Spectral editing of 13C NMR spectra via two-dimensional pulse techniques

JOURNAL

OFMAGNETIC

RESONANCE

81,50&5

11 (1989)

Spectral Editing of 13CNMR Spectra via Two-Dimensional Pulse Techniques NIEL.Y CHR. NIELSEN, * JENS CHR . MADSEN, * ,t HENRIK BILDSBE, * HANS J. JAKOBSEN, * AND OLE W. SBRENSEN$ *Department of Chemistry, University ofAarhus, DK-8000 Aarhus C, Denmark, and $ Laboratoriumfir Physikalische Chemie, Eidgeniissische Technische Hochschule, CH-8092 Zurich, Switzerland Received February 9,1988; revised June 22,1988 Two-dimensional pulse techniques for subspectral editing in 13C NMR spectroscopy are described. The experiments are compared with existing one-dimensional editing methods with respect to sensitivity, information content, and practical performance. In combination with a computer program for fully automatic extraction of one-dimensional edited subspectra and radiofrequency field strength information, the two-dimensional presented experiments are useful as setup experiments in 13C NMR. o 1989 Academic Press, Inc.

NMR pulse techniques for editing 13C or “N spectra into independent subspectra according to the number of attached protons have become popular during the past few years. Spectral editing techniques include the one-dimensional pulse sequences INEPT ( I, 2), DEPT (GL) (3, 4)) SEMUT (GL) (4,5), and APT or its improved version ESCORT (6). Furthermore, extended editing schemes providing information about the numbers of protons attached to the neighboring heteronuclei are also feasible. Such information can be achieved via 13C- 13C couplings using SEMINA ( 79) and SLAP- 1 /SLAP-2 editing sequences ( 9) or via ‘H- ‘H couplings using extended INEPT or DEPT experiments ( 2 0). In this paper we describe two-dimensional approaches to 13C subspectral editing using 2D analogs ( II, 12) of some of the editing experiments listed above. The combination with a computer program for fully automatic extraction of 1D edited subspectra makes these methods attractive as starting experiments especially in 13CNMR investigations. Obtained practically without any preknowledge, the output of the program is the decoupler RF field strength and proton multiplicities. For state-of-the-art high-Q probes this may often be of interest for samples varying widely in concentration or solvent. PRINCIPLES

OF 2D SPECTRAL

EDITING

In general, spectral editing distinguishes various 1,s (I = ‘H, S = r3C, “N, etc.) coupling networks through their different rotation angle dependencies of evolution 7 Present address: Research School of Chemistry, The Australian National University, Canberra A.C.T. 260 1, Australia. QO22-2364189

$3.00

Cop-t Q 1989 by Academic Ress, Inc. AU rights of reproduction in any form reservd.

500

2D EDITING

OF CARBON-1 3 SPECTRA

501

under a Hamiltonian %‘. This Hamiltonian can be that of heteronuclear scalar couplings xJ = ?rJ Ci 2 Ii,Sz (2,6) or that of the decoupler channel radiofrequency pulse xRF = -ylBI Ci Ii6 (3-5, 7-IO), where 4 refers to the phase ofthe pulse. All editing techniques contain a pulse sequence fragment which in its general form may be written F---e;, 180:--r.

111

For %’ editing, fl is fixed at 180” and T is varied. The opposite holds for #= editing, where 7 is fixed at (24-l and (9is varied. The resulting intensity functions fall into two categories, Category A: c sin 6 cos”-‘8, Category B: cosn&

