Iridium(III) hydrido N-heterocyclic carbene-phosphine complexes as catalysts in magnetization transfer reactions.

The hyperpolarization (HP) method signal amplification by reversible exchange (SABRE) uses para-hydrogen to sensitize substrate detection by NMR. The catalyst systems [Ir(H)2(IMes)(MeCN)2(R)]BF4 and [Ir(H)2(IMes)(py)2(R)]BF4 [py = pyridine; R = PCy3 or PPh3; IMes = 1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene], which contain both an electron-donating N-heterocyclic carbene and a phosphine, are used here to catalyze SABRE. They react with acetonitrile and pyridine to produce [Ir(H)2(NCMe)(py)(IMes)(PPh3)]BF4 and [Ir(H)2(NCMe)(py)(IMes)(PCy3)]BF4, complexes that undergo ligand exchange on a time scale commensurate with observation of the SABRE effect, which is illustrated here by the observation of both pyridine and acetonitrile HP. In this study, the required symmetry breaking that underpins SABRE is provided for by the use of chemical inequivalence rather than the previously reported magnetic inequivalence. As a consequence, we show that the ligand sphere of the polarization transfer catalyst itself becomes hyperpolarized and hence that the high-sensitivity detection of a number of reaction intermediates is possible. These species include [Ir(H)2(NCMe)(py)(IMes)(PPh3)]BF4, [Ir(H)2(MeOH)(py)(IMes)(PPh3)]BF4, and [Ir(H)2(NCMe)(py)2(PPh3)]BF4. Studies are also described that employ the deuterium-labeled substrates CD3CN and C5D5N, and the labeled ligands P(C6D5)3 and IMes-d22, to demonstrate that dramatically improved levels of HP can be achieved as a consequence of reducing proton dilution and hence polarization wastage. By a combination of these studies with experiments in which the magnetic field experienced by the sample at the point of polarization transfer is varied, confirmation of the resonance assignments is achieved. Furthermore, when [Ir(H)2(pyridine-h5)(pyridine-d5)(IMes)(PPh3)]BF4 is examined, its hydride ligand signals are shown to become visible through para-hydrogen-induced polarization rather than SABRE.

Introduction

Hyperpolarization (HP) methods are being used to improve the sensitivity of NMR and magnetic resonance imaging to substrate detection.[1] Signal amplification by reversible exchange (SABRE) corresponds to one such method in which the nuclear spin order from para-hydrogen (p-H2) is used to sensitize a substrate. In this case, the process does not involve a change in the chemical structure of the hyperpolarized substrate, as demonstrated in the more traditional p-H2-induced polarization (PHIP) technique pioneered by Weitekamp and Eisenberg.[2−4] In PHIP, detection of a product that contains two nuclei that were originally located in a single p-H2 molecule is achieved and therefore a simultaneous change in the chemical identity is required. Nonetheless, this approach is very powerful, and detection of the true reaction intermediates has been achieved,[5−7] the development of NMR theoretical approaches facilitated,[8,9] and the monitoring of hydrogenation products in organic transformation enabled.[10,11] The latter situation has led to significant interest in PHIP as a tool for the in vivo monitoring of living systems.[12] Furthermore, more recently, Koptyug et al. have used the same methods to follow a number of heterogeneous reactions.[13−15]

In SABRE, while the HP feedstock is still p-H2, a catalyst is now used as a scaffold to temporarily bind both H2 and the substrate.[16,17] Consequently, after ligand dissociation, there is no change in the chemical identity of the hyperpolarized species. Earlier studies used [Ir(COD)(PCy3)(py)]BF4, which contains the sterically bulky electron-donating phosphine PCy3 as a suitable catalyst.[17] More recently, greater HP efficiency has been demonstrated for the related 1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene (IMes) complex, [Ir(COD)(IMes)Cl].[18] Recent reviews illustrate the context of both SABRE and PHIP.[19,20]

The iridium–phosphine system [Ir(COD)(PCy3)(py)]+ is known as Crabtree’s catalyst and is usually associated with a role as either a hydrogenation or deuterium-transfer catalyst.[21] Evidence obtained through studies of the effect of the ligand electronic parameters by Tolman,[22] computation,[23] and calorimetric analysis of the bond strengths[24] has revealed that N-heterocyclic carbenes (NHCs) are more electron-donating than phosphines.[25] Furthermore, the work of Nolan and Buriak has shown that the stability and activity of Crabtree’s catalyst can be improved with such ligands.[26] They achieve this by combining a bulky NHC, such as IMes, with an appreciably encumbered phosphine (PPh3, PBn3, PMe2Ph, and PnBu3) to deliver the most robust and effective deuterium exchange catalyst.[27]

