Analysis and elimination of artifacts in indirect covariance NMR spectra via unsymmetrical processing.
Magnetic resonance in chemistry MRC (2005)
- PubMed: 16144032
Available from
Kirill Blinov's profile on Mendeley.
or
Abstract
Indirect covariance NMR offers an alternative method of extracting spin-spin connectivity information via the conversion of an indirect-detection heteronuclear shift-correlation data matrix to a homonuclear data matrix. Using an IDR (inverted direct response)-HSQC-TOCSY spectrum as a starting point for the indirect covariance processing, a spectrum that can be described as a carbon-carbon COSY experiment is obtained. These data are analogous to the autocorrelated 13C-13C double quantum INADEQUATE experiment except that the indirect covariance NMR spectrum establishes carbon-carbon connectivities only between contiguous protonated carbons. Cyclopentafuranone and the complex polynuclear heteroaromatic naphtho2',1':5,6-naphtho2',1':4,5thieno2,3-cquinoline are used as model compounds. The former is a straightforward example because of its well-resolved proton spectrum, while the latter, which has considerable resonance overlap in its congested proton spectrum, gives rise to two types of artifact responses that must be considered when using the indirect covariance NMR method.
Author-supplied keywords
Available from
Kirill Blinov's profile on Mendeley.
Page 1
Analysis and elimination of artifacts in indirect covariance NMR spectra via unsymmetrical processing.
MAGNETIC RESONANCE IN CHEMISTRY
Magn. Reson. Chem. 2005; 43: 999–1007
Published online 16 September 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mrc.1674
Analysis and elimination of artifacts in indirect
covariance NMR spectra via unsymmetrical processing
Kirill A. Blinov,1 Nicolay I. Larin,1 Mikhail P. Kvasha,1 Arvin Moser,2 Antony J. Williams2 and
Gary E. Martin3∗
1 Advanced Chemistry Development, Moscow Department 6 Akademik Bakulev Street, Moscow 117512 Russian Federation, Russia
2 Advanced Chemistry Development 110 Yonge Street 14th Floor Toronto M5C 1T4, Ontario Canada
3 Pfizer Global Research and Development Analytical Research and Development 7000 Portage Road Kalamazoo, Michigan 49001-0199, USA
Received 8 May 2005; Revised 27 June 2005; Accepted 29 June 2005
Indirect covarianceNMRoffers an alternativemethod of extracting spin–spin connectivity information via
the conversion of an indirect-detection heteronuclear shift-correlation data matrix to a homonuclear data
matrix. Using an IDR (inverted direct response)-HSQC-TOCSY spectrum as a starting point for the indirect
covariance processing, a spectrum that can be described as a carbon–carbon COSY experiment is obtained.
These data are analogous to the autocorrelated 13C–13C double quantum INADEQUATE experiment
except that the indirect covariance NMR spectrum establishes carbon–carbon connectivities only
between contiguous protonated carbons. Cyclopentafuranone and the complex polynuclear heteroaromatic
naphtho[2′,1′:5,6]-naphtho[2′,1′:4,5]thieno[2,3-c]quinoline are used as model compounds. The former is a
straightforward example because of its well-resolved proton spectrum, while the latter, which has
considerable resonance overlap in its congested proton spectrum, gives rise to two types of artifact
responses that must be considered when using the indirect covariance NMR method. Copyright 2005
John Wiley & Sons, Ltd.
KEYWORDS: indirect covariance NMR; IDR-HSQC-TOCSY; carbon–carbon vicinal correlation
INTRODUCTION
There have been several recently published reports that
have outlined the principles behind covariance NMR
spectroscopy.1,2 More recently, Bru¨schweiler and coworkers
described covariance NMR spectroscopy by singular value
decomposition for homonuclear 2D NMR data3 in addition
to indirect covariance NMR spectroscopy for use on het-
eronuclear 2D NMR data.4 The latter method was of interest
in that it provides the means of obtaining homonuclear 2D
NMR spectra of the insensitive nucleus of a heteronuclide
pair with detection via the sensitive nuclide. Beginning with
an HSQC-TOCSY data set, the resulting indirect covari-
ance NMR spectrum is the equivalent of an autocorrelated
13C–13C double quantum INADEQUATE spectrum.5 – 8 How-
ever, unlike the autocorrelated INADEQUATE spectrum,
which affords correlations between both protonated and
quaternary carbon resonances, the indirect covariance NMR
spectrum is more accurately described as a protonated car-
bon CC-COSY spectrum and is only capable of furnishing
connectivity information between contiguous protonated
carbons.
To explore this new method, we elected to use
data sets for several model compounds, which included
ŁCorrespondence to: Gary E. Martin, Analytical R & D
0200/259/277, Pfizer Global Research & Development, 7000
Portage Road, Kalamazoo, MI 49001-0199, USA.
E-mail: gary.e.martin@pfizer.com
the cyclopentafuranone (1, (3aR,4S,5R,6aS)-5-hydroxy-4-
(hydroxymethyl)hexahydro-2H-cyclopenta[b]furan-2-one)
and the complex polynuclear aromatic heterocycle naphtho
[20,10:5,6]naphtho[20,10:4,5]thieno[2,3-c]quinoline (2). The for-
mer is a reasonably simple molecule and nearly completely
resolved with only one pair of overlapping resonances at
an observation frequency of 500 MHz. In contrast, 2 exhibits
a very congested proton NMR spectrum with a number of
resonances partially or heavily overlapped even at 600 MHz.
The differences between the complexity of the proton spec-
tra of 1 and 2 lead to considerable differences in the ease
of interpretation, assignment of the data, and presence of
artifacts.
Copyright 2005 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2005; 43: 999–1007
Published online 16 September 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mrc.1674
Analysis and elimination of artifacts in indirect
covariance NMR spectra via unsymmetrical processing
Kirill A. Blinov,1 Nicolay I. Larin,1 Mikhail P. Kvasha,1 Arvin Moser,2 Antony J. Williams2 and
Gary E. Martin3∗
1 Advanced Chemistry Development, Moscow Department 6 Akademik Bakulev Street, Moscow 117512 Russian Federation, Russia
2 Advanced Chemistry Development 110 Yonge Street 14th Floor Toronto M5C 1T4, Ontario Canada
3 Pfizer Global Research and Development Analytical Research and Development 7000 Portage Road Kalamazoo, Michigan 49001-0199, USA
Received 8 May 2005; Revised 27 June 2005; Accepted 29 June 2005
Indirect covarianceNMRoffers an alternativemethod of extracting spin–spin connectivity information via
the conversion of an indirect-detection heteronuclear shift-correlation data matrix to a homonuclear data
matrix. Using an IDR (inverted direct response)-HSQC-TOCSY spectrum as a starting point for the indirect
covariance processing, a spectrum that can be described as a carbon–carbon COSY experiment is obtained.
These data are analogous to the autocorrelated 13C–13C double quantum INADEQUATE experiment
except that the indirect covariance NMR spectrum establishes carbon–carbon connectivities only
between contiguous protonated carbons. Cyclopentafuranone and the complex polynuclear heteroaromatic
naphtho[2′,1′:5,6]-naphtho[2′,1′:4,5]thieno[2,3-c]quinoline are used as model compounds. The former is a
straightforward example because of its well-resolved proton spectrum, while the latter, which has
considerable resonance overlap in its congested proton spectrum, gives rise to two types of artifact
responses that must be considered when using the indirect covariance NMR method. Copyright 2005
John Wiley & Sons, Ltd.
KEYWORDS: indirect covariance NMR; IDR-HSQC-TOCSY; carbon–carbon vicinal correlation
INTRODUCTION
There have been several recently published reports that
have outlined the principles behind covariance NMR
spectroscopy.1,2 More recently, Bru¨schweiler and coworkers
described covariance NMR spectroscopy by singular value
decomposition for homonuclear 2D NMR data3 in addition
to indirect covariance NMR spectroscopy for use on het-
eronuclear 2D NMR data.4 The latter method was of interest
in that it provides the means of obtaining homonuclear 2D
NMR spectra of the insensitive nucleus of a heteronuclide
pair with detection via the sensitive nuclide. Beginning with
an HSQC-TOCSY data set, the resulting indirect covari-
ance NMR spectrum is the equivalent of an autocorrelated
13C–13C double quantum INADEQUATE spectrum.5 – 8 How-
ever, unlike the autocorrelated INADEQUATE spectrum,
which affords correlations between both protonated and
quaternary carbon resonances, the indirect covariance NMR
spectrum is more accurately described as a protonated car-
bon CC-COSY spectrum and is only capable of furnishing
connectivity information between contiguous protonated
carbons.
