1Limits of ¹H and ¹³C NMR

Alkylation of an NH-indazole often gives both the N1 and the N2 isomer. With potassium carbonate in DMF, the two isomers form in almost equal amounts [8].

The same problem arises with pyrazoles, triazoles and tetrazoles [4, 13, 15]. The isomers have the same mass, and their ¹H and ¹³C spectra are similar. Two indirect methods are in common use:

• NOE. The substituent shows an NOE to H7 in the N1 isomer and to H3 in the N2 isomer [4].
• The ¹³C shift of C3. On average, C3 resonates near 135 ppm in N1-substituted indazoles and near 123 ppm in N2-substituted indazoles [3].

The NOE method requires a proton next to the substituent. It fails when C3 and C7 are both substituted. The C3 criterion is based on series averages, and a single compound may fall between them.

Protonation, tautomerism and metal coordination raise the same difficulty. In each case the electronic environment of one nitrogen changes, and ¹⁵N NMR is applied to all of these questions [2].

¹⁵N NMR measures the change at the nitrogen itself, not its indirect effect on neighbouring protons and carbons.

2Nitrogen types on the ¹⁵N scale

The ¹⁵N chemical-shift range spans about 1200 ppm [2]. Common nitrogen types appear in distinct regions of it (Figure 1).

An azole ring contains two kinds of nitrogen. A pyrrole-type nitrogen is bonded to hydrogen or to a substituent, and its lone pair is part of the aromatic system. A pyridine-type nitrogen has its lone pair in the ring plane. This lone pair is available for bonding to a proton or an alkyl group.

The two kinds resonate in different regions of the ¹⁵N scale. In N1-substituted indazoles, the pyridine-type N2 resonates between −45 and −66 ppm. The substituted, pyrrole-type N1 resonates between −173 and −192 ppm [3, 4].

Bond formation at the lone pair of a pyridine-type nitrogen moves its signal upfield by about 100 ppm or more. In 26 N-alkylated diazines, the mean shift of the alkylated nitrogen was −112 ppm [5]. Protonation of imines, oximes and pyridines in trifluoroacetic acid gives upfield shifts of 110–150 ppm [9].

Weaker interactions give smaller shifts. Hydrogen bonding in trifluoroethanol gives shifts of 8–28 ppm [9]. A change of solvent from cyclohexane to water moves the pyridine-type nitrogen of 1-methylazoles by up to about 30 ppm [18].

A shift of about 100 ppm indicates a new bond to the lone pair. A shift of 30 ppm or less indicates hydrogen bonding or a solvent effect.

3The experiments

Direct observation of ¹⁵N is difficult. The natural abundance of ¹⁵N is 0.37 %, and its receptivity is 3.84 × 10⁻⁶ relative to ¹H [1]. A direct ¹⁵N spectrum of a 300 g/mol compound at 30 mg/mL needs an overnight run, against 60 minutes for a ¹H–¹⁵N HMBC of the same sample [29].

Several experiments give ¹⁵N information at natural abundance. They differ in which nitrogens they detect:

The ¹H–¹⁵N HMBC is the routine choice. It detects the protons two or three bonds from each nitrogen, and the nitrogen shift is read from the cross-peaks [19]. Published examples indicate the scale of the experiment:

• 600 MHz, cryogenic probe: about 10–20 mg of sample, 20–30 minutes [5].
• 500 MHz: about 45 minutes per spectrum, with four scans per increment [7].
• 400 MHz: the indazole data in [3] were recorded at this field.

These figures apply to the instruments and samples reported. On another spectrometer, time and sample mass depend on the probe and the compound.

Long-range N–H coupling constants range from 2 to 16 Hz. An HMBC optimised for one value can miss correlations at another. Two spectra with different settings, for example 4 Hz and 10 Hz, recover the missing correlations [6].

The other experiments extend the method. A one-bond ¹H–¹⁵N correlation at natural abundance has been used, for example, to characterise hydrogen bonds between a catalyst and its substrate, in 29 minutes at 180 K [25]. The HCNMBC gives ¹⁵N shifts for nitrogens that do not appear in the HMBC [24].

The absence of a cross-peak is not evidence on its own. An assignment requires the shift pattern and several correlations.