Upon reaction of [Ir(COD)(PCy3)(py)]BF4 with pyridine and p-H2, [Ir(H)2(PCy3)(py)3]BF4 formed and proved to function very effectively as a SABRE catalyst.[17] Indeed, measurements under SABRE with its counterpart [Ir(H)2(IMes)(py)3]Cl yielded proton signal intensity gains in free pyridine of greater than 3000-fold.[18] These complexes completed the HP transfer step in a low magnetic field in order to facilitate the interactions necessary for spin-order transfer.[8] A growing number of studies have used the SABRE effect.[28−34] Both of these systems are examples of catalysts in which it is magnetic inequivalence in the equatorial plane that provides the necessary symmetry breaking that activates the two nuclei that were originally located within p-H2 and now reside on the metal as a pair of dihydride ligands. Hence while these systems contain three potential polarization acceptors in the form of pyridine, only the two that are trans to hydride fulfill this role and, consequently, the effective spin system involved in transfer is made up of just 2 hydride and 10 pyridine protons. We aim here to show that it is possible to improve on the efficiency of this polarization transfer catalyst by sharing the p-H2 spin order with fewer protons. While this has already been demonstrated by using deuterated substrates,[18,35] a redesign of the catalyst to include fewer exchangeable ligand sites offers another route to achieving this goal. In this paper, we describe how the combination of an electron-donating NHC and a phosphine produces a new set of high-activity catalysts for SABRE. The phosphines employed are PCy3 and PPh3, and they provide access to the well-defined complexes [Ir(H)2(IMes)(MeCN)2(PPh3)]BF4 and [Ir(H)2(IMes)(MeCN)2(PCy3)]BF4. The molecules that are hyperpolarized in this study correspond to pyridine and acetonitrile. It has previously been suggested that the HP efficiency of the SABRE catalyst is linked to both the lifetime of the metal complex and the strength of the magnetic field where polarization transfer occurs.[8] These concepts are further tested here by using a flow-polarization apparatus that has been previously described.[18,36] In particular, we explore the ligand-exchange processes and demonstrate how a number of new species are detectable, including characterization of a C–H bond activation product that acts as a resting state within the SABRE process. In this case, it is products of the type [Ir(H)2(NCMe)(py)(IMes)(PR3)]BF4 that are dominant in the catalyst medium. The effect of the magnetic field experienced by the catalyst at the point of polarization transfer (the polarization transfer field, or PTF) is monitored experimentally and shown to vary with the identity of the catalyst and detected substrate. We also demonstrate how breaking of the original p-H2 molecule symmetry through the chemical shift difference in the dihydride product [Ir(H)2(NCMe)(py)(IMes)(PR3)]BF4 enables the wide transfer of HP within the parent complex and hence their detection as examples of hyperpolarized reaction intermediates.

Experimental Section

[Ir(COD)(PCy3)(py)]PF6 (1) was obtained from Strem. [Ir(COD){CO(CH3)2}(IMes)]BF4 (2), [Ir(H)2(NCMe)2(IMes)(phosphine)]BF4 (3a and 4a), [Ir(H)2(py)2(IMes)(PPh3)]BF4 (3b), and [Ir(H)2(NCMe)(py)(IMes)(PPh3)]BF4 (3c) were prepared based on published literature methods, as described in the Supporting Information.[37] All NMR measurements were recorded on a Bruker Avance III series 400 or 500 MHz system. Samples were now typically dissolved in 3 mL of methanol-d4 (5 mM) and contained pyridine (5 or 20-fold excess). Single crystals of both 3c and 5 were grown from a benzene solution and analyzed on an Oxford Diffraction SuperNova diffractometer. The crystals were kept at 110.00(10) K during data collection.

[Ir(H)2(NCMe)2(IMes)(PPh3)]BF4 (3a)

1H NMR (400 MHz, CDCl3, 298 K): δ −21.45 (d, 2H, JHP = 17.27 Hz), 1.65 (s, 6H, NCCH3), 2.11 (s, 12H, CH3 of IMes), 2.38 (s, 6H, CH3), 6.95–7.34 (m, 21H, CH, PPh3, IMes). 13C{1H} NMR (101 MHz, CDCl3, 298 K): δ 1.0 (NCCH3), 17.9, 21.2 (CH3), 117.83 (NCMe), 122.2, 122.6 (NCH), 127.9, 128.9, 129.6 (d, CH, JCP = 11.3 Hz), 129.8, 132.7, 133.1 (CH), 133.9, 134.4 6 (d, CH, JCP = 11.0 Hz), 135.9, 137.6, 138.8 (−C=), 164.1 (d, NCN, JCP = 114.7 Hz). 31P{1H} NMR (162 MHz, CDCl3, 298 K): δ 18.96. 11B NMR (160 MHz, CD3CN, 298 K): δ −1.40 (11BF4, 81%), −1.42 (10BF4, 19%). 19F NMR (470 MHz, CD3CN, 298 K): δ −152.89 (10BF4, 19%), −152.94 (11BF4, 81%). ESI MS: 802.29 [M+ – NCMe].