To explore this new method, we elected to use
data sets for several model compounds, which included
ŁCorrespondence to: Gary E. Martin, Analytical R & D
0200/259/277, Pfizer Global Research & Development, 7000
Portage Road, Kalamazoo, MI 49001-0199, USA.
E-mail: gary.e.martin@pfizer.com
the cyclopentafuranone (1, (3aR,4S,5R,6aS)-5-hydroxy-4-
(hydroxymethyl)hexahydro-2H-cyclopenta[b]furan-2-one)
and the complex polynuclear aromatic heterocycle naphtho
[20,10:5,6]naphtho[20,10:4,5]thieno[2,3-c]quinoline (2). The for-
mer is a reasonably simple molecule and nearly completely
resolved with only one pair of overlapping resonances at
an observation frequency of 500 MHz. In contrast, 2 exhibits
a very congested proton NMR spectrum with a number of
resonances partially or heavily overlapped even at 600 MHz.
The differences between the complexity of the proton spec-
tra of 1 and 2 lead to considerable differences in the ease
of interpretation, assignment of the data, and presence of
artifacts.
Copyright 2005 John Wiley & Sons, Ltd.
Page 2
1000 K. A. Blinov et al.
RESULTS AND DISCUSSION
Analysis of artifact responses in indirect
covariance NMR spectra
The annotated IDR (inverted direct response)-HSQC-TOCSY
spectrum of cyclopentafuranone (1) is shown in Fig. 1. Direct
correlation responses correspond exactly to the correlations
of an HSQC spectrum and are inverted by the pulse sequence
used.9,10 Negatively phased direct correlation responses are
represented by black contours in Fig. 1. Relayed correlation
responses have a positive phase and are defined by red
contours. The mixing time used during the acquisition of
these data was 18 ms. The relatively short mixing time
provided coherence transfer to only vicinal neighbor protons
almost exclusively. Variations in the relayed response
intensity observed in Fig. 1 are a function of the size of
the vicinal coupling constant between the correlated protons
through which coherence was transferred.
The indirect covariance NMR spectrum of 1 obtained by
processing the IDR-HSQC-TOCSY frequency domain spec-
trum according to the method of Zhang and Bru¨schweiler4
is shown in Fig. 2. The well-resolved proton spectrum of 1,
which has only a single pair of overlapping proton reso-
nances, yields an indirect covariance NMR spectrum that is
straightforward to interpret and assign. The anisochronous
C9 methylene resonance at 61.9 ppm provides a convenient
starting point for the assignment of both the HSQC-TOCSY
spectrum in Fig. 1 as well as the indirect covariance NMR
spectrum shown in Fig. 2. The negative off-diagonal response
5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5
F2 chemical shift (ppm)
35
40
45
50
55
60
65
70
75
80
85
90
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
9 > 410 < 9
5 < 4
3 < 4
3 > 8b
8a < 3
2 < 3
2 > 6
5 > 6a
5 > 6b
6a 6b
5 < 6a/b
5 > 4
11 < 5
2 < 6a/b
2 > 3
9 < 4
3 > 4
7
8
O
1
3
2
4
5
6
OH
11
O
12
9 OH
10
Figure 1. Inverted direct response HSQC-TOCSY spectrum of
cyclopentafuranone (1) acquired with an 18-ms mixing time.
Proton–carbon direct correlation responses have negative
phase and are denoted by black contours; relayed coherence
responses have a positive phase and are denoted by red
contours. The numbering scheme used is shown by the
structure inset. Responses are labeled using the convention of
listing the downfield resonance followed by the upfield
resonance.
pair labeled C4–C9 (the convention of listing the upfield res-
onance is first used for the correlation labels) identifies this
carbon–carbon vicinal pair. Correlations from C4 to C3 and
to C5 are much weaker than the C4–C9 correlation as a con-
sequence of the small vicinal proton couplings to H4 (see the
proton reference spectrum plotted above the HSQC-TOCSY
spectrum in Fig. 1). Correlations linking C5 to C6 and C3 to
C8 (labeled C6–C5 and C8–C3, respectively) are consider-
ably more intense. Finally, the correlations linking both C6
and C3 to C2 are only partially resolved and are again weak
because of small vicinal proton–proton couplings.
The indirect covariance NMR spectrum shown in Fig. 2
also contains a pair of prominent positive off-diagonal
responses labeled as the C4–C6 Type I artifact response.
We have found that when using IDR-HSQC-TOCSY spectra
such as that shown in Fig. 1 for indirect covariance NMR
processing, two types of artifacts are observable due to
proton-resonance overlap. Zhang and Bru¨schweiler4 referred
to artifacts because of proton resonance overlap in general
terms but did not elaborate further on this aspect of indirect
covariance NMR spectra. Figure 3 shows a side-by-side
presentation of the upfield region of the indirect covariance
spectrum shown in Fig. 2 with the appropriate segment of
the HSQC-TOCSY spectrum from Fig. 1. Note that in the
right panel of Fig. 3, there are two IDRs corresponding to
H4–C4 and H6b–C6. Relayed coherence responses are also
observed from H3 ! H4 at the chemical shift of C3 and
from H9 ! H4 at the chemical shift of C9. The pair of
correlations in the right panel for the H4/C4 direct response
C4−C6 Type I
Artifact response
90 85 80 75 70 65 60 55 50 45 40 35 30
F2 chemical shift (ppm)
35
40
45
50
55
60
65
70
75
80
85
90
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
C2 C5
C9 C4 C3C6 C8
7
8
O
1
3
2
4
5
6
OH
11
O
12
9 OH
10
C3–C2
C6–C2
C4–C5
C6–C5 C3–C4
C8–C3
C4–C9
Figure 2. Indirect covariance NMR spectrum of
cyclopentafuranone (1) calculated according to the method of
Zhang and Bru¨schweiler.4 Both frequency axes are the
indirectly detected nuclide, 13C in this case. Negatively phased
off-diagonal responses defined by black contours correspond
to carbon–carbon vicinal connectivities. The numbering
scheme is shown on the structure inset. The single pair of
positively phased, off-diagonal responses (red contours)
correspond to a Type I artifact.
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
RESULTS AND DISCUSSION
Analysis of artifact responses in indirect
covariance NMR spectra
The annotated IDR (inverted direct response)-HSQC-TOCSY
spectrum of cyclopentafuranone (1) is shown in Fig. 1. Direct
correlation responses correspond exactly to the correlations
of an HSQC spectrum and are inverted by the pulse sequence
used.9,10 Negatively phased direct correlation responses are
represented by black contours in Fig. 1. Relayed correlation
responses have a positive phase and are defined by red
contours. The mixing time used during the acquisition of
these data was 18 ms. The relatively short mixing time
provided coherence transfer to only vicinal neighbor protons
almost exclusively. Variations in the relayed response
intensity observed in Fig. 1 are a function of the size of
the vicinal coupling constant between the correlated protons
through which coherence was transferred.
The indirect covariance NMR spectrum of 1 obtained by
processing the IDR-HSQC-TOCSY frequency domain spec-
trum according to the method of Zhang and Bru¨schweiler4
is shown in Fig. 2. The well-resolved proton spectrum of 1,
which has only a single pair of overlapping proton reso-
nances, yields an indirect covariance NMR spectrum that is
straightforward to interpret and assign. The anisochronous
C9 methylene resonance at 61.9 ppm provides a convenient
starting point for the assignment of both the HSQC-TOCSY
spectrum in Fig. 1 as well as the indirect covariance NMR
spectrum shown in Fig. 2. The negative off-diagonal response
5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5
F2 chemical shift (ppm)
35
40
45
50
55
60
65
70
75
80
85
90
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
9 > 410 < 9
5 < 4
3 < 4
3 > 8b
8a < 3
2 < 3
2 > 6
5 > 6a
5 > 6b
6a 6b
5 < 6a/b
5 > 4
11 < 5
2 < 6a/b
2 > 3
9 < 4
3 > 4
7
8
O
1
3
2
4
5
6
OH
11
O
12
9 OH
10
Figure 1. Inverted direct response HSQC-TOCSY spectrum of
cyclopentafuranone (1) acquired with an 18-ms mixing time.
