**1. Introduction**

The most important examples of inorganic polymers [1–3] are the polysiloxanes—(R2Si-O)n [4–8], polysilanes—(R2Si)n [9–11], and the polyphosphazenes—(R2P=N)n− [12–14]. They have been the subject of significant research due to desirable properties, such as biocompatibility, high thermal resistance, oxidative stability, UV resistance, interesting photoelectronic behavior, and high flexibility that are often difficult or impossible to achieve in carbon-based organic polymers. Much of the current research on polyphosphazenes is focused on applications in the life sciences as drug delivery vehicles and other biological applications [15–18].

The field of polyphosphazene synthesis has expanded significantly with the advent of more efficient and faster synthetic routes such as the ambient temperature preparation of poly(dichlorophosphazene) by the PCl5-initiated polymerization of *N*-trimethylsilyl-*P*-trichlorophosphoranimine, [13,19] the polymerization of *N*-trimethylsilyl-*P*-tris(2,2,2- trifluoroethoxy)phosphoranimine by antimony pentachloride initiator, [20] and poly (organophosphazenes) from partially halogenated phosphoranimines [21]. These reactions are examples of cationic polymerization of phosphoranimines. The cationic route has also led to polyphosphazenes with functional and terminal groups at the chain ends [22–27] as well as many other derivative copolymers with more complex architecture [28–32].

The first catalyzed/initiated polymerization of a phosphoranimine monomer by fluoride ion was previously reported (Scheme 1a) and then extensively investigated [33]. The successful conversion of these monomers to polyphosphazenes initiated/catalyzed by fluoride anion, trifluoroethoxide anion, and *N*-methylimidazole was described [34–36]. Polyphosphazene random and block copolymers with mixed alkoxy/alkoxyether substituents were subsequently prepared by the fluoride/anionic method and characterized [37]. Other examples include polyphosphazenes with alkyl or aryl substituents [38], polyphosphazenes with electronegative nitropropoxy groups bound to the phosphorus atom of the *N*-silylated phosphoranimine [39], polymerization of *N*-silyl-*P*-diethyl phosphoranimine with fluoride

and phenoxide

imine  initiators [40], and

> with water in the presence of

polymerized

The use of chain capping agents is a technique to prepare well-defined polyphosphazenes produced by anionic polymerization of phosphoranimines, since active chain segments are subject to late-stage condensation reactions that can increase molecular weight with diminished control of chain length. In Scheme 1b, active anionic chain ends of the forming polymer can react with a monomer molecule via chain growth condensation (with elimination of trimethylsilyl trifluoroethoxide), or it may react with a partially polymerized chain segmen<sup>t</sup> to advance molecular weight via a macrocondensation. The preparation of four *N*-alkyl phosphoranimine compounds and their efficiency as chain end-capping agents in polyphosphazene synthesis was briefly mentioned before [42].

*P*-tris(trifluoroethoxy)-*N*-trimethylsilyl

*N*-methylimidazole

 initiator in diglyme

phosphoran-

In this paper, we report more detailed experimental data obtained using these compounds, and present evidence of their incorporation as unique chain end groups during the formation of polyphosphazenes by the anionic route.

#### **2. Results and Discussion**

#### *2.1. Synthesis and Characterization of N-Alkyl Phosphoranimines*

The synthetic route to the four *N*-alkyl phosphoranimines prepared in this study was via the Staudinger reaction [43] is shown in Scheme 2, in which tris(2,2,2-trifluoroethyl) phosphite reacted with an alkyl azide to form the *P*-tris(2,2,2-trifluoroethoxy)-*N*-alkyl phosphoranimine with the evolution of nitrogen gas.

**Scheme 2.** Staudinger synthesis of *P*-tris(trifluoroethoxy)-*N*-alkylphosphoranimines.

The *N*-alkyl compounds formed consisted of the *N*-adamantyl, *N*-benzyl, *<sup>N</sup>*-*<sup>t</sup>*-butyl, and *N*-trityl phosphoranimines. Yields, boiling or melting points, densities, and refractive indices are shown in Table 1.


**Table 1.** Physical data for *N*-alkyl phosphoranimines.

The new compounds were characterized by NMR, fast atom bombardment, and GCmass spectrometry, with the resulting data presented in Table 2. The data obtained from the NMR spectra are consistent with the proposed structures of the four compounds, as are the mass spectrometry measurements for molecular weight of the parent ions.



\* Molecular ion; ˆ Partially decomposes on silica GC column; \*\* Molecular ion, FAB mass spec.