4Assigning N1 and N2 isomers

A 2026 open-access study reports a clear pair of indazole isomers [4]. The substituent is an N-Boc-pyrrolidin-3-yl group. It is on N1 in compound 19b and on N2 in compound 20b. Both compounds have the same ester group at C4, and both spectra were recorded in CDCl₃.

Three sets of cross-peaks define the assignment:

• Pyrrolidine CH₂ protons. The protons on C2′ and C4′ are three bonds from the substituted ring nitrogen and four bonds from the other. Of the two ring nitrogens, they correlate only with the substituted one: N1 in 19b and N2 in 20b.
• H3. It is two bonds from N2 and three bonds from N1. It correlates with both ring nitrogens in 19b and with N2 only in 20b.
• Pyrrolidine H3′. It is two bonds from the substituted ring nitrogen and three bonds from the other. It correlates with N2 in 19b and with both ring nitrogens in 20b.

The shifts confirm the assignment. The two ring nitrogens are 126.2 ppm apart in 19b and 49.1 ppm apart in 20b. The same pattern holds for all ten indazoles measured in the two open-access studies [3, 4].

Ring-nitrogen shifts of N1- and N2-substituted indazoles (ppm, nitromethane scale)

The N-substituent is a hydroxymethyl group in 2a–2d and an N-Boc-pyrrolidin-3-yl group in 19a–20c. The separation is the difference between the two ring-nitrogen signals of one molecule.

Solvent and ring substituents change individual values. Across the table, N1 of the N1 isomers varies by 19 ppm. Between the matched isomers of [4], each ring nitrogen moves by 82–107 ppm. The solvent and substituent effects are therefore too small to affect the assignment.

In all ten published examples, one ¹H–¹⁵N HMBC spectrum distinguishes the N1 and N2 isomers without a reference compound. The CH₂ cross-peaks identify the substituted nitrogen. The separation of the two ring-nitrogen signals confirms the assignment.

5Limits of the method

Two N1-substituted pyrazoles from the same study differ only in the position of the ester group: C3 in 14b and C5 in 15b. Their ¹⁵N shifts are almost identical. N1 and N2 are at −159.9 and −70.1 ppm in 14b, and at −162.1 and −66.3 ppm in 15b [4].

In this case, the NOE gives the assignment. H3′ of the pyrrolidine shows a strong NOE to the pyrazole H5 in 14b. Compound 15b has no H5 and shows no NOE [4].

The method also applies to other ring systems:

• 1,2,3-Triazoles. ¹H–¹⁵N HMBC at natural abundance distinguishes the 1,4-, 1,5- and 2,4-disubstituted isomers. Calculated shifts support the assignment [13].
• 1,2,4-Triazolopyrimidines. The [1,5-a] and [4,3-a] isomers are distinguished. The ¹⁵N data also show the conversion of one isomer into the other [14].
• Tetrazoles. A natural-abundance study revised an earlier assignment of the 2,5-disubstituted isomers. It also identified N4 as the site of protonation [15].

Calculated shifts provide an independent check. In one systematic study, DFT-calculated ¹⁵N shifts agreed with experiment within ±9.6 ppm for 95 % of the nitrogens tested, with primary amines excepted [16]. This error is small compared with the 82–107 ppm shift of each nitrogen between matched indazole isomers [4].

¹⁵N shifts identify the substituted nitrogen. They do not locate a substituent on the carbon framework, which requires cross-peaks and NOE data.

6Tautomers: which nitrogen bears the hydrogen

In an NH-azole, the hydrogen can move between the ring nitrogens. The nitrogen that bears the hydrogen is pyrrole-type, and the other nitrogen is pyridine-type. ¹⁵N NMR therefore shows which tautomer is present.

In solution the exchange is often fast, and the two nitrogens give one averaged signal. A 3,5-diarylpyrazole gives one signal at −143.5 ppm for both nitrogens in CDCl₃ [20].

In the solid, the hydrogen is fixed on one nitrogen. The same pyrazole then gives two signals: −183.6 ppm for the N–H nitrogen and −98.0 ppm for the other nitrogen [20]. Their mean, −140.8 ppm, is close to the averaged value in solution.

Low temperature slows the exchange in solution. At 170–175 K in THF-d₈, the tautomerism of NH-pyrazoles is frozen, and the shifts of the individual tautomers can be measured [26].