Compound 3b

1H NMR (400 MHz, methanol-d4, 296 K): δ −22.74 (d, 2H, JHP = 18.4 Hz), 1.97 (s, 12H, CH3 of IMes), 2.31 (s, 6H, CH3), 6.75–7.29 (m, 27H, CH, PPh3, IMes, py), 7.55 (t, 2H, JHH = 7.2 Hz, para 1H py), 7.55 (d, 4H, JHH = 4.5 Hz, ortho 1H py). 13C{1H} NMR (101 MHz, methanol-d4, 298 K): δ 17.9, 21.2 (CH3), 122.0, 122.3 (NCH), 123.6 (CH, py), 127.8, 128.7, 129.5 (d, CH, JCP = 11.3 Hz), 129.7, 132.5, 133.1 (CH), 133.9, 134.4, (d, CH, JCP = 11.0 Hz), 135.7 (CH, py), 135.9, 137.6, 138.8 (−C=), 149.7 (−C=, py), 164.1 (d, NCN, JCP = 114.7 Hz). 31P{1H} NMR (162 MHz, methanol-d4, 298 K): δ 22.1. ESI MS: 840.31 [M+ – py].

Compound 3c

This complex was prepared in situ in an NMR tube by adding 5 equiv of acetonitrile to a solution of 3b or 5 equiv of pyridine to a solution of 3a. 1H NMR (500 MHz, acetonitrile-d3, 298 K): δ −22.30 (dd, 1H, trans to pyridine, JHH = 7.21 Hz, JHP = 19.44 Hz), −21.10 (dd, 1H, trans to NCCH3, JHH = 7.21 Hz, JHP = 16.09 Hz), 1.52 (s, 3H, coordinated NCCH3), 1.80 (s, 6H, CH3), 2.00 (s, free NCCH3), 2.15 (s, 6H, 2 × −CH3), 2.36 (s, 6H, 2 × −CH3), 6.81 (d, 1H, −CH=, JHH = 0.77 Hz), 6.97 (d, 1H, CH, JHH = 0.77 Hz), 7.05–7.49 (m, 21H, CH, PPh3, IMes), 7.78 (t, 1H, para 1H pyridine), 8.72 (d, 2H, ortho 1H pyridine. 31P{1H} NMR (161 MHz, CD3CN, 298 K): δ 22.98. 11B NMR (160 MHz, CD3CN, 298 K): δ −1.05. 19F NMR (470 MHz, CD3CN, 298 K): δ −154.69 (10BF4, 19%), −154.74 (11BF4, 81%). 15N NMR (40.54 MHz, CDCl3, 263 K): δ 177.8 (coordinated NCCH3), 195.8 (NCN, IMes), 238.1 (coordinated pyridine).

[Ir(H)2(NCMe)2(IMes)(PCy3)]BF4 (4a)

1H NMR (400 MHz, methanol-d4, 296 K): δ −22.24 (d, 2H, JHP = 18.7 Hz), 1.32–1.96 (m, 30H CH2 of PCy3), 2.16 (s, 6H, CH3 of acetonitrile), 2.21 (s, 12H, CH3 of IMes), 2.38 (3H, m, CH of PCy3), 7.08 (m, 2H, CH, IMes. 13C{1H} NMR (101 MHz, CDCl3, 298 K): δ 1.4 (NCCH3), 16.0, 19.8 (CH3), 25.8, 26.2, 27.0, 28.4, 30.5, 32.8, 33.1 (CH2), 118.8 (NCMe), 122.5, 125.1 (NCH), 128.5 (CH), 135.8, 137.9, 138.5 (−C=), 173.9 (d, NCN, JCP = 118.0 Hz). 31P{1H} NMR (162 MHz, methanol-d4, 296 K): δ 18.24. ESI MS: 779.40 [M+ – 2NCMe].

NMR Data for [Ir(H)2(py)2(IMes)(PCy3)]BF4 (4b)

1H NMR (400 MHz, methanol-d4, 296 K): δ −23.83 (d, 2H, JHP = 18.7 Hz), 1.32–1.96 (m, 30H, CH of PCy3), 2.19 (s, 12H, CH3 of IMes), 2.41 (s, 6H, CH3 of IMes), 7.21 (s, 4H, CH, IMes), 7.66 (t, 2H, meta 1H pyridine), 8.01 (s, 2H, CH, IMes), 8.18 (t, 1H, para 1H pyridine), 8.95 (d, 2H, ortho 1H pyridine). 31P{1H} NMR (162 MHz, methanol-d4, 298 K): δ 21.9.

NMR Data for [Ir(H)2(NCMe)(py)(IMes)(PCy3)]BF4 (4c)

1H NMR (400 MHz, methanol-d4, 296 K): δ −23.42 (dd, 1H, JHP = Hz), −21.98 (dd, 1H, JHP = Hz), 1.13–1.73 (m, 30H, CH2 of PCy3), 1.83 (s, 6H, CH3 of IMes), 2.15 (s, 6H, CH3 of IMes), 2.19 (s, 3H, CH3, acetonitrile), 2.38 (s, 6H, CH3 of IMes), 6.90 (s, 2H, −CH=, IMes), 7.07 (s, 2H, −CH=, IMes), 7.16 (t, 2H, JHH = 6.3 Hz, meta 1H pyridine), 7.18 (s, 2H, −CH=, IMes), 7.81 (t, 1H, JHH = 8.3 Hz, ortho 1H pyridine), 8.23 (d, 2H, JHH = 4.3 Hz, para 1H pyridine). 31P{1H} NMR (162 MHz, methanol-d4, 298 K): δ 22.73.