Elemental analyses of these compounds for found and theoretical percentages are displayed in Table 3. The determined percentages of elements from the analyses are in good agreemen<sup>t</sup> with the assigned structures.


**Table 3.** Elemental analyses of *N*-alkyl phosphoranimines.

Notes: Fluorine interferes with oxygen determination. \* Sample size insufficient for P-analysis.

> FTIR spectra further confirmed the structures with the appearance of the broad absorption bands of the P=N moieties between 1270–1310 cm<sup>−</sup>1, and the characteristic absorptions of aliphatic or aromatic protons as expected for the particular structure. The data collected are shown in Table 4, and an example IR spectrum of the *N-*adamantyl phosphoranimine is shown in Figure 1.



**Figure 1.** FT-IR spectrum of *P*-tris(trifluoroethoxy)-*N*-adamantyl phosphoranimine.

#### *2.2. Reactivity of N-Alkyl Phosphoranimines*

The phosphoranimines are air-sensitive and hydrolyze easily to the corresponding phosphoramidite esters, (CF3CH2O)2P(=O)-NHR, as confirmed by mass spectrometry and NMR. For example, the GC-mass spectrum of the *N*-adamantyl phosphoranimine in (wet) diethyl ether showed two major peaks with fragments that are consistent for the phosphoranimine (molecular ion 477), and the corresponding *N*-adamantyl phosphoramidate ester (molecular ion 395). A third peak in trace amount is in good agreemen<sup>t</sup> with the 1-azidoadamantane starting reagent, and all three species contain the adamantyl fragment (135).

The proton-coupled 31P-NMR spectrum of the GC-MS sample showed the phosphoranimine signal at −27.4 ppm, and the new species at +7.0 ppm. The new signal is a sextet with a coupling constant of 7.3 Hz, as expected for <sup>3</sup>*J*POCH coupling.

The 1H-NMR spectrum of this sample showed the ethyl ether proton signals at 1.18 ppm (t) and 3.45 ppm (q); two distinct sets of the different adamantyl protons from 1.58 to 2.10 ppm; amidate proton at 2.80 ppm (d); and trifluoroethoxy protons at 4.27 ppm (p). The amidate doublet has a coupling constant of <sup>2</sup>*J*PNH = 9 Hz. This data supports the structure assignment for the hydrolysate.

The phosphoramidate ester can arise from nucleophilic attack by water (or OH−) on the electrophilic phosphorus atom of the phosphoranimine, as shown in Scheme 3.

**Scheme 3.** Reaction of *P*-tris(trifluoroethoxy)-*N*-adamantylphosphoranimine with H2O.

Similar behavior was reported in the study of an *N*-silylated phosphoranimine [44]. This could be a general reaction of phosphoranimines bearing trifluoroethoxy (or other leaving) groups bound to the phosphorus center.

A sample of the *N*-benzyl phosphoranimine was exposed to air and was converted to a colorless crystalline solid with a δp = +8.3 ppm, which is consistent with the corresponding *N*-benzyl phosphoramidate ester, (CF3CH2O)2P = (O)-NHCH2Ph [45].

In addition to the phosphoramidate ester major product, a trace signal at −152.5 ppm was also observed. This high field resonance arises from the hexacoordinate species (CF3CH2O)6P<sup>−</sup>, in which the phosphorus center carries a formal negative charge. This compound was previously prepared and its 31P-NMR chemical shift was reported as −154.6 ppm [46]. Electronegative groups stabilize the phosphorus hexacoordinated state [47].

This species has also been observed as a trace component of polymerizing mixtures of the *P*- tris(2,2,2-trifluoroethoxy)-*N*-trimethylsilyl phosphoranimine monomer, and in unreacted fractions of tris(2,2,2-trifluoroethyl) phosphite that were distilled from reaction mixtures. It is probably formed by a thermal rearrangemen<sup>t</sup> of phosphorus compounds bearing trifluoroethoxy ligands.

A pentacoordinated phosphorus compound was observed in the 31P-NMR spectrum of a sample of the clear colorless *N*-benzyl phosphoranimine that had crystallized after several weeks in a desiccator. Its chemical shift of −72 ppm is typical of five-coordinate phosphorus compounds as previously reported [45], and sufficiently electronegative ligands tend to stabilize such species [45]. The penta-(2,2,2-trifluoroethoxy) phosphorane, P(OCH2CF3)5, was previously prepared and its 31P-NMR chemical shift reported as −76.6 ppm [46]. Another example of a penta-coordinated phosphorus compound with an *N*-benzyl ligand is (CF3)3 (F)P[N(CH3) (CH2Ph)] with a δp = −68.8 ppm [48].

