exclusive anti-Markovnikov addition had taken place within the NMR detection limit. Integration of the resonances in the 31 P NMR spectra indicated around 90% conversion to the Pdisubstituted species (Fig. 1) giving a random copolymer consisting of [Ph(octyl)PBH 2 ] and [PhHPBH 2 ] units.
The molar mass of the product was determined by gel permeation chromatography (GPC) using polystyrene standards. A bimodal distribution was observed (M n ¼ 83 000 Da, PDI ¼ 1.21 and M n < 3000 Da) with a high molar mass polymer/ low molar mass polymer peak ratio of 3 : 7. 4a We interpret the presence of these two fractions as evidence that growth in molar mass by alkene addition is accompanied by main-chain cleavage (vide infra). We postulate that this chain cleavage is caused by undesired radical-induced side reactions such as backbiting and b-scission reactionsprocesses commonly invoked in the photodegradation of organic polymers. 16 Analysis of the above reaction mixture by NanoSpray electrospray ionisation mass spectrometry (ESI-MS, positive mode, DCM solvent), showed a repeat unit of 234.2 m/z, which corresponds to a successive loss of [Ph(octyl)PBH 2 ]. As expected for conversion of 80-90%, repeat units of 122.0 m/z corresponding to loss of [PhHPBH 2 ] were also observed. The maximum observed m/z was around 3000, much lower than that observed by GPC; however, this is analogous with previous characterisation of polyphosphinoboranes 4a,h and polyaminoboranes, 17 and is a noted limitation of ESI-MS for molar mass determination of these polymers. 18 Matrix-assisted laser desorption/ionization time of ight mass spectrometry (MALDI-TOF MS) was also undertaken in an attempt to overcome the low m/z detection limit of ESI-MS; however, no high molar mass fraction was detected suggesting problems with the ionisation of these materials under MALDI conditions. A limitation of this methodology is that the hydrophosphination was slow, requiring 20 h to achieve 90% conversion. A variety of different conditions was therefore investigated to optimise this reaction (Table 1). Given the success of UVpromoted hydrophosphination (Table 1, entry 1), the introduction of a photoinitiator was investigated: addition of 10 mol% 2,2-dimethoxy-2-phenylacetophenone (DMPAP) to the reaction mixture and irradiation under UV light in THF showed a marked increase in reaction rate (90% conversion in 1 h, entry 2). Signicantly, the reaction could be scaled up to 2 mmol without loss of activity (entry 3); whereas, for UV irradiation without an initiator, a drastic reduction in reaction rate was observed upon scale up (entry 4). Decreasing the amount of DMPAP to 1 mol% led to a slower reaction rate (entry 5) and increasing the amount of DMPAP to 30 mol% did not accelerate the reaction further (entry 6). Lowering the reaction temperature to 0 C also resulted in a lower conversion aer 1 h (entry 7). The reaction proceeded equally well in THF or chlorobenzene, and a slight increase in conversion aer 0.25 h was observed when using toluene or 1,2-dichlorobenzene (entries 8-11). Changing the solvent did not have a signicant effect on the molar mass prole of the resulting polymer according to GPC analysis and because of the higher volatility and therefore easier removal of THF from the polymer products, THF was used for all subsequent reactions. Yields and molar masses obtained when reactions are carried out in air were comparable to those obtained using dry and degassed solvents under a nitrogen atmosphere.