Results and Discussion

Complexes 3a, 3b, and 4a were prepared and characterized by multinuclear NMR spectroscopy. Key resonances are listed in the Experimental Section. The hydride ligands in these species appear as simple phosphorus-coupled doublets at δ −21.40, −22.70 and −22.24, in methanol-d4 at 298 K and at δ −21.45, −22.80, and −22.42 in CDCl3. These resonances are used to monitor the chemical evolution of these systems as they react with pyridine and acetonitrile. Two control samples were prepared to do this. The first sample contained 3a in methanol-d4 and 5 equiv of pyridine and 3 bar of H2. The second contained 3b in CDCl3 and 1 equiv of NCMe and 3 bar of H2. In both cases, the hydride signals of 3a or 3b were replaced by a common species that yielded a pair of coupled hydride ligand signals at δ −21.02 and −22.18. This product, 3c, is the sole product in both reactions and confirms that 3c is thermodynamically more stable than either 3a or 3b, as illustrated by Scheme 1 and Figure 1.

Scheme 1

Equilibration of 3a or 3b with Pyridine and Acetonitrile Leading to 3c as the Dominant Product

Figure 1

Hydride region of two 1H NMR spectra used to illustrate the conversion of 3a into 3c upon the addition of pyridine. These traces were recorded at 273 K in CDCl3.

The peak integrals for these two hydride resonances of 3c exhibit the expected 1:1 ratio, and we note that a coordinated NCMe signal appears at δ 1.48. In 3c, the methyl substituents of the IMes ligand appear as three distinct sets, each of six protons, in accordance with rapid rotation about the metal–carbene and N–C bonds (see the Experimental Section). The hydride ligand resonances of 3c are also well-resolved and connect, according to nuclear Overhauser effect methods, with proton signals at δ 6.99 and 1.48, which are attributed to bound pyridine and CH3CN ligands, respectively. These connections arise because of through-space interactions and confirm the cis-ligand arrangement that is shown in Figure 1. In contrast, when a 1H–15N HMQC NMR spectrum is recorded for this system (in CDCl3), where the 15N label is present at natural abundance, as shown in Figure 2, a different set of connections become visible. Now the hydride signal that appears at lower field connects with a coordinated acetonitrile 15N signal at δ 177.8 through a 15.3 Hz JH15N splitting; it also connects to its CH3 partner at δ 1.48. The second, higher field, hydride signal at δ −22.18 possesses a JHN splitting of 18.5 Hz and is located trans to pyridine. The 15N resonance of the associated coordinated pyridine ligand of 3c appears at δ 239.1 in this NMR spectrum. In addition to these two connections, the δ 6.99 (JHN = 10.6 Hz) signal of the −CH=CH– protons in the imidazolium ring of the IMes ligand shows a strong correlation peak to a third 15N signal at δ 195.7 in this experiment. These measurements therefore locate key 15N ligand resonances in 3c and give a high level of confidence in our product assignment; we note that these measurements were achieved in only a few hours without 15N labeling. The characterization of 3c was further confirmed by X-ray crystallography at 110 K. Figure 3 shows the corresponding ORTEP for 3c, while Table 1 lists key bond lengths and angles.

Figure 2

2D 1H–15N HMQC NMR spectrum of unlabeled 3c recorded at 263 K in CDCl3, with the resulting 15N data providing diagnostic information on the ligand arrangement, as discussed in the text (relative expansions in the boxed areas are indicated).

Figure 3

ORTEP diagrams of the cations 3c and 5 as determined by X-ray diffraction.

Table 1

Bond Distance (Å) and Angles (deg) for Complexes 3c and 5

	3c	5
Ir–P1	2.2832(9)	2.3347(9)
Ir–H1	1.603	1.604
Ir–C1	2.068(4)	2.046(3)
Ir–C4		2.101(3)
Ir–N3	2.160(3)	2.090(3)
Ir–N4	2.107 (5)	2.154(3)
C1–Ir–P1	168.29(11)	165.48(10)
C4–Ir–P1		94.00(10)
C1–Ir–C4		80.24(13)
N3–Ir–N4	91.6(2)	85.55(11)
N3–Ir–H1	176.58
N4–Ir–H1	175.85	179.16
N3–Ir–C1	93.96(13)	94.82(12)
N4–Ir–C4	97.54(14)	92.06(12)01
P1–Ir–N3	94.14(7)	91.41(8)
P1–Ir–N4	90.67(11)	100.54(8)

Characterization of 4c

A series of related studies were then undertaken on the PCy3 analogues of 3. These studies revealed that, upon dissolution of 4a in methanol-d4 and the addition of 5 equiv of pyridine and 3 bar of H2, 4c was formed in a manner similar to that of 3c. NMR data for 4a–4c are presented in the Experimental Section.