c constant

Pal

Pbl

where fi is identified with either the J evolution angle 2%J7 or the Ilip angle 0 = y$i7P, depending on the type of editing. Category A editing is usually associated with coherence transfer from I to S spins whereas category B editing is employed primarily when the observed signals are based on native S spin magnetization. Category A examples are INEPT ( %J editing) and DEPT (.%& editing); category B examples are AFT or ESCORT ( xJ editing) and SEMUT (2~ editing). In 1D editing, subspectra are obtained from appropriate linear combinations of spectra recorded using different rotation angles P. Alternatively, the separation may be achieved through a Fourier transformation of the signal intensity dependencies on 27 or TV. This leads directly to 2D spectral editing where the pulse sequences can be derived from the corresponding 1D pulse schemes using the J evolution period ( 27) or the pulse length ( rP) as the t, evolution period. Consequently, the multiplicity information is found in the F, dimension of the resulting 2D spectra. In addition to proton multiplicities, the 2D spectra provide information about the interaction parameter responsible for multiplet splittings in the E; dimension, i.e., heteronuclear JcoupIings in %J editing and decoupler RF field strength in ZW editing. Such experiments have been introduced as heteronuclear J spectroscopy ( 13) and 2D calibration of RF field strengths ( I I, 12). Fourier analysis of the t i dependencies for the two categories of experiments in Eq. [2] leads to the F, dimension muItiplet patterns depicted in Fig. 1. If t, is equal to 27 (xJ editing) or 7P (xRr editing) the peak separations in F, are J or 2 ( yi Bi / 27r), respectively. Although the multiplicity information can be obtained by inspection of the 2D spectra, it is normally desirable to perform full editing into independent s&spectra each containing only one multiplicity. This requires linear combinations of Fz SW tions taken at F, = kpJ/2 or F, = kp(rIBd2u) with p = 0, 1, 2, 3 for x. and %‘,, editing, respectively. Unfortunately, variations in J couplings would impede %‘J editing to a degree that makes this approach impracticable. In contrast no such complications occur in %‘W editing because decoupler off-resonance effects can be safely neglected on modern spectrometers ( 12). In the following we therefore concentrate on %& editing. From Fig. 1 it is apparent that the n = 2 and n = 3 subspectra are delivered directly by the F2 sections for F, = 2 ( yIBl/27r) and F1 = 3 (r~Bd2tr). The n = 0 and n = 1 subspectra are obtained by linear combination of two F2 sections:

502

NIELSEN

ET AL.

FIG. 1. Stick plots of F, dimension I,$, n = 0, 1, 2, and 3, multiplet structures for 2D category A and category B editing experiments. The peak separations are 279B,/2r and Jfor SW and zJediting, respectively.

n = 0 by combining F, = 0 with F, = 2 (y,&/2a) (category B only), and n = 1 by combining F, = (y,&/27r)with F, = 3(yIB1/2a). In practice 2D editing methods have their main merit when the multiplicity information and the decoupler RF field strength are desired. If y1B1/2x is already known with an acceptable accuracy, a sensitivity gain can be obtained by editing via 1D methods (vide infra). A reasonable compromise between sensitivity and calibration accuracy in 2D 13C- ‘H experiments is N = 16-32 tl increments combined with appropriate selection of the spectral width SW ( 12)) [31 where nmax is the highest occurring proton multiplicity (i.e., normally nmax = 3). Under these conditions yi BI/27r can be determined with an accuracy of about 2.3 and 1.1% for N = 16 and N = 32, respectively ( 12). APPLICATIONS

OF 2D SPECTRAL

EDITING

Three pulse schemes for 1D and 2D xm spectral editing are outlined in Fig. 2. These are the category A sequences (a) DEPT and (b) extended DEPT and the category B sequence (c) SEMUT. In 2D versions 0 = yr BItl and for ideal timings of the T and 7’ delays (i.e., 7 = 7’ = (2 ‘J&-l) the sequences produce t, dependencies as expressed in Eq. [ 21 and thereby Fi multiplet patterns as displayed in Fig. 1. In Fig. 3 is shown an experimental i3C 2D DEPT spectrum of cholesterol recorded with the purpose of simultaneous spectral editing and calibration of the decoupler RF field strength; i.e., N = 32 and SW = 60,000 Hz. The edited subspectra (lower half) are generated as linear combinations of Fz sections extracted from the 2D spectrum (upper half) as described above. The CH, spectrum (all carbons) is obtained