As with the case in which no photoinitiator was used, a bimodal molar mass distribution was observed upon analysis of the polymer product of entry 2 by GPC (Fig. 2). In an effort to minimise any molar mass decline accompanying this reaction, an analogous reaction was attempted using blue light instead of UV light; however, no reaction was observed (ESI Table S1, entry 1 †). While the targeted hydrophosphination did not occur under these conditions, use of blue light irradiation together with photocatalyst 9-mesityl-10-methylacridinium perchlorate and diphenyliodinium triate did result in the desired reaction taking place (25% conversion aer 16 h) (ESI Table S1, entry 2 †); however, given the sluggish nature of this reaction, this methodology was not pursued further. DMPAP (10 mol%) THF 1 h 88 Di-tert-butyl nitroxide (10 mol%) a All reactions were carried out with 0.2 mmol of [PhPHBH 2 ] n and one equivalent of 1-octene in a borosilicate NMR tube in 0.5 mL solvent and irradiated under UV light at 20 C unless stated otherwise. UV irradiation was carried out using a 125 W medium-pressure mercury lamp. b Determined by 31 P NMR integrations, conversion ¼ x/(x + y) Â 100. c 2 mmol of [PhPHBH 2 ] n , one equivalent of 1-octene and 5 mL THF. d Reaction carried out at 0 C. TEMPO is well known to reversibly bind to organic radical species leading to its pioneering use in the eld of nitroxidemediated polymerisation (NMP). 19 This reversible binding establishes an activation-deactivation equilibrium which reduces the concentration of active radical species giving a more controlled polymer growth. Given the success of NMP protocols to control radical reactions, we investigated the effect of addition of TEMPO to the hydrophosphination of 1-octene with 1. When 100 mol% of TEMPO was added to an NMR tube containing 1 (0.2 mmol), 1-octene (0.2 mmol), DMPAP (0.02 mmol) and THF (0.5 mL), no reaction was observed aer irradiation for 1 h (Table 1, entry 12). However, when instead, 10 mol% of TEMPO was added under otherwise analogous reaction conditions, the conversion aer 1 h was comparable to the case where no TEMPO was added (compare entries 13 and 2). Furthermore, upon characterisation of the molar mass of the polymer using GPC, it was now found that signicantly more high molar mass material remained (peak ratio 7 : 3 high molar mass polymer/low molar mass polymer), suggesting that polymer degradation during the course of the reaction was signicantly reduced (Fig. 2). We postulate that the TEMPO acts to reduce the concentration of reactive radicals via reversible binding to the radical species produced from the photoinitiator resulting in a more controlled hydrophosphination without detrimental side reactions that cause chain cleavage. A similar degree of conversion was found when an alternative nitroxide, di-tert-butyl nitroxide, was used in place of TEMPO (entry 14).
We also found that it was possible to carry out the hydrophosphination of 1-octene using 1 thermally at 60 C in THF using 10 mol% AIBN as an initiator. This thermally-induced hydrophosphination is signicantly slower than the UV mediated version (taking 27 h to reach 90% conversion, Fig. S1 †); however, this allowed for convenient monitoring of the reaction by 31 P NMR (vide infra).

Mechanistic studies
We propose that the reaction of poly(phenylphosphinoborane) and 1-octene in the presence of 10 mol% DMPAP and irradiation under UV light takes place via a radical chain reaction in which a radical initiator (Inc) forms from the photolysis of DMPAP (Scheme 2A), and subsequently abstracts a hydrogen atom from phosphorus on the polymer chain (Scheme 2B). This then adds to the alkene in an anti-Markovnikov fashion to give the most stable secondary radical based on the alkyl chain (Scheme 2C). To continue the radical chain reaction, a hydrogen is then abstracted from another position on the polymer chain (Scheme 2D). This is analogous to the mechanism reported by Gaumont and co-workers for the microwave irradiation-induced hydrophosphination of alkenes using secondary phosphine-boranes. 15 Introduction of TEMPO into this system has an interesting effect: UV irradiation of 1 (0.2 mmol), 1-octene (0.2 mmol), DMPAP (0.02 mmol) and THF (0.5 mL) alone at 20 C, shows 75% conversion from 1 to the P-disubstituted polymer aer just 10 minutes (determined by 31 P NMR spectroscopy of the crude reaction mixture). However, in contrast, when 10 mol% TEMPO was present in an analogous reaction mixture, there was minimal conversion to the P-disubstituted polymer aer 10 minutes of UV irradiation (Fig. S2 †). Nevertheless, analysis by 31 P NMR spectroscopy of both reactions aer 1 h of irradiation shows comparable degrees of conversion of around 90% (Fig. S3 †). This suggests that the addition of TEMPO causes an induction period for the hydrophosphination reaction. We also explored an analogous thermal reaction using AIBN and TEMPO wherein an NMR tube were charged with 1 (0.1 mmol), 1-octene (0.1 mmol), AIBN (0.01 mmol), TEMPO (0.01 mmol) and THF (0.5 mL) and was placed in an oil bath at 60 C. The reaction was monitored by 31 P NMR spectroscopy. A clear induction period was observed, with no detectable conversion by 31 P NMR spectroscopy aer 1 h but around 10% conversion aer 2 h, with continually increasing conversion thereaer (Fig. 3). We postulate that the induction periods that we observe are caused by reversible reaction of TEMPO with the radical species produced from the photodegradation of DMPAP under UV light (Scheme 2E) or by thermal degradation of AIBN. The adducts formed could then break down initiating the hydrophosphination reaction. The formation of the 2-cyanopropyl-TEMPO adduct (Fig. 4A) has been reported previously from the heating a solution of AIBN and TEMPO in toluene, 20 and so it is plausible that we are also forming this species prior to any reaction with the polymer. We also attempted to isolate an adduct between DMPAP and TEMPO. The photodegradation of DMPAP has been reported to yield several products, 21 a number of which could conceivably react with TEMPO complicating any investigation. Nevertheless, analysis of the crude reaction mixture aer the irradiation of equimolar amounts of DMPAP with TEMPO in THF by ESI mass spectrometry showed signals that correspond to 2,2,6,6-tetramethylpiperidin-1-yl benzoate fragments (Fig. 4B), as well as hydrogenated TEMPO supporting our hypothesis that an adduct forms between TEMPO and radicals derived from DMPAP (Fig. S4 †).