It is well-known that species such as 3c and 4c can undergo C–H activation of an o-methyl group of the mesityl ring after ligand loss. Aldridge and co-workers[38] reported such an observation in the solid-state chemistry of Ir(IMes)(IMes′)(H)Cl, while Torres et al.[37] demonstrated that [Ir(H)2(IMes)(NCMe)2(PPr3)]PF6 forms [IrH(IMes′)(NCMe)2(PPr3)]PF6 upon treatment with ethylene. In this case, upon exposure of the cyclometalated complex to H2, the dihydride is re-formed. We were therefore not surprised when a sample of 4a that was placed under N2 underwent H2 reductive elimination to form a new product. This species proved to correspond to [IrH(IMes′)(NCMe)2(PCy3)]BF4 (5). Isolation of complex 5 led to the collection of X-ray-quality crystals, which yielded the structure shown in Figure 3. The Ir–phosphine bond lengths in 5 and the related complex [IrH(IMes′)(NCMe)2(PPr3)]PF6 are 2.3347(9) and 2.3367(17) Å and hence indistinguishable. In contrast, the corresponding Ir=C1 bond lengths are slightly different at 2.046(3) vs 1.973(7) Å, in accordance with the greater electron-donating power of PCy3 vs PiPr3. The corresponding Tolman cone angles change from 170° to 160° for these ligands. The difference in the phosphine steric bulk changes the structure such that in 5 the equatorial N–Ir–C and N–Ir–H bond angles are larger than those for [IrH(IMes′)(NCMe)2(PPr3)]PF6.

Kinetics of Ligand Exchange in 3c and 4c

It should be clear that the dominant species present in these solutions is 3c or 4c. A series of 1H-selective EXSY NMR experiments were therefore performed in order to probe the ligand-exchange processes undergone by 3c and 4c. These data are presented in the Supporting Information (section 4), but our findings are summarized here. In the first instance, 3c was observed to undergo magnetization transfer between free and bound pyridine, free and bound acetonitrile, and the two hydride sites, as illustrated in Scheme 2. When the experimentally observed rate of transfer of magnetization from bound pyridine into free pyridine was monitored as a function of the pyridine concentration, saturation behavior indicative of a dissociative process was evident at high pyridine concentrations. The limiting rate constant for magnetization transfer into free pyridine was determined to be 1.15 ± 0.1 s–1. However, when a similar sample was examined, where the initial concentrations of 3c and pyridine were 0.0152 and 0.0342 M, respectively, the rate of magnetization transfer into free pyridine reduced to zero with increase in the acetonitrile concentration. This information suggests that it is actually acetonitrile dissociation that accounts for the observed pyridine magnetization transfer. Hence, this process happens by secondary ligand exchange through unstable 3b as indicated in Scheme 2 rather than the direct dissociation of pyridine. This deduction was confirmed when the effect of the acetonitrile concentration on the rate of transfer of bound to free acetonitrile was explored. Now exchange rate saturation was again seen, but the maximum rate constant for ligand loss proved to be 1.95 ± 0.1 s–1. This value is twice that of the limiting pyridine rate, a link that is expected if 3b were to be involved because once formed there is now an equal probability of losing the magnetically labeled pyridine or the freshly bound pyridine ligand. Furthermore, the observed rate constant for hydride site interchange is smaller than this and is affected by the ligand excess, in accordance with the 16-electron reaction intermediate retaining unique hydride ligand identities as expected.[39]

Scheme 2

Magnetization Transfer Pathways Revealed through EXSY Examination of 3 (R = Ph) or 4 (R = Cy) with Pyridine and Acetonitrile as a Function of the Reagent Concentration

The color scheme shows how labeled magnetization can propagate through the system, based on the indicated dissociative reactions. Solvation of the intermediates is to be expected, and distinct hydride ligand sites are retained in these complexes.

In contrast, when 4c is examined, a limiting reaction rate is again reached as both the pyridine and acetonitrile concentrations are increased. However, now the limiting rate constant for pyridine loss is 3.1 ± 0.1 s–1, while that for acetonitrile loss is half that, at 1.4 ± 0.1 s–1. Hence, for 4c, the mechanism seems to be inverted, with the main ligand loss process corresponding to loss of pyridine. Consequently, trapping with acetonitrile now leads to 4a, where two identical equatorial ligands reside. The temperature dependence of the observed magnetization transfer rates between the bound and free ligands is detailed in the Supporting Information.

It is interesting to note that when the hydride ligands of 3c are probed by EXSY spectroscopy, no exchange into free H2 is observed. This situation marks a significant difference in behavior from that previously reported for 1 and 2, where H2 loss is relatively facile. These data therefore suggest that 3c and 4c are suitable as SABRE catalysts because they undergo ligand exchange on the NMR time scale. We now describe a series of SABRE-type measurements that are designed to explore this hypothesis.