One-bond ¹H–¹⁵N correlations and coupling constants show which nitrogen is bonded to the hydrogen. With these data and the ¹H and ¹³C shifts, 5-substituted pyrazolo[3,4-c]pyridines were shown to exist mainly as the N1–H tautomer [27].

The nitrogen that bears the hydrogen resonates in the pyrrole-type region. An averaged signal indicates fast exchange. A solid-state or low-temperature spectrum shows the individual tautomer.

7Locating the proton in a salt

Protonation is analogous to alkylation, and the ¹⁵N shift responds in the same way. A pyridine-type nitrogen moves upfield by about 100 ppm on protonation [10].

An amine nitrogen behaves differently. Protonation moves its signal by only about 10 ppm, and downfield [10]. In a base with an amine and an aromatic nitrogen, ¹⁵N NMR therefore shows clearly whether the aromatic nitrogen is protonated.

In a base with two amine nitrogens, the shift changes are small. A pH titration is then the better method. It follows the ¹H, ¹³C and ¹⁵N shifts of each nitrogen and gives a separate pKa for each amine [7].

The size of a protonation shift depends on the solvent and the counter-ion. Pyridine hydrochlorides show large solvent effects, whereas N-methylation shifts are almost independent of solvent and counter-ion [11]. In salts of one base with three different acids, the shift of the protonated nitrogen correlated almost linearly with the pKa of the acid [12].

Two practical rules follow:

• Compare the salt with its free base. Measure both in the same solvent and at similar concentration.
• Interpret an intermediate shift as partial protonation. Proton exchange in solution is faster than the NMR time scale. Each nitrogen therefore shows a population-weighted average shift [7].

The distinction between a salt and a cocrystal is a separate question. It concerns the position of the proton in the solid. Solid-state NMR answers it by measuring the N–H distance at natural abundance [17].

In solution, a protonated aromatic nitrogen shifts by about 100 ppm relative to the free base in the same solvent. A base with two amine nitrogens requires a pH titration. A solid requires solid-state NMR.

8The site of metal coordination

A metal bound to a pyridine-type nitrogen acts on the lone pair in the same way as a proton or an alkyl group. The bound nitrogen moves upfield, and the other nitrogens move little [20].

The pyrazole of Figure 5 illustrates this. In its Au(I), Pd(II) and Ag(I) complexes, the metal-bound N2 moved by −51 to −59 ppm relative to the free pyrazole in the solid. N1 moved by less than 5 ppm [20].

The size of the coordination shift depends on the metal and on the other ligands. A review of 105 complexes of Pd(II), Pt(II), Au(III), Co(III), Rh(III), Ir(III), Pd(IV) and Pt(IV) with pyridine-type ligands gives coordination shifts of about −30 to −150 ppm [21]. Their magnitude increases in the order Au(III) < Pd(II) < Pt(II), and Rh(III) < Co(III) < Pt(IV) < Ir(III) [21].

When a ligand has several possible donor nitrogens, the shifts identify the one that binds:

• Platinum(II) complexes of 9-methyladenine. The coordinated ring nitrogen moved 60 ppm upfield. It also showed a coupling of about 60 Hz to the phosphorus of the trans phosphine ligand [22].
• Ruthenium(II) complexes of a pyridyl-pyrimidine ligand. ¹H–¹⁵N HMBC spectra at natural abundance established which of the two pyrimidine nitrogens binds the metal [23].

A metal moves the bound nitrogen upfield by tens of ppm or more, while the unbound nitrogens move little. The shift identifies the donor atom.

9Reporting a ¹⁵N shift

Two reference scales are in use. IUPAC recommends nitromethane at 0 ppm, with liquid ammonia as the alternative for aqueous and biochemical work [1].

There is no single conversion constant between the two scales. The difference is 380.2 ppm for older spectrometers with the field perpendicular to the sample, and 381.7 ppm for superconducting magnets [2]. The IUPAC reference frequencies give 380.4 ppm [1].

The spread of 1.5 ppm arises from bulk magnetic susceptibility, which depends on the orientation of the sample in the field [2]. It does not affect an N1/N2 assignment, where the differences are about 100 ppm. It matters when a measured value is compared with a literature value to within a few ppm.

This article uses the nitromethane scale throughout. Negative values are upfield (more shielded).

A reported ¹⁵N shift should state the scale, the reference and the solvent.