Since these proposed adducts closely resemble alkoxyamine compounds that are commonly used as initiators in NMP, 22 we sought to determine if alkoxyamines could facilitate the reaction of 1-octene with 1. Heating an NMR tube charged with 1 (0.1 mmol), 1-octene (0.1 mmol), toluene (0.5 mL) and 0.01 mmol of the commercially available N-tert-butyl-N-(2- methyl-1-phenylpropyl)-O-(1-phenylethyl)hydroxylamine (Fig. 4C) to 100 C, resulted in the desired hydrophosphination reaction taking place, albeit much more slowly than using our photoinitiated system (Fig. S5 †). Signicantly when this alkoxyamine was used, no induction period was observed supporting our assertion that adduction formation is involved in the rst step of the photoinitiated hydrophosphination in the presence of DMPAP and TEMPO.
In nitroxide mediated polymerisations it is generally accepted that the nitroxide is able to reversibly bind to the growing polymer chain and this mediates the reaction resulting in a controlled polymer growth. In order to test whether TEMPO is binding to phosphorus-based radicals on the polyphosphinoborane main chain, 1 (0.2 mmol) was irradiated with DMPAP (0.2 mmol) and TEMPO (0.2 mmol) in THF (0.5 mL) at 20 C. Analysis of the crude reaction mixture by 31 P NMR spectroscopy aer 4 h, showed the emergence of a very minor signal at 120 ppm which we tentatively assign to the polymer bound to TEMPO (Fig. S6 †) due to the similarity in chemical shi to the recently reported Ph 2 POTEMP ( 31 P d ¼ 110.8 ppm); 23 however, no evidence of binding of TEMPO to the polymer chain could be observed by mass spectrometry. Addition of 1octene and continued irradiation resulted in the disappearance of this signal at 120 ppm and the emergence of the signal at À23.5 ppm which corresponds to the hydrophosphination of 1octene by 1 (Fig. S7 †).

Large scale syntheses and properties of P-disubstituted polyphosphinoboranes
Following the success of this new hydrophosphination methodology, we targeted the isolation of a series of polymers to investigate the difference in their physical properties. To this end we targeted various degrees of substitution of poly(phenylphosphinoborane) using 1-octene by varying the reaction stoichiometry (0.1 eq. 1-octenepolymer 2, 0.6 eq.polymer 3, 1 eq.polymer 4, and 2 eq.polymer 5) (Scheme 3). We also targeted other alkenes: allylbenzene (1 eq.polymer 6), allyl pentauorobenzene (1 eq.polymer 7), and 1H,1H,2H-per-uorohexene (1 eq.polymer 8). The synthesis of these polymers followed the same procedure, 1 (2 mmol), DMPAP (0.2 mmol), TEMPO (0.2 mmol), and alkene were added to a vial and dissolved in THF (5 mL). The reaction mixture was irradiated under UV light for 2 h at 20 C for 2-4 and 6-8. For polymer 5, the reaction mixture was irradiated for 24 h at 20 C. The polymers were isolated by precipitation from THF into H 2 O/ isopropanol (1 : 1 v/v) at À20 C (polymers 3, 4, 5, and 8) or from DCM into pentane at À78 C (polymer 2, 6, and 7) and then dried under vacuum at 40 C for at least 48 h. The polymers were isolated as light-yellow solids except for 4 and 5 which were pale yellow-brown gums. The discolouration for these polymers Scheme 2 Proposed reaction mechanism for the UV-induced hydrophosphination of alkenes using 1 in the presence of DMPAP (and the effect of addition of TEMPO to the reaction mixture).  likely originates from small amounts of residual iron species from the polymerisation of phenylphosphine-borane using [FeCp(CO) 2 OTf]. The 11 B NMR spectra of the resultant polymers showed little change from that of the parent poly(phenylphosphinoborane) (a broad singlet at around À34 ppm). 