Teeter-Tautomers

2,4-pentanedione (aka acetylacetone, acetylacetonato or acacH) is not only a ubiquitous ligand (AND ligand precursor!) for beautifully coloured organometallic complexes[1-3], but it is also a very interesting molecule with which to broach the subject of the rate of exchange processes and NMR!

In solution, acacH exists in two tautomeric forms: keto and enol.  A tautomer is a structural/constitutional isomer that is effectively interconverted through an acid catalyzed mechanism.

Why is this remarkable?  Well, of course it has ramifications in things like reactivity and mechanistic study, but for the sake of this discussion I’ll focus on dynamic NMR.  When I say ‘dynamic NMR’ I am referring to molecules that are involved in some sort of equilibrium – whether we’re talking about rotomers (which, are often confused with, but should be distinguished from non-interconvertable diastereomers[4]), tautomers or conjugate acid-base pairs.  Regardless of the exchange mechanism, we’re talking about a process in which two distinct forms (A and B) are interconverted.  If you think of these two forms having two unique resonant frequencies (δa and δb), the exchange process between A and B can be:
(a) fast;
(b) slow

It is not uncommon to refer to these rates relative to the NMR time scale.  But unfortunately, the term ‘NMR time scale’ is ambiguous and largely misused.[5]

In order to understand the differences we observe in spectra with different exchange rates, it helps to review our old friend Heisenberg’s Uncertainty Principle.  If there are two distinct species (again, this is the ‘A’ and ‘B’ we just defined) they have resonance frequencies measured in Hz of νa and vb:

Where h is Planck’s constant and τ is the lifetime of a given state

The probability of observing each frequency (va and vb) depends on how long (τ) each state (A and B) is ‘around’ in solution.

In case (a), where exchange is fast at room temperature at 60 MHz (such as the pKa determination of pyridinium/pyridine) we observe only 1 sharp resonance (δc).

Why? Well, in this regime, the lifetime is short (τ), and the uncertainty of finding our molecule in one state (A or B) is large relative to the change in chemical shift (Δv) and Δv collapses into one sharp line.  So for a rate to be considered (a) fast – the distance in chemical shifts must be much smaller than the exchange rate (Δv << k). [5,6]

It is important to remember here that we typically work with chemical shifts (δ) in ppm which are normalized for field strength, but they are measured as chemical shifts (v) in Hz (where Hz = 1/s).  Δv, of course, is dependent on the strength of magnetic field (Bo) the 1H NMR data was acquired at simply because Δv at 400 MHz >> Δv at 60 MHz (i.e., high field has inherently better chemical shift distribution).

For slow exchange (case (b)), the change in chemical shift is large relative to the exchange rate (Δv > k).  Here, τ is long – very, very long – and accordingly the uncertainty associated with each species decreases.  So now, instead of seeing a time-averaged resonance (δc), so we observe both A and B simultaneously at their own unique resonance frequencies (δa and δb or va and vb).

Now it is important to note that a fast and slow exchange can be interconverted!  We know that rate constants (k) are related to equilibrium constants (K), which, of course, are also related to temperature!

For a fixed Bo, a fast exchange can be changed be turned into a slow exchange by lowering the temperature (and vice versa).  Again, changing the temperature changes the equilibrium constant (K) and, correspondingly, the uncertainty of knowing the resonance frequency of species A or B.  The temperature at which two lines converge into one (and slow exchange into an apparent fast one) is known as the coalescence point.  The coalescence point, then is a time-averaged signal (δc = x δa + y δb) that depends on the molar fraction of A and B.

For acetylacetone, in the slow exchange regime, we see both keto (2,4-pentanedione) and the enol (4-hydroxy-pent-3-en-2-one) simultaneously.  At room temperature in CDCl3, these can be distinguished at 60 MHz:

Because the equilibrium is so slow, we can integrate each peak directly to determine the ratio of tautomers in solution.  To reiterate, we expect this ratio to be dependent on (a) solvent; and (b) temperature.

Solvent effect are illustrated in the 1H stacked plot below (generated in MNova) at ~30oC in d8-toluene, CDCl3 and d6-DMSO.  The enol form is strongly favoured in non-polar solvents like d8-toluene (87.6% enol) and chloroform-d (82.5% enol), but in polar hydrogen bonding solvents like d6-DMSO the keto form becomes a larger component of the reaction mixture (61.4 % enol).

The data was taken at 30oC because this is the temperature the NMReady-60 runs at.  Like any equilibrium constant, this ratio is in a fine balance, and can be manipulated by temperature.  But before I get into that:

Small digression on stability – The homogeneity of a Bo generated by a permanent magnet is vulnerable to changes in atmospheric temperature.  As homogeneity is critical to obtain good NMR data, we have to rigorously control the temperature of the magnet.  There are a number of protocols in place to help isolate the magnet from fluctuations in its external environment – one of these is to heat the magnet slightly above the room temperature of a controlled laboratory environment (hence the ~30oC).

Because of this, the current model of the NMReady-60 is not currently compatible with a controlled variable temperature (VT) NMR experiment.  However, this does not mean that quality data cannot be generated for a sample that is not at the same temperature as the magnet.  To demonstrate this I have included example data taken over a 60oC range (~0 to 60 oC).

The qualitative temperature effect is shown below for a series of 1H NMR spectra taken in d6-DMSO.  At 5 oC the enol is present in ~68%, as the temperature increases the ratio starts to totter over to the keto side and the percentage of enol decreases – 30 oC (~61%), 65 oC (~58%).

These rough experiments on ratio can be done in any solvent and is summarized below for both CDCl3 and d6-DMSO.

Okay – so, yeah, that got long so to recap:

1) Equilibria have a rate associated with it that is temperature dependent.

2) When discussing the rate relative to the ‘NMR time scale’ to discuss you must refer specifically to:
(a) a specific magnetic field and nuclei that is being observed (because Δv is dependent)
(b) a specific temperature (the characteristics of an equilibrium depend on temperature)
(c) whether we observe 2 individual signals for (δa and δb) or just one (δc).

3) A keto-enol tautmerization is slow at 60 MHz in the 0-65 oC temperature range!

[1] JACS, 2003, 125, 5622
[2] JACS, 2003, 125, 11911
[3] Eur. J. Inorg. Chem., 1998, 1285
[4] J. Org. Chem., 2012, 77, 5198
[5] Facey, Glen University of Ottawa NMR Facility Blog [Viewed May 5/2015]
[6] Bryant, R. G. J. Chem. Educ. 1983, 60, 933