Using 3 and 4 as SABRE Catalysts

In order to monitor these complexes under SABRE conditions rather than PHIP, a flow apparatus that has been previously described was employed.[36] This equipment allows for the automated collection of hyperpolarized NMR data. A series of identical solutions were prepared, containing the same amount of specified catalyst (1, 2, 3a, 3b, and 4a) and a 20-fold excess of pyridine. The total 1H NMR signal enhancement of the three proton sites of pyridine was then plotted as a function of the strength of the PTF. Table 2 summarizes these data, while Figure 4 illustrates a typical 1H NMR spectrum of the organic region to demonstrate the impact of SABRE and Figure 5 shows the corresponding PTF plot for the signal amplification of the meta resonance of pyridine as a function of the catalyst identity.

Table 2

Total Pyridine Proton Signal Enhancement (Metal Concentration 5.5 mM; 20-Fold Excess of Pyridine) Measured in Conjunction with the Described SABRE–Flow Apparatus for the Indicated PTF with the Corresponding Experimentally Determined Rate of Ligand Loss

catalyst precursor	field/G	absolute value	ligand loss rate constant (s–1)
1	40/90	130	0.46
2	110	646	6.26
3a	140	405	1.94
3b	130	372	1.94
4	130	685	3.1

Figure 4

(a) One-scan 1H NMR spectrum showing the hyperpolarized pyridine signals produced under SABRE with 4. (b) Thermal control spectrum illustrating the scale of signal enhancement.

Figure 5

Plot showing how the PTF controls both the degree of pyridine meta proton signal enhancement and the phase of the signal for transfer in a methanol-d4 solution using 2 (blue), 1 (red), 3c (purple), and 4c (green) SABRE catalysts when 1 equiv of NCMe and 20 equiv of pyridine are present in solution.

Surprisingly, the level of proton signal enhancement observed using 4c is comparable to that found for 2 even though the ligand-exchange rates are slower. Additionally, the level of signal enhancement seen in the proton signals of acetonitrile, phosphine, and IMes ligands of 4c is substantial, albeit at a lower level than that of pyridine. We further note that when 3c and 4c are employed as magnetization transfer catalysts, transfer into the protons of NCMe is observed. The level of enhancement seen for the acetonitrile protons in these systems reaches a maximum with a PTF of 20 G, as shown in Figure 6.

Figure 6

Plot showing the PTF dependence on the absolute signal enhancement observed for the protons of (■) acetonitrile (11-fold excess) and (○) pyridine (31-fold excess), produced in a methanol-d4 solution using the SABRE catalyst 3c at 291 K.

Exploring the Reactions of 3c and 4c by PHIP

This involved taking a solution of 3a, with added pyridine, under p-H2, and collecting a 1H NMR spectrum at 298 K using a 45° excitation pulse. Now strong PHIP was seen for the signals that arise from the two hydride ligands of 3c, as shown in Figure 7. No signals for either 3a or 3b were visible in this NMR spectrum, and these data are consistent with the previous section, which detailed how 3c is the thermodynamic product of the reaction. These NMR spectra, however, do reveal the presence of a further minor reaction product that is not readily visible without PHIP. This product appears as a pair of mutually coupled hydride signals at δ −11.89 (ddd, JHH = 4.8, JP = 23.6, and JP = 133.8 Hz) and δ −20.45 (multiplet, JHH = 4.5, JP = 16.0, and JP = 11.3 Hz), which possess couplings to two inequivalent 31P centers. These centers were located at δ 2.87 (trans to the hydride resonating at δ −11.89) and 6.30 (cis to both hydrides) in the corresponding 2D 1H–31P NMR correlation. The corresponding 1H–15N HMQC NMR experiment revealed that the hydride signal at δ −20.45 connected with an acetonitrile 15N signal at δ 196.7 through a trans coupling. These results confirm that this additional species is the ligand redistribution product [Ir(H)2(NCMe)(IMes)(PPh3)2]BF4 (6), with a cis–cis hydride and phosphine ligand arrangement.

Figure 7

1H NMR spectrum of the hydride region of a methanol-d4 solution of 3c under p-H2, revealing the pairs of hydride resonances for 3c (major) and 6 (minor).

Exploring the Reactions of 3c and 4c by SABRE

When an analogous sample was examined using the flow apparatus, after a p-H2 exposure time of 30 s, the SABRE effect was observed in some of the resonances of 3c and 6. For example, the signal for the o-phenyl proton of the phosphine ligand of 3c now appears as an emission peak, a triplet at δ 7.05, which simplifies upon 31P decoupling into the expected doublet. A weaker signal at δ 6.99 is visible for an o-phenyl proton of the phosphine in 6. The signal intensities seen for these hyperpolarized ligand resonances proved to vary with the PTF used for SABRE and, in this case, were maximized when set to 50 and 90 G, respectively. The o-pyridine proton signal for 3c can also be seen as a SABRE-enhanced signal at δ 7.78 in these measurements. Upon examination of the region around δ 2.0, hyperpolarized signals are also detected at δ 2.05, 1.54, and 1.32 for free CH3CN and bound CH3CN in 3c and 6, respectively. These resonances proved to have maximum intensity when the PTF was 20 G. Additionally, when the PTF field was increased to 140 G, signals for the corresponding methyl protons of the IMes ligands in 3c and 6 were enhanced; they appear at δ 1.88, 2.13, and 2.39 for 3c and at δ 2.36 and 3.05 for 6. These observations have therefore demonstrated that the SABRE effect can be used to observe key resonances in these complexes at high sensitivity. They also reveal that the PTF can be used to select key resonances.