2D EDITING 90::

a

180°

503

SPECTRA

8,

90”

180° t

13C

OF CARBON-13

t “b@-

90: ‘H

b

180" =I

% x1-t

: 90”

13C

t’

5 180"

t

t “b

F1~.2.Pulsesequences(a)DEPT(3,4),(b)extendedDEPT(IO),and(c)SEhrlUT(#,S)forlDand 2D ( I I, 12) subspectral editing. In 2D versions, the duration of the 0 pulse is incremented to constitute the t, evolution period. The delays 7 and #should be chosen as 7 = [ 2(&,, + 0.146(5- J&))]-’ and 7’ = [2(5,, - 0.146(5,, - Jmi.))]-’ to reduce Jcros~ talk (4,5), ~&re Jh” < J< J,, represents the actual range of J couplings. The homonuclear spin-echo period in (b) is adjusted to rI = ( 2 J,, )-I.

by integrating signals from the three F2 sections. The decoupler 90” pulse width is determined to tW- = 33.0 + 0.3 ps. Additional information about the surrounding ‘H- ‘H coupling networks can be achieved using the extended DEPT editing scheme depicted in Fii. 2b. During the extended spin-echo period (2q) prior to polarization transfer the signal is an&We modulated by a factor (10) nkzi cos( 7rJii1,2q) due to homonuclear coupIings between the directly attached proton(s) Ii and the surrounding protons Ik. Under the assumption of approximately equal magnitudes for the ‘H- ‘H couplings, adjustment of q = (2J&r causes sign modulation of the signals according to the parity of the number of distant protons Ik; i.e., even and odd numbers produce positive and negative line intensities, respectively, compared to normal DEPT. Figure 4 shows fully edited 13C spectra (lower half) obtained from a 13C 2D extended DEPT spectrum (upper half) of a mixture of three aliphatic compounds: 1,3dibromobutane, diethyl ethylmalonate, and isobutyl acetate. For organic molecules in which one proton multiplicity prevails, such as aromatic and carbohydrate compounds, ordinary 1,s editing is not worthwhile, but interpretation of the spectra can be greatly facilitated by the use of the extended DEPT sequence providing information about the number of protons on the neighboring bonded carbon atoms, Provkled the only object of the experiment is assignment, the 1D extended DEPT-edited CH (0 = 90”) spectrum is sufficient. The applicability of the extended DEPT 0 = 90

504

NIELSEN

ET AL.

I1

-20. III1 -10. I

2 Ato.---ILlo-

I

I

111

II I

II1111 I

II I

IIIII

---_----_-_ I

'

I

IIII II II 20.

[II I

307 120

100

60

60

LO

20

F2(ppm)

%

Ill

120

100

60

60

LO

I

20

F2(PP~) FIG. 3. (Upper half) Contour plot of a 2D 13CDEPT spectrum ( 7 = 3.95 ms, T’ = 3.74 ms) of cholesterol (0.08 M in CDCIS) obtained on a Varian XL-300 spectrometer using N = 32 t, increments for an F, spectral width SW = 60,000 Hz. (Lower half) Edited subspectra generated as linear combinations of F2 sections through resonances F, = ~,&/2r, p = 1,2, and 3 as described in the text; the CH, spectrum was obtained as a sum of the three F2 sections. The de-coupler RF field strength was determined to be y,B,/2~ = 7570 + 50 Hz corresponding to tw( ‘H) = 33.0 + 0.3 ps.