Proton–carbon direct correlation responses have negative
phase and are denoted by black contours; relayed coherence
responses have a positive phase and are denoted by red
contours. The numbering scheme used is shown by the
structure inset. Responses are labeled using the convention of
listing the downfield resonance followed by the upfield
resonance.
pair labeled C4–C9 (the convention of listing the upfield res-
onance is first used for the correlation labels) identifies this
carbon–carbon vicinal pair. Correlations from C4 to C3 and
to C5 are much weaker than the C4–C9 correlation as a con-
sequence of the small vicinal proton couplings to H4 (see the
proton reference spectrum plotted above the HSQC-TOCSY
spectrum in Fig. 1). Correlations linking C5 to C6 and C3 to
C8 (labeled C6–C5 and C8–C3, respectively) are consider-
ably more intense. Finally, the correlations linking both C6
and C3 to C2 are only partially resolved and are again weak
because of small vicinal proton–proton couplings.
The indirect covariance NMR spectrum shown in Fig. 2
also contains a pair of prominent positive off-diagonal
responses labeled as the C4–C6 Type I artifact response.
We have found that when using IDR-HSQC-TOCSY spectra
such as that shown in Fig. 1 for indirect covariance NMR
processing, two types of artifacts are observable due to
proton-resonance overlap. Zhang and Bru¨schweiler4 referred
to artifacts because of proton resonance overlap in general
terms but did not elaborate further on this aspect of indirect
covariance NMR spectra. Figure 3 shows a side-by-side
presentation of the upfield region of the indirect covariance
spectrum shown in Fig. 2 with the appropriate segment of
the HSQC-TOCSY spectrum from Fig. 1. Note that in the
right panel of Fig. 3, there are two IDRs corresponding to
H4–C4 and H6b–C6. Relayed coherence responses are also
observed from H3 ! H4 at the chemical shift of C3 and
from H9 ! H4 at the chemical shift of C9. The pair of
correlations in the right panel for the H4/C4 direct response
C4−C6 Type I
Artifact response
90 85 80 75 70 65 60 55 50 45 40 35 30
F2 chemical shift (ppm)
35
40
45
50
55
60
65
70
75
80
85
90
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
C2 C5
C9 C4 C3C6 C8
7
8
O
1
3
2
4
5
6
OH
11
O
12
9 OH
10
C3–C2
C6–C2
C4–C5
C6–C5 C3–C4
C8–C3
C4–C9
Figure 2. Indirect covariance NMR spectrum of
cyclopentafuranone (1) calculated according to the method of
Zhang and Bru¨schweiler.4 Both frequency axes are the
indirectly detected nuclide, 13C in this case. Negatively phased
off-diagonal responses defined by black contours correspond
to carbon–carbon vicinal connectivities. The numbering
scheme is shown on the structure inset. The single pair of
positively phased, off-diagonal responses (red contours)
correspond to a Type I artifact.
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
Page 3
Elimination of artifacts in NMR spectra 1001
C4−C6 Type I
Artifact Response
C4-C6
1.9 1.8 1.7 1.6 1.5
F2 chemical shift (ppm)
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
64 62 60 58 56 54 52 50 48 46 44 42 40 38 36 34
F2 chemical shift (ppm)
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
7
8
O
1
3
2
4
5
6
O
12
9 OH
10
OH
11
C3–C4
C8–C3
C4–C9
C4–C9
C3–C4
C6−C9 Type II
Artifact response
Figure 3. Composite presentation of a segment of the indirect covariance NMR spectrum (left panel) and the corresponding region
of the HSQC-TOCSY spectrum (right panel). Relative to the spectrum shown in Fig. 2, the threshold for the lowest contour is
somewhat lower. The positive off-diagonal correlations in the left panel labeled C4–C6 Type I artifact response arise owing to the
overlap of the H4 and H6 protons in the data from the HSQC-TOCSY spectrum as shown in both panels connected by a solid red
line. The very weak pair of correlations in the left panel connected by the dashed black line labeled C6–C9 Type II artifact again
arises as a consequence of the overlap of H4 and H6. The direct response from H6b at the chemical shift of C6 and the relayed
response from the C9 methylene resonances to H4 at the chemical shift of C9 give rise to the negative-phase off-diagonal Type II
artifact responses. A slightly deeper threshold level was used when plotting the data shown in this figure vs the data shown in Fig. 2,
allowing the Type II response to be observed.
and the relayed H3 ! H4/C3 response gives rise to the weak
C3–C4 vicinal carbon–carbon correlation response in the left
panel as expected in the indirect covariance NMR spectrum.
A second pair of responses for the H4/C4 direct response
and the relayed H9 ! H4 response at the C9 chemical
shift affords the C4–C9 vicinal carbon–carbon correlation
response, as expected.
The positive-phase off-diagonal responses in the indirect
covariance spectrum arise because of H4/H6b proton-
resonance overlap; the direct responses for H6b/C6 and
H4/C4 connected by the solid red line in the right panel
define this correlation, which we have elected to label
as a Type I artifact. The positive phase of this type of
artifact response makes it possible to visually eliminate
them from consideration when interpreting any indirect
covariance NMR spectrum. Type I artifact responses can
arise between a pair of IDRs as in this case or between
a pair of positive-phase relayed coherence responses (see
the discussion of the indirect covariance NMR spectrum of
naphtho[20,10:5,6]naphtho[20,10:4,5]thieno[2,3-c]quinoline (2)
later). In both cases, the responses in the indirect covariance
for these artifacts have a positive phase. It is also worth
noting at this point that when conventional HSQC-TOCSY
data, rather than IDR-HSQC-TOCSY data, are subjected to
indirect covariance processing, all responses in the spectrum
have the same phase (positive) rendering the Type I artifacts
indistinguishable from any other response in the spectrum,
thereby further complicating the interpretation of the data.
The weak, negative-phase off-diagonal response in the
indirect covariance spectrum shown in the left panel of
Fig. 3 and labeled C6–C9 Type II artifact response arises
because of the IDR for H6b/C6 and the positive-phase
relayed response for H9 ! H4 at the C9 chemical shift.
This pair of responses gives rise to a very weak negative-
phase off-diagonal response in the indirect covariance NMR
spectrum that looks like a legitimate vicinal carbon–carbon
correlation response. The response, however, is a type
of artifact response that we have labeled Type II. This
type of artifact response is indistinguishable from the
carbon–carbon vicinal correlation responses sought in
the experiment, and thus represents, in our opinion, a
shortcoming of the method. While the single Type II response
visible in the indirect covariance NMR spectra shown in
Figs 2 and 3 is very weak and thus of little concern, similar
Type II responses observed in the indirect covariance NMR
spectrum of naphtho[20,10:5,6]-naphtho[20,10:4,5]thieno[2,3-
c]quinoline (2) have intensity comparable to the legitimate
carbon–carbon vicinal connectivity correlations and are thus
more problematic.
It should also be noted that the complete complement
of artifact responses being analyzed here and described as
Type I and Type II artifact responses is also observed when
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
C4−C6 Type I
Artifact Response
C4-C6
1.9 1.8 1.7 1.6 1.5
F2 chemical shift (ppm)
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
64 62 60 58 56 54 52 50 48 46 44 42 40 38 36 34
F2 chemical shift (ppm)
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
7
8
O
1
3
2
4
5
6
O
12
9 OH
10
OH
11
C3–C4
C8–C3
C4–C9
C4–C9
C3–C4
C6−C9 Type II
Artifact response
Figure 3. Composite presentation of a segment of the indirect covariance NMR spectrum (left panel) and the corresponding region
of the HSQC-TOCSY spectrum (right panel). Relative to the spectrum shown in Fig. 2, the threshold for the lowest contour is
somewhat lower. The positive off-diagonal correlations in the left panel labeled C4–C6 Type I artifact response arise owing to the
overlap of the H4 and H6 protons in the data from the HSQC-TOCSY spectrum as shown in both panels connected by a solid red
line. The very weak pair of correlations in the left panel connected by the dashed black line labeled C6–C9 Type II artifact again
arises as a consequence of the overlap of H4 and H6. The direct response from H6b at the chemical shift of C6 and the relayed
response from the C9 methylene resonances to H4 at the chemical shift of C9 give rise to the negative-phase off-diagonal Type II
artifact responses. A slightly deeper threshold level was used when plotting the data shown in this figure vs the data shown in Fig. 2,
allowing the Type II response to be observed.
and the relayed H3 ! H4/C3 response gives rise to the weak
C3–C4 vicinal carbon–carbon correlation response in the left
panel as expected in the indirect covariance NMR spectrum.