The mass spectrum of the crystallized sample of the *N*-benzyl phosphoranimine showed a high mass peak of 351. The pentacoordinate phosphorus compounds shown in Scheme 4 are potential isomers with MW = 353 and are reasonable structural assignments for the species observed. They may exist in equilibrium with their phosphonium salts of the general formula, R4P+R− [45].

**Scheme 4.** Pentacoordinated phosphorus structures.

#### *2.3. Polymerization Studies*

Bulk Polymerization with Addition of *N*-alkyl Phosphoranimines

Experiments were conducted in which small samples of these compounds were added to polymerizing mixtures of the tris(2,2,2-trifluoroethoxy) phosphoranimine monomer and corresponding polyphosphazene. The initial experiment probed the reactivity of the *N*-alkyl phosphoranimine in bulk polymerization. *<sup>P</sup>*-tris(2,2,2-trifluoroethoxy)-*<sup>N</sup>*trimethylsilyl phosphoranimine monomer, 1.5 mmol (CF3CH2O)3P = N-Si(CH3)3, was treated with 1 mol% tetrabutylammonium fluoride (TBAF) in a dry NMR tube and heated at 150 ◦C for 15 min. Then, 0.6 mmol of *N*-benzyl phosphoranimine was injected by syringe, mixed, and heated for another 15 min.

The reaction was cooled to room temperature, and 31P-NMR spectra confirmed polymer formation, along with a few percent oligomers. There were two very small doublets (+8.5, −2.5 ppm, *J*PNP = 73 Hz) in the spectrum of the reaction mixture (less than 2%) which indicate formation of a phosphazene dimer as a minor side product, probably formed by a coupling reaction of monomer and the *N*-benzyl compound. A small signal at −72 ppm indicated the presence of ca. 2% of the pentacoordinate phosphorane discussed above.

The sample was then heated for an additional 15 min and final spectra recorded. A control reaction without *N*-benzyl phosphoranimine was run concurrently. The polymer samples were dissolved in diglyme and precipitated from excess cold chloroform, then re-dissolved in diglyme and precipitated again from 90/10 CHCl3/MeOH with thorough washing of the white polymer solids.

The 1H-NMR spectra of the isolated polymer in d6-acetone solution showed the polymer signal at 4.55 ppm, *n*-butyl protons from the TBAF initiator, and new signals at 7.35 ppm indicating the aromatic protons of the benzyl group (Figure 2).

**Figure 2.** 1H-NMR spectrum of *N*-benzyl-capped poly(bis(trifluoroethoxy)phosphazene).

The 31P-NMR spectrum of this sample showed the polymer signal at −7.05 ppm, but no signals for the *N*-benzyl phosphoranimine, dimer, or phosphorane. The aromatic proton signals were thus ascribed to reaction of the *N*-benzyl compound with the polymer chain. Integration of the polymer and benzyl proton signal ratio correlated with incorporation of one *N*-benzyl phosphoranimine residue per chain, indicative of a functionalized chain-end. Similar results were obtained using the other *N*-alkyl phosphoranimines. After heating

6 mmol of monomer and 1 mol% TBAF initiator for 15 min at 150 ◦C, 3 mmol of the *N*-alkyl compound (neat or in solution) was added by syringe and heating continued for a total time of 1 h.

The polymer samples were precipitated twice as described before, and the 1H-NMR spectra showed signals for polymer, the *n*-butyl groups of the TBAF, and the corresponding alkyl group. The presence of the trityl-capped chains is particularly distinctive by the multiple aromatic proton signals of the formed polymer product, as seen in Figure 3. The proton signals for the adamantyl and *t*-butyl chain-end groups are partially obscured by the TBAF initiator residue, but they can be seen and identified. The 31P-NMR spectra of the experimental polymer samples showed no signals for unreacted *N*-alkyl phosphoranimines. The control reaction of monomer and TBAF showed proton signals for polymer and *n*butyl groups, and GPC confirmed higher molecular weight. The yield of the polymer was significantly higher, indicating a more complete monomer conversion.

**Figure 3.** 1H-NMR spectrum of *N*-trityl-capped poly(bis(trifluoroethoxy)phosphazene).

The molecular weights of the polymer samples obtained show the effect of the addition of the *N*-alkyl compounds. As seen in Table 5, the molecular weight and yield of polymer were significantly reduced by the addition of the *N*-alkyl compound compared to the control.

**Table 5.** Molecular weights, dispersity, yields of *<sup>N</sup>*-capped polyphosphazenes (bulk polymerization at 150 ◦C 1% TBAF, 3 mmol *N*-alkyl phosphoranimine addition, heated for 1 h).