experiment for assignment of CH resonances according to the parity of the protons on neighboring carbons is demonstrated experimentally in Fig. 5 with 13C spectra of (a) methyl benzoate, (b) Br@ucopyranose pentaacetate, (c, e) 9-bromophenanthrene, and (d, f) 9-acetylphenanthrene. The spectra (a), (b), (e), and (f) are recorded using the extended DEPT sequence in Fig. 2b, whereas the standard 13C spectra (c) and (d) are inserted for comparison. The 7i delay is normally matched to three-bond ‘H- ‘H couplings, i.e., corresponding to magnitudes of 3JHH = 6- 12 Hz. Figure 5a illustrates that the equally intense ortho and meta i3C resonances in monosubstituted benzenes can be clearly distinguished using extended DEPT. In the spectrum of &Dglucopyranose pentaacetate (Fig. 5b) the C5 resonance is unambiguously assigned to the line at highest field for the two close-lying resonances between

2D EDITING

70

60

OF CARBON-13

50

40 F2 IPP~

30

20

10 0

0

0 a

505

SPECTRA

0

0

0

I

I

I

CHn I

I”‘,I.,I~I/‘)““,‘(“‘,I”“““‘I”“I””/’I”/”””’/’1”’ 70 60 50

LO F2 bpm)

30

20

10

FIG. 4. “C 2D extended DEPT spectra (r, = 76.9 ms, T = 3.89 ms, and 7’ = 3.42 ms) of a mixture of aliphatic compounds: 1,3dibromobutane, dietbyl ethylmalonate, and isobutyl acetate ( 1.O, 0.6, and 0.6 M, respectively, in acetone-d,) recorded using N = 32 and SW = 60,000 Hz. A 2D contour plot (upper half) is shown along with edited subspectra CH, CHa, and CH3 and CH, (lower half). Positive and negke line intensities are associated with carbons where the numbers of homonuclear coupling partners to the directly attached proton(s) are even and odd, respectively. Marked signals in the ID CH, spectrum refer to the assignments: 0 = I ,3dibromobutane, Cl = diethyl ethylmalonate, and A = isobutyl acetate. The average 90” decoupler pulse width was determined as t& ‘H) = 33.8 + 0.3 ps.

72 and 73 ppm; the somewhat reduced intensity for the negative C5 resonance is caused by a spread in the magnitudes for the three 3JHH couplings involving H5. The assignment of the resonance at 6 = 91.5 ppm as Cl (Fig. 5b) also follows direct& from this spectrum. In the 1Cline spectra for the two 9-X-substituted phemmthrerres (X = Br, Fig. 5c, and X = CH3C0, Fig. 5d), the extended DEPT experiments are useful for partial assignments of the crowded region between 122 and 134 ppm. The four negative lines in each of the Fig. 5e and 5f spectra obviously must be assigned to the resonances for C 1, C4, C5, and C8. Of these C4 and C5 may be safely assigned to the two peaks at about 122.6 ppm in accordance with the assignment for phenanthrene ( 14). The low intensities observed for some of the negative lines in Figs. 5e

506

NIELSEN

E -08 CD

=

ET

AL

=I -

2D EDITING

507

OF CARBON-1 3 SPECTRA

and 5f most likely result from strong coupling effects and not a spread in 3JHH values, because the intensities are quite insensitive to changes in the extended rI delay. SENSITIVITY

OF 2D SPECTRAL

EDITING

In evaluation of 2D methods for subspectral editing and RF field strength calibration it seems obligatory to consider the achievable sensitivity as compared to the sensitivity of the corresponding two 1D experiments required for providing the equivalent information. In Ref. (12) the proposed 2D sequences are proven to always provide significantly higher sensitivity of calibration than the earlier applied 1D analogs. In this section comparison with respect to sensitivity of subspectral editing is discussed. The relevant comparison is between edited subspectra obtained from a series of 1D experiments with a total of N X M FIDs and a 2D experiment with M FIDs coadded for each of the N t, sampling points, in both cases recorded under identical experimental conditions within the same total time. The relative sensitivity [(S/ N)&( S/N) ,n] n for an 1,s subspectrum can, within the assumption of exponential decay of the signal envelope in both time dimensions, be derived using the formulas given by Aue et al. ( 1.5) . First of all, the CH2 and CH3 subspectra are obtained directly from the F, = 2( yi&/2a) and F, = 3( y,Bi/2a) F2 sections. Without numerical weighting in the tl dimension we find