31 P NMR chemical shis of these isolated polymers were found at around À24 ppm. As expected, a singlet was observed in the 1 H-coupled 31 P NMR spectra alongside a doublet at d ¼ À48.9 ppm corresponding to [PhHPBH 2 ] units in all polymers except 5. From the 31 P NMR spectra, the degree of conversion to the P-disubstituted polymer could be calculated. When 1 eq. alkene was used, conversions of between 72 and 82% were observed (Table 2, polymers 4, 6, 7, and 8). Different degrees of substitution could be obtained by varying the reaction stoichiometry (compare polymers 2-5). To obtain the fully Pdisubstituted polymer 5, a greatly extended reaction time and two equivalents of 1-octene were required. We postulate that this is due to reactive sites becoming less accessible as conversion approaches 100%. The successful incorporation of the alkene was conrmed by ESI-MS and for each polymer, fragments corresponding to [PhRPBH 2 ] repeat units could be detected. The molar masses of these polymers were determined by GPC relative to polystyrene standards and were found to range from M n ¼ 81 000 to 130 000 Da (PDIs ¼ 1.1-1.9). No change in the 31 P NMR spectra or GPC chromatograms was detected aer the solid polymers were exposed to air for 6 months, indicating that these polymers are air-stable. These polymers also appear to be water-stable as addition of a few drops of water to a THF solution of these polymers (5 mg in 1 mL THF) and leaving open to air for 24 h at 20 C also resulted in no change in the NMR spectra or GPC chromatograms.
The thermal properties of functionalised polyphosphinoborane polymers 2-8 were investigated by thermogravimetric analysis (TGA, N 2 atmosphere, heating rate 10 C min À1 ) and differential scanning calorimetry (DSC, heating rate 10 C min À1 ) ( Table 2). Thermal stability was quantied by comparing T 5%the temperature at which the polymer loses 5% of its original mass. P-Disubstituted polymers were found to have slightly higher T 5% values than 1, except for 2 and 8 which were marginally lower. This increased thermal robustness relative to the starting [PhPHBH 2 ] n polymer is promising for further utility of these modied polymers. The thermal stability of the octyl substituted polymers increased up to around 60% insertion (compare data for polymers 1 and 3), but little further increase was observed with additional alkene insertion (polymers 4 and 5). The onset of mass loss has been attributed to thermally induced H 2 -loss leading to further polymer degradation pathways. 3b These results suggest that the presence of an octyl group at every other repeat unit is sufficient to suppress the inter-chain P-H/B-H interaction required for H 2 elimination. However, higher degrees of insertion presumably enhances P-B backbone ssion due to steric pressure and the concomitant molar mass decline is likely to reduce the thermal stability of the polymer. It has also been postulated that thermally induced crosslinking is important for thermal stability of polyphosphinoboranes. As the number of P-disubstituted units increases, this would become increasingly difficult as there are both fewer sites available for crosslinking and a higher steric bulk reducing favourable interactions between polymer chains.
Reecting the random addition of 1-octene along the polymer backbone, only one glass transition temperature (T g ) was observed for 2-8 ( Table 2). The T g values for 2-5 are lower than that for 1, which is ascribed to the presence of long alkyl side chains that increase the polymer free volume and therefore reduce T g . Consistently, the T g values for 2-5 also show an inverse relationship with the extent of alkene insertion as expected for greater incorporation of a long alkyl chain. Polymers 4 and 5 have glass transition temperatures signicantly below room temperature and are gums whereas the other polymers are Scheme 3 Reaction conditions for the hydrophosphination of alkenes with 1. glassy solids. Polymers 6-8 have T g values that are higher than for 1, which we tentatively ascribe to greater steric interactions between the uorinated and/or aryl groups in the polymer side chains increasing the rigidity of the polymer.