A second pair of responses for the H4/C4 direct response
and the relayed H9 ! H4 response at the C9 chemical
shift affords the C4–C9 vicinal carbon–carbon correlation
response, as expected.
The positive-phase off-diagonal responses in the indirect
covariance spectrum arise because of H4/H6b proton-
resonance overlap; the direct responses for H6b/C6 and
H4/C4 connected by the solid red line in the right panel
define this correlation, which we have elected to label
as a Type I artifact. The positive phase of this type of
artifact response makes it possible to visually eliminate
them from consideration when interpreting any indirect
covariance NMR spectrum. Type I artifact responses can
arise between a pair of IDRs as in this case or between
a pair of positive-phase relayed coherence responses (see
the discussion of the indirect covariance NMR spectrum of
naphtho[20,10:5,6]naphtho[20,10:4,5]thieno[2,3-c]quinoline (2)
later). In both cases, the responses in the indirect covariance
for these artifacts have a positive phase. It is also worth
noting at this point that when conventional HSQC-TOCSY
data, rather than IDR-HSQC-TOCSY data, are subjected to
indirect covariance processing, all responses in the spectrum
have the same phase (positive) rendering the Type I artifacts
indistinguishable from any other response in the spectrum,
thereby further complicating the interpretation of the data.
The weak, negative-phase off-diagonal response in the
indirect covariance spectrum shown in the left panel of
Fig. 3 and labeled C6–C9 Type II artifact response arises
because of the IDR for H6b/C6 and the positive-phase
relayed response for H9 ! H4 at the C9 chemical shift.
This pair of responses gives rise to a very weak negative-
phase off-diagonal response in the indirect covariance NMR
spectrum that looks like a legitimate vicinal carbon–carbon
correlation response. The response, however, is a type
of artifact response that we have labeled Type II. This
type of artifact response is indistinguishable from the
carbon–carbon vicinal correlation responses sought in
the experiment, and thus represents, in our opinion, a
shortcoming of the method. While the single Type II response
visible in the indirect covariance NMR spectra shown in
Figs 2 and 3 is very weak and thus of little concern, similar
Type II responses observed in the indirect covariance NMR
spectrum of naphtho[20,10:5,6]-naphtho[20,10:4,5]thieno[2,3-
c]quinoline (2) have intensity comparable to the legitimate
carbon–carbon vicinal connectivity correlations and are thus
more problematic.
It should also be noted that the complete complement
of artifact responses being analyzed here and described as
Type I and Type II artifact responses is also observed when
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
Page 4
1002 K. A. Blinov et al.
a conventional HXQC-TOCSY (X D M or S) is used as the
source of the data for indirect covariance processing. In
the case of the conventional HXQC-TOCSY spectrum, all
of the artifact responses will have identical phases since all
of the responses in the spectrum are positive. Hence Type
I artifact responses will be indistinguishable from Type II
artifact responses or from legitimate carbon–carbon vicinal
correlation responses. In this regard, it is preferable to utilize
IDR-HSQC-TOCSY when indirect covariance processing is
contemplated. The use of HSQC rather than the HMQC-
based variant of the experiment is also preferable based
on the improved F1 resolution of the former, which is
comparable to the difference in F1 resolution between an
HMQC and HSQC experiment.11
Moving from cyclopentafuranone (1) to naphtho[20,10:
5,6]naphtho[20,10:4,5]-thieno[2,3-c]quinoline (2), which has
a much more complex and overlapped proton spectrum
even at 600 MHz, as expected, both the IDR-HSQC-
TOCSY and indirect covariance NMR spectra are con-
siderably more complex. The IDR-HSQC-TOCSY spec-
trum of 2 is shown in Fig. 4. The total assignment of 2,
based on the use of IDR-HSQC-TOCSY data, has been
reported.11,12 For purposes of a discussion of the indi-
rect covariance NMR spectrum of 2, we will focus on
two isolated two-spin systems, H5/C5–H6/C6 (9.09/134.4
and 8.52/125.4 ppm, respectively) and H14/C14–H15/C15
(7.84/132.5 and 8.53/129.8 ppm, respectively). As would be
expected from the overlap of H6 and H15 resonating at
8.52 and 8.53 ppm, respectively, 2 has considerable poten-
tial for the observation of artifact responses in the indirect
covariance NMR spectrum. The location of the two-spin
systems on which the discussion of the indirect covari-
ance NMR spectrum will be based is shown by 3. The
desired vicinal carbon–carbon correlations are denoted in
the structure by the solid black lines. From the analysis
of the indirect covariance NMR spectrum of 2 shown in
Fig. 5, the solid red and dashed black lines, correspond-
ing to Type I and Type II artifacts, respectively, are also
shown. It is important to note that the lines signifying the
resonances leading to Type II artifacts cannot be drawn
in arbitrarily; the spectrum must be analyzed to deter-
mine which resonances are associated with Type II artifact
responses.
S
15
14
N
6
5
3
H5
H15 H6 H14
9.1 9.0 8.9 8.8 8.7 8.6 8.5 8.4 8.3 8.2 8.1 8.0 7.9 7.8
F2 chemical shift (ppm)
125
126
127
128
129
130
131
132
133
134
135
136
137
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
S
15
14
N
6
5
C6
C5
C15
C14
Figure 4. IDR-HSQC-TOCSY spectrum of
naphtho[20,10:5,6]naphtho[20,10:4,5]-thieno[2,3-c]quinoline (2).
Correlation pathways for the H5/C5–H6/C6 and
H14/C14–H15/C15 resonant pairs are shown. The H6 and H15
protons, which resonate at 8.52 and 8.53 ppm, respectively,
correspond to the completely overlapped ‘doublet’ centered at
¾8.255 ppm. As a consequence of the overlap of these proton
resonances, both Type I and Type II artifact responses are
expected in the indirect covariance NMR spectrum of 2.
Artifact responses will also be expected for the overlapped
proton resonances near 8.05 and 7.80 ppm.
Referring to the IDR-HSQC-TOCSY spectrum of 2 shown
in Fig. 4, the correlations for the two two-spin systems are
denoted by solid black lines. The vicinal carbon–carbon
correlation for C5–C6 is expected in the indirect covari-
ance NMR spectrum at the carbon shifts, 134.4–125.4 ppm.
In a similar fashion, the vicinal carbon–carbon correla-
tion for C14–C15 will involve the carbons at 132.5 and
129.8 ppm. These correlations are denoted by solid black
lines in the indirect covariance NMR spectrum of 2 shown in
Fig. 5. With the sole exception of the correlation between
C2 and C3, which is not observed, all of the expected
vicinal correlation responses are observed in the spectra
shown in Figs 5 and 6. The vicinal spin system comprising
H2/C2–H3/C3 is problematic in terms of the HSQC-TOCSY
experiment, in general. The four resonances comprising
this contiguous protonated carbon pair are 7.79/134.7 and
7.83/134.4 ppm, respectively. In cases such as this, with very
similar proton and carbon chemical shifts for the resonant
pair, the relayed coherence response is frequently obscured
by the much more intense direct response. To an extent,
the IDR-HSQC-TOCSY experiment ameliorates the prob-
lem. However, in this case, both pairs of resonances are so
close that a response overlap in the two-dimensional NMR
data matrix and the consequent loss of the connectivity
information is almost unavoidable. There are two alterna-
tives in dealing with such a situation. One is to acquire
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
a conventional HXQC-TOCSY (X D M or S) is used as the
source of the data for indirect covariance processing. In
the case of the conventional HXQC-TOCSY spectrum, all
of the artifact responses will have identical phases since all
of the responses in the spectrum are positive. Hence Type
I artifact responses will be indistinguishable from Type II
artifact responses or from legitimate carbon–carbon vicinal
correlation responses. In this regard, it is preferable to utilize
IDR-HSQC-TOCSY when indirect covariance processing is
contemplated. The use of HSQC rather than the HMQC-
based variant of the experiment is also preferable based
on the improved F1 resolution of the former, which is
comparable to the difference in F1 resolution between an
HMQC and HSQC experiment.11
Moving from cyclopentafuranone (1) to naphtho[20,10:
5,6]naphtho[20,10:4,5]-thieno[2,3-c]quinoline (2), which has
a much more complex and overlapped proton spectrum
even at 600 MHz, as expected, both the IDR-HSQC-
TOCSY and indirect covariance NMR spectra are con-
siderably more complex. The IDR-HSQC-TOCSY spec-
trum of 2 is shown in Fig. 4. The total assignment of 2,
based on the use of IDR-HSQC-TOCSY data, has been
reported.11,12 For purposes of a discussion of the indi-
rect covariance NMR spectrum of 2, we will focus on
two isolated two-spin systems, H5/C5–H6/C6 (9.09/134.4
and 8.52/125.4 ppm, respectively) and H14/C14–H15/C15
(7.84/132.5 and 8.53/129.8 ppm, respectively). As would be
expected from the overlap of H6 and H15 resonating at
8.52 and 8.53 ppm, respectively, 2 has considerable poten-
tial for the observation of artifact responses in the indirect
covariance NMR spectrum. The location of the two-spin
systems on which the discussion of the indirect covari-
ance NMR spectrum will be based is shown by 3. The
desired vicinal carbon–carbon correlations are denoted in
the structure by the solid black lines. From the analysis
of the indirect covariance NMR spectrum of 2 shown in
Fig. 5, the solid red and dashed black lines, correspond-
ing to Type I and Type II artifacts, respectively, are also
shown. It is important to note that the lines signifying the
resonances leading to Type II artifacts cannot be drawn
in arbitrarily; the spectrum must be analyzed to deter-
mine which resonances are associated with Type II artifact
responses.