In related experiments, bulk polymerizations with monomer-equivalent additions of *N*-benzyl phosphoranimine (NBP) at specific intervals during 150 ◦C polymerization using 1 mol% TBAF initiator showed that the polymer molecular weight was terminated

upon addition compared to controls (\*) with no *N*-benzyl compound (Table 6). Neither conversion nor molecular weight increased after addition of the NBP.

**Table 6.** Effect of addition time of *P*-Tris(2,2,2-trifluoroethoxy)-*N*-trimethylsilyl phosphoranimine on conversion and molecular weight (bulk at 150 ◦C, 1% TBAF).


**\***controls, no *N*-benzyl phosphoranimine.

The effect of various amounts of *N*-benzyl phosphoranimine was examined by preparing samples containing monomer, 1 mol% TBAF initiator, and *N*-benzyl compound in the ratios shown in Table 7. After heating at 150 ◦C for 45 min, the molecular weights of the polymers obtained were measured by GPC and conversion determined by 31P-NMR spectra integration. These data show that at least 2% *N*-benzyl phosphoranimine was needed to effectively limit polymer molecular weight and conversion of monomer, with 20% inhibiting polymerization.

**Table 7.** Effect of various monomer: *N*-benzyl phosphoranimine ratios on conversion and molecular weight.


The polymer molecular weight, dispersity, and yield are decreased by the *N*-alkyl phosphoranimine addition. However, the data also show that perfect stoichiometric agreemen<sup>t</sup> between the amount of the capping agen<sup>t</sup> and degree of polymerization (DP) was not always observed. For example, the 20:1 and 50:1 samples have DPs of about 20 and 50, respectively, but the other ratios exhibit some variability. The lower than expected DP of the 100:1 sample may arise from other transfer reactions in the system.

Using the data from Table 7, a graph was constructed, as shown in Figure 4. The following Mayo equation was used to determine the chain transfer to the capping agen<sup>t</sup> constant, Ctr,X, and the constant for transfer to monomer, Ctr,M:

$$1/\text{DP}\_{\text{n}} = \text{C}\_{\text{tr},\text{M}} + \text{C}\_{\text{tr},\text{X}} \text{ [X]/[\text{M}]}.$$

From the graph, the slope of the line is the chain transfer constant to capping agen<sup>t</sup> and the y-intercept is the chain transfer to monomer constant. From the equation, the values for Ctr,X and Ctr,M are 8.54 × 10−<sup>1</sup> and 7.2 × <sup>10</sup>−3, respectively. There was a very small signal from trimethylsilyl protons in the 1H-NMR spectrum of precipitated polymer at 0.12 ppm (Figure 2), which indicated that some chains could be terminated by trimethylsilyl groups, albeit in trace amount.

**Figure 4.** Mayo plot of 1/DP as function of [X]/[M] ratio.

A final experiment was conducted in which a 1:1.7 mol mixture of monomer and *N*adamantyl phosphoranimine with 1% (mol) TBAF was heated for several hours at 150 ◦C, well beyond the normal time required for complete conversion of monomer to polymer in the presence of the fluoride. Analysis of the reaction mixture by 31P-NMR showed that only 13% conversion of the monomer occurred, producing polymer and oligomers, with 87% of the spectrum signals arising from unreacted monomer and the *N*-adamantyl compound. Under these forcing conditions, the polymerization was retarded to a major extent, but not completely inhibited by the *N*-adamantyl phosphoranimine. The amount of polymer in the reaction mixture was less than 7% and could not be isolated for inspection of end groups.

Significantly, no phosphorus doublets were observed in the 31P-NMR spectrum, thus, the monomer and *N*-adamantyl compound did not react to form a phosphazene dimer. This result and that of the end-capping experiments indicate that the *N*-adamantyl phosphoranimine reacts preferentially with active chain ends.

The active species, particularly in the early chain-growth stage of the polymerization, should be an anionic polymer chain-end [49]. The phosphorus atom of the *N*-alkyl phosphoranimine has similar reactivity as in the *N*-silylated monomer (Ctr,X ~1) but forms an unreactive group. The stable alkyl group of the new phosphoranimine leads to termination of the chain, whereas the labile silyl group of the monomer results in continued propagation of the chain, as shown in Scheme 5.

Chain termination with these compounds is a degradative transfer reaction in which an active chain is terminated and a new active species, trifluoroethoxide anion, is produced. The number of active species in the system is unchanged, and the anion can initiate a new chain, as reported in an earlier study [49,50].