Exploring the Reactions of 3c by SABRE While Employing a Range of Deuterated Ligands

We have built further on these SABRE studies by reexamining these reactions in conjunction with deuterium labeling, with the aim of making the ligand resonances even more intense. For example, when pyridine-d5, rather than pyridine-h5, is used, the SABRE-enhanced signals seen for the o-phenylphosphine protons of 3c and 6 proved to be reduced in intensity by around 80%. In contrast, the corresponding signals for the CH3CN ligands of 3c and 6 were, however, now 20% stronger, with the methyl resonances of their IMes groups showing an 8 fold increase under optimum conditions rather than the 4-fold seen in the experiments with pyridine-h5. The signal enhancements for the hydride ligand signals of 3c were also increased by 20% over those seen with pyridine-h5. The effects of deuteration are therefore substantial and confirm that, by a reduction in the number of protons in the spin system, the signal enhancement can be focused onto the remaining resonances. For the phosphine, however, the opposite is true, and hence we can deduce that its signals are actually enhanced via transfer from the protons of pyridine.

Interestingly, when a mixture of pyridine-h5 and pyridine-d5 is employed, the hydride ligands of 3b now become visible as strongly PHIP-enhanced signals. These hydride resonances are strongly amplified by virtue of the creation of a small chemical shift difference between them; the result of this process is the creation of two AB-type hydride resonances. Consequently, when a 45° excitation pulse is employed, these signals dominate, although the signals for 3c and 6 are still visible. Under SABRE conditions, enhanced signals for 3b are also visible at δ 7.09 (the o-phenyl proton of the phosphine), at δ 7.98 (the o-pyridine proton), and at δ 2.16 for its IMes ligand in this experiment.

Furthermore, when PPh3-d15 is employed instead of PPh3-h15, the hydride ligand signals of 3c and 6 appear to be 27% larger and 26% smaller, respectively. One further benefit of deuteration is that the 1H resonances of the bound pyridine-h5 ligands in these species can now be clearly seen because there is no longer any overlap with what would be PPh3-h15 resonances. The important aspects of this behavior are illustrated in Figure 8. We note, however, that separation of these signals can also be achieved by varying the PTF and, consequently, either route can be used to confirm the resonance assignment. This means that even if deuteration is not chemically viable, characterization is still possible.

Figure 8

Plot of variation of the NCMe 1H signal intensity versus PTF when a 3c-derived catalyst system is employed in conjunction with the specified deuterium labeling. A total of 3 mL of a methanol-d4 solution of 3c (5.5 mM) with an 11-fold excess of NCMe and a 31-fold excess of pyridine relative to metal was employed in these room temperature measurements.

We have also studied the effect of deuteration on the polarization of free NCMe. When pyridine-d5 was used instead of pyridine-h5, the enhancement level of the NCMe protons is 25 times higher. Furthermore, if a IMes-d22 ligand is used, this increases further to an absolute enhancement of 49-fold. When both IMes-d22 and pyridine-d5 are employed, the enhancement level rises yet further to 117-fold. These data are illustrated in Figure 8 and therefore demonstrate that it is possible to produce a viable NCMe HP transfer catalyst. We note that the cyanide function plays a role in many important drugs such as AstraZeneca’s Anastrozole, which is marketed as Arimidex.[40]

We further note that when p-H2 is used to examine the original solution of 3a in the presence of a very small amount of pyridine (0.1-fold based on iridium) with methanol-h4 (120-fold) in methanol-d4, while weak signals for 3a dominate, PHIP-enhanced signals are seen for both 3c and 6 in addition to those of two further products. These additional products yield pairs of hydride signals at δ −20.56 (dd, JHH = 6.1 Hz, JPH = 12.9 Hz) and −22.54 (dd, JHH = 5.8, JPH = 18.6 Hz) (3d) and at δ −20.81 (dd, JPH = 18 Hz) and δ −21.88 (dd, J = 17 Hz) (3e). They are therefore both monophosphine-containing. Species 3d was assigned as [Ir(H)2(MeOH)(py)(IMes)(PPh3)]+ on the basis of a series of multinuclear NMR observations. It is worth noting, in the corresponding 1H–1H COSY experiment, the hydride signal of 3d at δ −20.56 coupled with a hydride signal at δ −22.54 and a −CH3 proton signal of a bound methanol-h4 ligand at δ 2.99. In the corresponding 1H–31P HMQC NMR experiment, a 31P signal at δ 6.54 was seen for 3d. In contrast, 3e is [Ir(H)2(NCMe)(py)2(PPh3)]+, and while its NMR signals are too weak for full characterization, a proton signal of the axial pyridine ligand was located at δ 9.04. In both of the newly detected complexes, 3d and 3e, it is the chemical shift of the hydride ligands that secures their identity.