S
15
14
N
6
5
3
H5
H15 H6 H14
9.1 9.0 8.9 8.8 8.7 8.6 8.5 8.4 8.3 8.2 8.1 8.0 7.9 7.8
F2 chemical shift (ppm)
125
126
127
128
129
130
131
132
133
134
135
136
137
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
S
15
14
N
6
5
C6
C5
C15
C14
Figure 4. IDR-HSQC-TOCSY spectrum of
naphtho[20,10:5,6]naphtho[20,10:4,5]-thieno[2,3-c]quinoline (2).
Correlation pathways for the H5/C5–H6/C6 and
H14/C14–H15/C15 resonant pairs are shown. The H6 and H15
protons, which resonate at 8.52 and 8.53 ppm, respectively,
correspond to the completely overlapped ‘doublet’ centered at
¾8.255 ppm. As a consequence of the overlap of these proton
resonances, both Type I and Type II artifact responses are
expected in the indirect covariance NMR spectrum of 2.
Artifact responses will also be expected for the overlapped
proton resonances near 8.05 and 7.80 ppm.
Referring to the IDR-HSQC-TOCSY spectrum of 2 shown
in Fig. 4, the correlations for the two two-spin systems are
denoted by solid black lines. The vicinal carbon–carbon
correlation for C5–C6 is expected in the indirect covari-
ance NMR spectrum at the carbon shifts, 134.4–125.4 ppm.
In a similar fashion, the vicinal carbon–carbon correla-
tion for C14–C15 will involve the carbons at 132.5 and
129.8 ppm. These correlations are denoted by solid black
lines in the indirect covariance NMR spectrum of 2 shown in
Fig. 5. With the sole exception of the correlation between
C2 and C3, which is not observed, all of the expected
vicinal correlation responses are observed in the spectra
shown in Figs 5 and 6. The vicinal spin system comprising
H2/C2–H3/C3 is problematic in terms of the HSQC-TOCSY
experiment, in general. The four resonances comprising
this contiguous protonated carbon pair are 7.79/134.7 and
7.83/134.4 ppm, respectively. In cases such as this, with very
similar proton and carbon chemical shifts for the resonant
pair, the relayed coherence response is frequently obscured
by the much more intense direct response. To an extent,
the IDR-HSQC-TOCSY experiment ameliorates the prob-
lem. However, in this case, both pairs of resonances are so
close that a response overlap in the two-dimensional NMR
data matrix and the consequent loss of the connectivity
information is almost unavoidable. There are two alterna-
tives in dealing with such a situation. One is to acquire
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
Page 5
Elimination of artifacts in NMR spectra 1003
136 135 134 133 132 131 130 129 128 127 126 125
F2 chemical shift (ppm)
125
126
127
128
129
130
131
132
133
134
135
136
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
C6–C5
C15–C14
C6–C15C6–C14
C15–C3,
C15–C5
C4–C3
C14–C5
S
15
14
N
6
5
Figure 5. Indirect covariance NMR spectrum of
naphtho[20,10:5,6]naphtha-[20,10:4,5]thieno[2,3-c]quinoline (2).
The data were processed according to the method of Zhang
and Bru¨schweiler.3 Connectivities are shown for the C5–C6
and C14–C15 spin systems whose correlation pathways are
defined in the IDR-HSQC-TOCSY spectrum shown in Fig. 4.
The expected vicinal carbon–carbon connectivities between
C5–C6 and C14–C15 are denoted by off-diagonal correlations
linked by solid black lines. The vicinal carbon–carbon
correlation for C3–C4 (solid black line) overlaps the Type II
C15–C13 and C15–C5 artifact responses (dashed black line)
in the spectrum. The Type II artifact response labeled C6–C14
and the two Type I artifact responses labeled C6–C15 and
C14–C5 (denoted by solid red lines) are resolved and do not
represent multiple responses.
the data without broadband 13C decoupling during the
acquisition period. In the resulting coupled HXQC-TOCSY
or -ROESY spectrum [X D M (multiple) or D S (single)],
the direct responses are displaced by š1JCH/2, allow-
ing the relayed responses to be observed. Several reports
in the literature have used this approach successfully for
HMQC-TOCSY13 and HMQC-ROESY.14 The other alterna-
tive is to completely suppress the direct responses using
the SDR (suppressed direct response)-HSQC-TOCSY pulse
sequence.15 Rather than using a pulse sequence element
comprising 180° pulses applied to both proton and carbon
as in the IDR experiment, the SDR variant of the exper-
iment applies a 90°, 1H pulse and a 180°, 13C pulse to
completely suppress the direct response. Either of these
approaches can presumably establish the H2/C2–H3/C3
correlation. However, in the former case, calculation of
the indirect covariance NMR spectrum from a 13C cou-
pled HSQC-TOCSY experiment can produce only artifact
responses because of resonance overlap. In the case of the
SDR-HSQC-TOCSY spectrum, an indirect covariance NMR
spectrum cannot be calculated as there will be, in princi-
ple, only the relayed response at any given proton chemical
shift.
136.5 136.0 135.5 135.0 134.5 134.0 133.5 133.0 132.5 132.0 131.5 131.0 130.5 130.0
F
2
chemical shift (ppm)
129.5
130.0
130.5
131.0
131.5
132.0
132.5
133.0
133.5
134.0
134.5
135.0
135.5
136.0
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
C15
C10 C16C14
C17
C12 C13
C1 C2 C5
C11
C3
11–10
12–15 16 1514–15
17–16
12–14 5–14
13–1
12–13
13–2
1–2
12–2
12-1
2–15
15–3, 15–5
3–4
Figure 6. Annotated expansion of the indirect covariance NMR
spectrum of naphtho[20,10:5,6]naphtho[20,10:4,5]
thieno[2,3-c]quinoline (2) shown in Fig. 5. Legitimate
carbon–carbon vicinal correlations are denoted by solid black
lines. Type I artifact responses are denoted by solid red lines.
Type II artifact responses are denoted by dashed black lines. A
13C reference spectrum processed with 10-Hz exponential
broadening is plotted above the contour plot; a projection
through the F1 frequency domain is plotted along the left side
of the contour plot. Volume integration of the peaks in the
indirect covariance spectrum was used to determine the
overlap of the 3–4 carbon–carbon vicinal correlation response
(solid black line) with the Type II artifact responses labeled
15–3 and 15–5 (dashed black line).
As expected from the analysis of the IDR-HSQC-TOCSY
spectrum shown in Fig. 4, a pair of Type I responses is
observed in the indirect covariance NMR spectrum shown
in Fig. 5 between C5–C14 and C6–C15. These are denoted
in Fig. 5 by solid red lines. The Type I artifact responses
arise from a pair of either direct or relayed responses from
different spin systems as denoted by the solid red lines
shown for the strongly overlapped H15/H16 resonances
in F2 in Fig. 4. In addition, multiple Type II responses
are observed in Fig. 5 between the C15–C3, C15–C5, and
C6–C14 resonant pairs, these correlations are denoted by
dashed black lines. The Type II artifact responses arise
for a direct and relayed response, again from different
spin systems, as defined by the dashed black lines at
the F2 frequency of the overlapped H15/H16 resonances
in Fig. 4. A vicinal correlation response between C3 and
C4 overlaps the Type II C15–C3 and C15–C5 artifact
responses and is denoted by the solid black line in
Fig. 5.