[

(S/N)ZD

(SIN)

ID

1

m+2 =

~~~,+2),(~+~~T:ff~+2)[1

- exp(-t?Y~~f~m+2~)1 “&n+2)h

,

m =o,

1.

PW

[41

The various components of Eq. [4] shall now be explained. c(,+~),(~+~) is a Fourier coefficient with the following general definitions for the two categories of 2D experiments: ,

parity(n)

= parity(p),

OGp=sn

[5a]

FIG. 5. 13C-extended DEPT-edited 13C-‘H spectra of some aromatic and carbohydrate compounds. (a) Methyl benzoate (66%, v/v, in CDC13, 4 scans, r = 7’ = 3.13 ms, ri = 62.5 ms); (b) @-~-@ucopyranose pentaacetate (0.4 Min CDC13, 256 scans, 7 = r’ = 3.45 ms, ri = 83.3 ms); (e) 9-bromopbenanthmne (0.4 MinCDC13,64scans,s=#=3.13 m s ,q = 62.5 ms); and (f) 9-acetylphenantbrene (0.5 Min CDC13, 16scans,~=s’=3.13ms,~I=55.5ms).Forcomparison(c)and(d)showstandard’3CSpectfaobtained for the compounds in (e) and ( f) with the same number of scans, respectively.

508

NIELSEN

ET

AL.

fn( 7, ,@ describes the sensitivity for an 1,s subspectrum generated as a linear combination of 1D spectra recorded using different rotation angles 0. For standard 1D category A and B experiments, rotation angles of 19~= p, 90”, and 180” - ,L3and I!& = O”, & 180” - p, and 180” with weights (relative number of scans) 77:2:9 and 1:v:Q: 1, respectively, are recommended. The optimum values of 9 and @can be found in Ref. (16). Finally, T::j is the t, -domain T2 relaxation time of the assumed exponential decay for an F, = miB,/2~ resonance. In the present experiments Tit; is generally dominated by RF inhomogeneity to a degree that allows neglect of homogeneous contributions; i.e., 1 / T$fd N p/ T$‘, where T$f,‘+ is the inhomogeneous broadening contribution for p = 1. The C (category B only) and CH subspectra require linear combination of F2 sections through F, = m(yiBi/27r) and F, = (m + 2)(y1B1/2?r) with m = 0 and 1, respectively. The scaling factor to be applied to the (m + 2) section is given by R,

= _

C(m+ZLm

Ti:i[

(1)

1 - exp( -t;““/

CM+~W+~) T2,(m+2)[1

T$‘A)]

- exp(-G”a/T$m+2~)l.

That leads to sensitivity ratios for C and CH subspectra according to

[

(S/N)ZD (SIWID

1 m =

‘I2 TifA[ 1 - exp( -tf”“/

cm,, fm(v,

8)

max

t1

Tit;)] 2

m=o,

1. [71

It should be noted that influences from RF inhomogeneity on signals obtained in the 1D experiments have been neglected in Eqs. [ 41 and [ 71. This approximation is acceptable only for a large number oft, increments (e.g., N > 10-12) as used in the experiments of practical relevance. Further, it should be kept in mind that the above analysis is based on comparison of peak heights in subspectra obtained from analogous 1D and 2D experiments. Thus, in terms of the 2D experiment, [(S/ N)2D ] n corresponds to the sensitivity of a single (or a linear combination of) F2 section(s) through the intensity maxima F, = nyi&/27r (and Fl = (II + 2)rrB1/27r). Depending on tt”“, the extent and profile of decoupler RF inhomogeneity, and weighting functions applied in the F, direction, [ ( S/N) 2D] n may be improved by instead integrating signals from a region of F, frequencies centered around the relevant intensity maxima. However, practically no gain in sensitivity is achieved with the assumptions of Eqs.[4]-[7]andtr”/T$+