These unusual deductions are supported by Figure 9 where part a shows the one-scan 1H NMR spectrum that was recorded after 30 s of bubbling of p-H2 through a methanol-d4 solution of 3c at a PTF of 130 G. In the hydride region of this spectrum, PHIP-enhanced signals are seen for 3a, 3d, 3e, and 6. In contrast, the organic region shows polarized and diagnostic proton signals for bound PPh3, the −CH3 groups of the IMes ligands, the pyridine ligands, and the NCMe ligands in these species. When the corresponding 1H OPSY NMR spectrum (Figure 9b) was collected, the corresponding p-H2-enhanced signals now appear very clearly, while those of 3a are suppressed, as expected, because of its high symmetry. Figure 9c shows the corresponding normal 32-scan 1H NMR spectrum that was recorded under true Boltzmann conditions for comparison. These observations are therefore fully consistent with the expected redistribution of ligands in solution that allow the formation of an array of low-concentration species that are not readily detectable by normal NMR methods.

Figure 9

(a) 1H NMR signals from hyperpolarized pyridine, PPh3, the −CH3 groups of IMes ligands, and the NCMe ligands that are detected as a consequence of SABRE. The enhanced signals in the hydride region are detected through PHIP. This NMR spectrum is the result of bubbling p-H2 through a methanol-d4 solution of 3c at 130 G for 30 s prior to transfer into the NMR probe for observation. (b) Corresponding one-scan 1H OPSY NMR spectrum of this sample. (c) Corresponding 32-scan 1H NMR spectrum, recorded when the sample magnetization yields signals of an intensity that reflects the normal Boltzmann distribution rather than the HP conditions. The sample consisted of 3 mL of methanol-d4 solution of 3c (5.5 mM) and a 20-fold excess of pyridine.

Conclusions

This study has revealed that when 3a, 3b, and 4a are exposed to a solution containing H2, pyridine, and acetonitrile, the dominant products are 3c and 4c. The ligand-exchange pathways of 3c and 4c were explored using EXSY methods. These studies revealed that 3c undergoes preferential acetonitrile loss, while for 4c, pyridine is lost preferentially. These processes are dissociative in nature and show the expected saturation behavior at high ligand concentrations. We further note that when a solution of 4a is left under N2, the CH bond activation product 5 is obtained.

Utilization of 3c and 4c as SABRE catalysts has also revealed that substantial enhancements in both pyridine and acetonitrile can be achieved. In this case, it is the chemical and magnetic inequivalence of the hydride ligands that provides the route to SABRE activity. This is reflected in these systems by both the axial and equatorial ligands seeing different hydride couplings and hence receiving SABRE transfer. The efficiency of the HP of free pyridine by 4a is comparable with that of Ir(COD){CO(CH3)2}(IMes)]BF4. We note that in the case of 4a the spin polarization derived from p-H2 is now shared with just 7 in-plane protons, while in 2, it is 12. Consequently, we confirm that one route to maximize the SABRE effect is to introduce fewer protons into the complex at the point of magnetization sharing. This effect overcomes its slower ligand-exchange rate compared to 2. We also note that while 3 is better at polarizing acetonitrile, the reverse is true for 4, with pyridine being optimally polarized. This observation is consistent with the change in the ligand-exchange mechanism, where 3 dissociates acetonitrile and 4 pyridine.

We have also shown that it is possible to detect hyperpolarized signals that are diagnostic of the ligands surrounding the metal centers through SABRE. In this case, [Ir(H)2(MeOH)(py)(IMes)(PPh3)]BF4 (3d) and [Ir(H)2(NCMe)(py)2(PPh3)]BF4 (3e) were detected as a direct result of this process. These observations therefore illustrate a new route to using p-H2-derived magnetization to characterize reaction intermediates. Two concepts have been explored here to aid this process. In the first, the effect of the magnetic field experienced by the sample at the point of polarization transfer has been examined and is related to the efficiency of ligand HP. These data revealed that it is possible to differentiate signals in the same NMR region by collecting a series of 1H NMR spectra at different PTF values. We have also completed an extensive array of deuterium-labeling studies. These have confirmed that it is possible by reducing the extent of magnetization sharing to dramatically increase the level of transfer into the remaining 1H sites in the catalyst system. This has been used not only to increase the level of pyridine and acetonitrile polarization but to aid in the characterization of the reaction intermediates 3d and 3e. We further note that when 3b is formed as the isotopomer [Ir(H)2(h5-pyridine)(d5-pyridine)(IMes)(PPh3)]BF4, the corresponding hydride ligand signals now become visible through PHIP rather than the SABRE effect. This is a direct consequence of the small chemical shift difference that exists between them and further illustrates how magnetic symmetry underpins both SABRE and PHIP observations.

Consequently, these measurements established that SABRE can be used to probe the NMR signature of the ligand sphere surrounding such hyperpolarized complexes. By utilizing the PTF to change the intensity of nearly overlapping resonances, it is possible to clearly distinguish them. We hope these examples help to illustrate a novel route to monitor reactivity. The results presented here also reveal how important it is to consider the role of all species present in solution when quantifying and explaining the appearance of SABRE-derived spectra.