Type I artifacts, since they arise from pairs of responses
in the IDR-HSQC-TOCSY spectrum with identical phases
will appear inverted in the indirect covariance spectrum, as
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
136 135 134 133 132 131 130 129 128 127 126 125
F2 chemical shift (ppm)
125
126
127
128
129
130
131
132
133
134
135
136
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
C6–C5
C15–C14
C6–C15C6–C14
C15–C3,
C15–C5
C4–C3
C14–C5
S
15
14
N
6
5
Figure 5. Indirect covariance NMR spectrum of
naphtho[20,10:5,6]naphtha-[20,10:4,5]thieno[2,3-c]quinoline (2).
The data were processed according to the method of Zhang
and Bru¨schweiler.3 Connectivities are shown for the C5–C6
and C14–C15 spin systems whose correlation pathways are
defined in the IDR-HSQC-TOCSY spectrum shown in Fig. 4.
The expected vicinal carbon–carbon connectivities between
C5–C6 and C14–C15 are denoted by off-diagonal correlations
linked by solid black lines. The vicinal carbon–carbon
correlation for C3–C4 (solid black line) overlaps the Type II
C15–C13 and C15–C5 artifact responses (dashed black line)
in the spectrum. The Type II artifact response labeled C6–C14
and the two Type I artifact responses labeled C6–C15 and
C14–C5 (denoted by solid red lines) are resolved and do not
represent multiple responses.
the data without broadband 13C decoupling during the
acquisition period. In the resulting coupled HXQC-TOCSY
or -ROESY spectrum [X D M (multiple) or D S (single)],
the direct responses are displaced by š1JCH/2, allow-
ing the relayed responses to be observed. Several reports
in the literature have used this approach successfully for
HMQC-TOCSY13 and HMQC-ROESY.14 The other alterna-
tive is to completely suppress the direct responses using
the SDR (suppressed direct response)-HSQC-TOCSY pulse
sequence.15 Rather than using a pulse sequence element
comprising 180° pulses applied to both proton and carbon
as in the IDR experiment, the SDR variant of the exper-
iment applies a 90°, 1H pulse and a 180°, 13C pulse to
completely suppress the direct response. Either of these
approaches can presumably establish the H2/C2–H3/C3
correlation. However, in the former case, calculation of
the indirect covariance NMR spectrum from a 13C cou-
pled HSQC-TOCSY experiment can produce only artifact
responses because of resonance overlap. In the case of the
SDR-HSQC-TOCSY spectrum, an indirect covariance NMR
spectrum cannot be calculated as there will be, in princi-
ple, only the relayed response at any given proton chemical
shift.
136.5 136.0 135.5 135.0 134.5 134.0 133.5 133.0 132.5 132.0 131.5 131.0 130.5 130.0
F
2
chemical shift (ppm)
129.5
130.0
130.5
131.0
131.5
132.0
132.5
133.0
133.5
134.0
134.5
135.0
135.5
136.0
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
C15
C10 C16C14
C17
C12 C13
C1 C2 C5
C11
C3
11–10
12–15 16 1514–15
17–16
12–14 5–14
13–1
12–13
13–2
1–2
12–2
12-1
2–15
15–3, 15–5
3–4
Figure 6. Annotated expansion of the indirect covariance NMR
spectrum of naphtho[20,10:5,6]naphtho[20,10:4,5]
thieno[2,3-c]quinoline (2) shown in Fig. 5. Legitimate
carbon–carbon vicinal correlations are denoted by solid black
lines. Type I artifact responses are denoted by solid red lines.
Type II artifact responses are denoted by dashed black lines. A
13C reference spectrum processed with 10-Hz exponential
broadening is plotted above the contour plot; a projection
through the F1 frequency domain is plotted along the left side
of the contour plot. Volume integration of the peaks in the
indirect covariance spectrum was used to determine the
overlap of the 3–4 carbon–carbon vicinal correlation response
(solid black line) with the Type II artifact responses labeled
15–3 and 15–5 (dashed black line).
As expected from the analysis of the IDR-HSQC-TOCSY
spectrum shown in Fig. 4, a pair of Type I responses is
observed in the indirect covariance NMR spectrum shown
in Fig. 5 between C5–C14 and C6–C15. These are denoted
in Fig. 5 by solid red lines. The Type I artifact responses
arise from a pair of either direct or relayed responses from
different spin systems as denoted by the solid red lines
shown for the strongly overlapped H15/H16 resonances
in F2 in Fig. 4. In addition, multiple Type II responses
are observed in Fig. 5 between the C15–C3, C15–C5, and
C6–C14 resonant pairs, these correlations are denoted by
dashed black lines. The Type II artifact responses arise
for a direct and relayed response, again from different
spin systems, as defined by the dashed black lines at
the F2 frequency of the overlapped H15/H16 resonances
in Fig. 4. A vicinal correlation response between C3 and
C4 overlaps the Type II C15–C3 and C15–C5 artifact
responses and is denoted by the solid black line in
Fig. 5.
Type I artifacts, since they arise from pairs of responses
in the IDR-HSQC-TOCSY spectrum with identical phases
will appear inverted in the indirect covariance spectrum, as
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
Page 6
1004 K. A. Blinov et al.
shown in Fig. 5. Type I artifacts can be visually rejected
on the basis of their phase during the analysis of an
indirect covariance NMR spectrum. In contrast, since the
Type II artifacts arise from a negatively phased, direct
and a positively phased relay response, albeit from dif-
ferent spin systems, they will have positive phase, and
there is no convenient means of visually identifying
these responses in a conventional indirect covariance spec-
trum other than through the analysis of the spectrum.
We will demonstrate, however, that these responses can
be manipulated and subsequently eliminated through a
modification of the indirect covariance–processing proto-
col described later followed by symmetrization as used
with diagonally symmetric spectra such as COSY and
TOCSY.
One other point is worth considering if we assume
that the transfer efficiency during the isotropic mixing
period is identical for all of the relayed coherence trans-
fer responses in an HSQC-TOCSY spectrum. This will
only be approximately valid in a case such as that rep-
resented by 2 when all of the vicinal proton–proton cou-
pling constants are more or less identical. In such a case,
Type II artifact responses, which arise only once in a
spectrum, e.g. the response for C6–C15 (125.4–129.7 ppm
in Fig. 6), will have only half of the observed response
intensity of a legitimate carbon–carbon vicinal correlation
response, e.g. C5–C6 (134.5–125.4 ppm in Fig. 6), in the indi-
rect covariance spectrum. This statement is based on the
observation that the C5–C6 vicinal carbon–carbon corre-
lation responses in the indirect covariance spectrum arise
from the pairs of direct and vicinal relayed responses
observed at the proton frequencies of both H5 and H6
in the F2 frequency domain. This observation was con-
firmed for the indirect covariance spectrum of 2 shown
in Fig. 6 by volume integration of the responses in the data
matrix.
The more complex and congested downfield region of
the indirect covariance NMR spectrum of 2 is presented in
Fig. 6. The correlations are denoted by the same convention
used in Fig. 5. To illustrate the distribution of Type I
and II artifact responses relative to the structure of the
molecule, all of the artifact correlations observed in Figs 5
and 6 are collected on 4. The expected carbon–carbon
vicinal correlation responses between contiguous protonated
carbons are not shown. Although the total number of artifact
responses might initially be assumed to render indirect
covariance NMR undesirable, it should be kept in mind
that all of the Type I artifact responses can be ignored on
simple visual inspection of the data matrix. There are only
a total of five Type II artifact responses contained in data
shown in Figs 5 and 6. Four of the Type II artifacts have
response intensity that makes them indistinguishable from
authentic carbon–carbon vicinal correlation responses by
casual visual inspection. Since the vicinal proton–proton
coupling constants are relatively uniform in a molecule
such as 2, the Type II artifact responses are, however,
distinguishable based on volume integration as discussed
above.