Crosslinked poly(phenylphosphinoborane)
Following the success of the insertion of alkenes into P-H bonds of 1, we sought to extend this methodology to other polyphosphinoborane-based so materials. We found that when 1 is irradiated with 0.1 eq. of 1-octene, a signicant shoulder is detected to the high molar mass polymer fraction (Fig. S12 †). We assign this to competitive polymer cross-linking (by P-P or P-B bond formation) at low degrees of substitution. This hypothesis is supported by irradiation of 1 with 10 mol% DMPAP in the absence of alkene, which yielded material with very high molar mass (>400 000 Da, Fig. S63 †). Further irradiation under these conditions results in the formation of insoluble material suggesting a higher degree of cross-linking. To investigate the potential of hydrophosphination of dienes to achieve controlled cross-linking, a solution containing 1, 1,5hexadiene (15 mol%), DMPAP (10 mol%), and TEMPO (10 mol%) was irradiated in THF at 20 C for 24 h. A so, pale yellow solid was obtained, which showed reversible organogel behaviour upon exposure to excess THF or vacuum (Fig. 5). This material was puried by repeated extraction with THF until the washings were colourless. Drying of this material under vacuum yields a pale-yellow brittle solid. This material undergoes reversible organogel swelling behaviour: if le in THF for 48 h, the material swells to 210% of its original mass; subsequent application of vacuum reverts the gel back to its brittle phase. No glass transition temperature was detected when the material was analysed by DSC. The ceramic yield of this crosslinked poly(phenylphosphinoborane) was found to be 54%, slightly higher than for non-crosslinked polyphosphinoboranes. These properties are promising for further utility of crosslinked polyphosphinoboranes and the use of different polymer precursor and crosslinking agents should yield gels with markedly different properties.

Synthesis and characterisation of a water-soluble bottlebrush polyphosphinoborane
We also explored the formation of a polyphosphinoborane bottlebrush polymer via the graing-to reaction of 1 with two equivalents of poly(ethylene glycol) methyl ether methacrylate in the presence of DMPAP (10 mol%) and TEMPO (10 mol%) in THF (Scheme 4). Aer UV irradiation for 2 h and subsequent removal of THF from the resultant solution and redissolution in CDCl 3 , a graing density of 58% was determined by integration of the 31 P NMR spectrum. This is in the range typically found for graing-to approaches to bottlebrush polymer formation (typically graing densities are <60% for graing-to approaches). 24 This polymer was found to be water-soluble (the rst water soluble polymer with a polyphosphinoborane backbone). No signicant change in the chemical shis of the 31 P NMR peaks was observed whether CDCl 3 or D 2 O was used as the solvent indicating that the polymer is water stable, although signicant broadening of the signals is observed when D 2 O is the solvent (compare Fig. S72 and S73 †). In order to remove any excess poly(ethylene glycol) methyl ether methacrylate, dialysis was performed using MW 12-14 kDa cut-off dialysis tubing in water. The successful removal of the poly(ethylene glycol) methyl ether methacrylate was conrmed by comparison of the dynamic light scattering trace for 9 and for poly(ethylene glycol) methyl ether methacrylate ( Fig. S75 and S76 †). The resulting polymer had a M n of 156 000 Da and a PDI of 1.34 determined by GPC in THF. The thermal properties of this polymer were investigated by DSC and TGA. No T g was detected by DSC analysis; however, a T m at 40 C was observed for the PEG side chains. The T 5% was found to be 300 C, signicantly higher than for other linear polyphosphinoboranes. This suggests that the presence of the long PEG chains imparts signicant additional thermal stability, and this bodes well for future research into applications of this interesting class of polyphosphinoborane polymers.

Conclusions
We have achieved the synthesis of P-di(organosubstituted) polyphosphinoboranes using a mild, scalable, photoinitiated process for inserting olens into the P-H bonds of preformed Pmonosubstituted derivatives under benchtop conditions. The use of DMPAP and TEMPO and UV irradiation serves to minimise molar mass decline during the course of this hydrophosphination reaction and facilitated the formation of random copolymers with controlled functionalisation as well as fully P-disubstituted derivatives. Investigations into the mechanistic reason behind the favourable effect of TEMPO addition suggested that reversible binding of TEMPO to radical species formed during the reaction could be preventing deleterious side reactions from occurring which lead to polymer degradation. The material properties of the new high molar mass polymers are tunable by the choice of alkene employed. We also describe the synthesis of the rst controllably crosslinked polyphosphinoborane, a material that exhibits organogel behaviour, and the synthesis of a water-soluble bottlebrush polymer featuring a polyphosphinoborane backbone. The results described offer promise for unlocking new applications for polyphosphinoboranes and relevant work in the area is currently underway in our group.

Conflicts of interest
There are no conicts of interest to declare.