S
15
14
10
11
13
12
8
N
6
5
16
17
4
3
2
1134.9
134.7
134.4
129.8
134.5
125.4
144.4
135.3
136.3
134.6
131.4
132.5
129.7
131.2
4
Elimination of artifact responses through
unsymmetrical covariance processing
The phase characteristics of the responses contained in
the IDR-HSQC-TOCSY spectrum allow the direct (negative
phase) and relayed or TOCSY responses (positive phase) to
be considered separately. Hence, as shown schematically in
Fig. 7, the IDR-HSQC-TOCSY spectrum can be decomposed
into a pair of data matrices, one containing only the negative-
phase direct responses and the other the positive-phase
relayed or TOCSY responses. After decomposition of the
IDR-HSQC-TOCSY spectrum into a positive- and negative-
phase matrix, the pair of matrices can be subjected to the
covariance procedure according to the following scheme.
In the usual covariance-processing protocol, diagonally
symmetric peaks arise from equal rows:
Row I ð Row J
and
Row J ð Row I
In the modified covariance procedure proposed in this
work, diagonally symmetric peaks arise from different rows
in the fashion:
CRow I ð Row J
and
CRow J ð Row I
This process is shown schematically in Fig. 8. When the
process is repeated for the other pair of responses that
comprise the usual pattern of responses in an IDR-HSQC-
TOCSY spectrum, the intensity of the diagonally symmetric
responses is effectively doubled (assuming equivalent cou-
pling constants and efficiency of magnetization transfer
during the mixing period).
In the case of artifact responses, generally, these responses
will only occur at a single proton chemical shift, that of the
overlap of the proton resonances in question. Consequently,
the artifact responses in the phase-separated data matrices,
when subjected to covariance processing, will give only a
single response, rather than a diagonally symmetric pair of
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
shown in Fig. 5. Type I artifacts can be visually rejected
on the basis of their phase during the analysis of an
indirect covariance NMR spectrum. In contrast, since the
Type II artifacts arise from a negatively phased, direct
and a positively phased relay response, albeit from dif-
ferent spin systems, they will have positive phase, and
there is no convenient means of visually identifying
these responses in a conventional indirect covariance spec-
trum other than through the analysis of the spectrum.
We will demonstrate, however, that these responses can
be manipulated and subsequently eliminated through a
modification of the indirect covariance–processing proto-
col described later followed by symmetrization as used
with diagonally symmetric spectra such as COSY and
TOCSY.
One other point is worth considering if we assume
that the transfer efficiency during the isotropic mixing
period is identical for all of the relayed coherence trans-
fer responses in an HSQC-TOCSY spectrum. This will
only be approximately valid in a case such as that rep-
resented by 2 when all of the vicinal proton–proton cou-
pling constants are more or less identical. In such a case,
Type II artifact responses, which arise only once in a
spectrum, e.g. the response for C6–C15 (125.4–129.7 ppm
in Fig. 6), will have only half of the observed response
intensity of a legitimate carbon–carbon vicinal correlation
response, e.g. C5–C6 (134.5–125.4 ppm in Fig. 6), in the indi-
rect covariance spectrum. This statement is based on the
observation that the C5–C6 vicinal carbon–carbon corre-
lation responses in the indirect covariance spectrum arise
from the pairs of direct and vicinal relayed responses
observed at the proton frequencies of both H5 and H6
in the F2 frequency domain. This observation was con-
firmed for the indirect covariance spectrum of 2 shown
in Fig. 6 by volume integration of the responses in the data
matrix.
The more complex and congested downfield region of
the indirect covariance NMR spectrum of 2 is presented in
Fig. 6. The correlations are denoted by the same convention
used in Fig. 5. To illustrate the distribution of Type I
and II artifact responses relative to the structure of the
molecule, all of the artifact correlations observed in Figs 5
and 6 are collected on 4. The expected carbon–carbon
vicinal correlation responses between contiguous protonated
carbons are not shown. Although the total number of artifact
responses might initially be assumed to render indirect
covariance NMR undesirable, it should be kept in mind
that all of the Type I artifact responses can be ignored on
simple visual inspection of the data matrix. There are only
a total of five Type II artifact responses contained in data
shown in Figs 5 and 6. Four of the Type II artifacts have
response intensity that makes them indistinguishable from
authentic carbon–carbon vicinal correlation responses by
casual visual inspection. Since the vicinal proton–proton
coupling constants are relatively uniform in a molecule
such as 2, the Type II artifact responses are, however,
distinguishable based on volume integration as discussed
above.
S
15
14
10
11
13
12
8
N
6
5
16
17
4
3
2
1134.9
134.7
134.4
129.8
134.5
125.4
144.4
135.3
136.3
134.6
131.4
132.5
129.7
131.2
4
Elimination of artifact responses through
unsymmetrical covariance processing
The phase characteristics of the responses contained in
the IDR-HSQC-TOCSY spectrum allow the direct (negative
phase) and relayed or TOCSY responses (positive phase) to
be considered separately. Hence, as shown schematically in
Fig. 7, the IDR-HSQC-TOCSY spectrum can be decomposed
into a pair of data matrices, one containing only the negative-
phase direct responses and the other the positive-phase
relayed or TOCSY responses. After decomposition of the
IDR-HSQC-TOCSY spectrum into a positive- and negative-
phase matrix, the pair of matrices can be subjected to the
covariance procedure according to the following scheme.
In the usual covariance-processing protocol, diagonally
symmetric peaks arise from equal rows:
Row I ð Row J
and
Row J ð Row I
In the modified covariance procedure proposed in this
work, diagonally symmetric peaks arise from different rows
in the fashion:
CRow I ð Row J
and
CRow J ð Row I
This process is shown schematically in Fig. 8. When the
process is repeated for the other pair of responses that
comprise the usual pattern of responses in an IDR-HSQC-
TOCSY spectrum, the intensity of the diagonally symmetric
responses is effectively doubled (assuming equivalent cou-
pling constants and efficiency of magnetization transfer
during the mixing period).
In the case of artifact responses, generally, these responses
will only occur at a single proton chemical shift, that of the
overlap of the proton resonances in question. Consequently,
the artifact responses in the phase-separated data matrices,
when subjected to covariance processing, will give only a
single response, rather than a diagonally symmetric pair of
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
Page 7
Elimination of artifacts in NMR spectra 1005
IDR-HSQC-TOCSY data matrix
Row I
Row J
Positive matrix
(relayed responses)
Negative matrix
(direct)
Direct responsesRelayed orTOCSY
responses
(positive)
Row I
Row J
Row I
J
Figure 7. Schematic representation of the decomposition of an
IDR-HSQC-TOCSY data matrix into a positive and negative
matrix based on response phase. Inverted direct responses are
collected in the negative matrix while the TOCSY or relayed
responses are collected in the positive matrix.
Row I
Positive matrix
(long-range)
Row j
Row I
Negative matrix
(direct)
Row j
Covariance matrix
zRow I
Co
lu
m
n
J
Co
lu
m
n
I
Row J
Figure 8. Schematic representation of the covariance
processing of an IDR-HSQC-TOCSY spectrum decomposed
into positive (TOCSY or relayed responses) and negative (direct
correlation responses) data matrices. For a vicinally coupled
pair of resonances in the IDR-HSQC-TOCSY spectrum, e.g.
represented by the H5/C6, H5/C5, H6/C6 and H6/C5
responses in Fig. 4, the diagonal carbon–carbon correlation
response arises as shown above from the multiplication of
CRow I ð Row J and CRow J ð Row I. The phase of
the diagonally symmetric carbon–carbon vicinal correlation
response is negative, denoted by the black responses in the
covariance matrix on the right.
responses as in the case of a legitimate carbon–carbon vicinal
correlation response. This result is shown schematically in
Fig. 9.
To illustrate what we have shown schematically in
Figs 7–9, the IDR-HSQC-TOCSY data set presented in Fig. 4
was used as an example. To simplify the presentation for
purposes of this discussion, all of the responses in the
Row I
Positive matrix
(long-range)
Row J
Row I
Negative matrix
(direct)
Row J
Covariance matrix
Row I
Co
lu
m
n
J
Co
lu
m
n
I
Row J
Figure 9. Schematic representation of the behavior of artifact
responses during covariance processing. The positive and
negative-phase responses that would give rise to a Type II
occur only at the proton shift in what would correspond to
column I. When the corresponding rows are multiplied,
CRow I ð Row J, a single response is produced in the
resulting covariance matrix at the intersection of Row I and
Column J. There is no diagonally symmetric response as in the
case of a legitimate carbon–carbon vicinal correlation
response shown schematically in Fig. 8.
data matrix except for those responsible for causing the
appearance of artifact responses in Fig. 5 were replaced by
noise taken from a random location on the noise floor of
the data matrix. The correspondingly modified IDR-HSQC-
TOCSY data matrix is shown in the left panel of Fig. 10.
When these data are subjected to the covariance procedure,
the resulting data matrix is shown in the right panel of Fig. 10.
Legitimate carbon–carbon vicinal correlation responses are
diagonally symmetric as expected. Artifact responses are
diagonally asymmetric and lend themselves to removal by
the symmetrization routine that has been used for many
years with COSY spectra. When the full IDR-HSQC-TOCSY
data matrix shown in Fig. 4 is subjected to the covariance-
processing protocol shown schematically in Figs 7–9, the
resulting spectrum is shown in Fig. 11.
CONCLUSIONS
Indirect covariance NMR represents a new direction in that it
offers the means of obtaining a homonuclear 2D NMR spec-
trum of the insensitive nuclide of a heteronuclear resonant
pair via detection of the sensitive member of the pair. Increas-
ing availability and access to cryogenic NMR probes16 – 18 can
be expected to enhance the utilization of lower sensitivity het-
eronuclear 2D NMR experiments such as the HSQC-TOCSY
experiment that is processed to yield the indirect covariance
NMR spectra discussed in this report.4 The acquisition of
IDR-HSQC-TOCSY rather than conventional HSQC-TOCSY
is advantageous in that Type I responses caused by overlap-
ping proton resonances have phase characteristics opposite
from the vicinal carbon–carbon correlation responses of
interest, allowing them to be quickly eliminated by simple
visual inspection. Type II artifact responses are generally
visually indistinguishable from legitimate carbon–carbon
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
IDR-HSQC-TOCSY data matrix
Row I
Row J
Positive matrix
(relayed responses)
Negative matrix
(direct)
Direct responsesRelayed orTOCSY
responses
(positive)
Row I
Row J
Row I
J
Figure 7. Schematic representation of the decomposition of an
IDR-HSQC-TOCSY data matrix into a positive and negative
matrix based on response phase. Inverted direct responses are
collected in the negative matrix while the TOCSY or relayed
responses are collected in the positive matrix.
Row I
Positive matrix
(long-range)
Row j
Row I
Negative matrix
(direct)
Row j
Covariance matrix
zRow I
Co
lu
m
n
J
Co
lu
m
n
I
Row J
Figure 8. Schematic representation of the covariance
processing of an IDR-HSQC-TOCSY spectrum decomposed
into positive (TOCSY or relayed responses) and negative (direct
correlation responses) data matrices. For a vicinally coupled
pair of resonances in the IDR-HSQC-TOCSY spectrum, e.g.
represented by the H5/C6, H5/C5, H6/C6 and H6/C5
responses in Fig. 4, the diagonal carbon–carbon correlation
response arises as shown above from the multiplication of
CRow I ð Row J and CRow J ð Row I. The phase of
the diagonally symmetric carbon–carbon vicinal correlation
response is negative, denoted by the black responses in the
covariance matrix on the right.
responses as in the case of a legitimate carbon–carbon vicinal
correlation response. This result is shown schematically in
Fig. 9.
To illustrate what we have shown schematically in
Figs 7–9, the IDR-HSQC-TOCSY data set presented in Fig. 4
was used as an example. To simplify the presentation for
purposes of this discussion, all of the responses in the
Row I
Positive matrix
(long-range)
Row J
Row I
Negative matrix
(direct)
Row J
Covariance matrix
Row I
Co
lu
m
n
J
Co
lu
m
n
I
Row J
Figure 9. Schematic representation of the behavior of artifact
responses during covariance processing. The positive and
negative-phase responses that would give rise to a Type II
occur only at the proton shift in what would correspond to
column I. When the corresponding rows are multiplied,
CRow I ð Row J, a single response is produced in the
resulting covariance matrix at the intersection of Row I and
Column J. There is no diagonally symmetric response as in the
case of a legitimate carbon–carbon vicinal correlation
response shown schematically in Fig. 8.
data matrix except for those responsible for causing the
appearance of artifact responses in Fig. 5 were replaced by
noise taken from a random location on the noise floor of
the data matrix. The correspondingly modified IDR-HSQC-
TOCSY data matrix is shown in the left panel of Fig. 10.
When these data are subjected to the covariance procedure,
the resulting data matrix is shown in the right panel of Fig. 10.
Legitimate carbon–carbon vicinal correlation responses are
diagonally symmetric as expected. Artifact responses are
diagonally asymmetric and lend themselves to removal by
the symmetrization routine that has been used for many
years with COSY spectra. When the full IDR-HSQC-TOCSY
data matrix shown in Fig. 4 is subjected to the covariance-
processing protocol shown schematically in Figs 7–9, the
resulting spectrum is shown in Fig. 11.
CONCLUSIONS
Indirect covariance NMR represents a new direction in that it
offers the means of obtaining a homonuclear 2D NMR spec-
trum of the insensitive nuclide of a heteronuclear resonant
pair via detection of the sensitive member of the pair. Increas-
ing availability and access to cryogenic NMR probes16 – 18 can
be expected to enhance the utilization of lower sensitivity het-
eronuclear 2D NMR experiments such as the HSQC-TOCSY
experiment that is processed to yield the indirect covariance
NMR spectra discussed in this report.4 The acquisition of
IDR-HSQC-TOCSY rather than conventional HSQC-TOCSY
is advantageous in that Type I responses caused by overlap-
ping proton resonances have phase characteristics opposite
from the vicinal carbon–carbon correlation responses of
interest, allowing them to be quickly eliminated by simple
visual inspection. Type II artifact responses are generally
visually indistinguishable from legitimate carbon–carbon
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
Page 8
1006 K. A. Blinov et al.
9.0 8.5 8.0
F2 chemical shift (ppm)
124
126
128
130
132
134
136
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
134 132 130 128 126
F2 chemical shift (ppm)
125
126
127
128
129
130
131
132
133
134
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
Vicinal carbon–carbon
correlation (diagonally
symmetric)
Artifact responses
(asymmetric)
Figure 10. The IDR-HSQC-TOCSY spectrum of 2 is shown in the left panel. All of the responses in the data matrix except for those
responsible for the artifacts observed in Fig. 5 have been replaced by noise taken from a random location on the noise floor of the
spectrum. By subjecting the data matrix on the left to decomposition, as shown schematically in Fig. 7, followed by covariance
processing as illustrated schematically in Figs 8 and 9, the data matrix shown in the right panel is the indirect covariance result of
the manipulation. The asymmetric nature of the Type II artifact responses allows their removal by simple symmetrization of the type
used with homonuclear COSY data.
136 135 134 133 132 131 130 129 128 127 126 125
F2 chemical shift (ppm)
124
125
126
127
128
129
130
131
132
133
134
135
136
137
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
C6–C5
C15–C14
C4–C3
Figure 11. When the full IDR-HSQC-TOCSY spectrum of 2
shown in Fig. 4 is subjected to the processing scheme
illustrated by Figs 7–9, the experimental result is shown above.
Correlations for three vicinal carbon–carbon correlations are
denoted by solid black lines. Two of these correspond to the
C5–C6 and C14–C15 correlations. The other arises from the
C3–C4 correlation and is shown because one of the
correlations from this diagonally symmetric pair overlaps the
Type II artifact response that was shown in the right panel of
Fig. 10. A 10-Hz broadened 13C reference spectrum is plotted
above the contour plot to provide a better representation of the
position of all of the carbon resonances in the spectrum of 2.
correlations and can only be differentiated from the latter
by the analysis of the indirect covariance spectrum in par-
allel with the IDR-HSQC-TOCSY data or through volume
integration in the particular case when all of the vicinal pro-
ton–proton homonuclear couplings are nearly identical as in
the case of a polynuclear aromatic or heteroaromatic such as
2. In the congested region of the 13C spectrum of IDR-HSQC-
TOCSY spectrum of 2 ranging from about 133–137 ppm,
the indirect covariance NMR spectrum (compare Figs 4 and
6) may offer some advantages over the IDR-HSQC-TOCSY
spectrum in terms of interpretability, although the issue of
Type II artifact responses must be considered.
In the second half of the present study, we have shown
that a modified or unsymmetrical covariance-processing
scheme applied to IDR-HSQC-TOCSY data can be used
in conjunction with symmetrization to remove artifacts
from indirect covariance NMR spectra, thereby eliminating
ambiguities that could arise in the interpretation of these
data when dealing with an unknown chemical structure. This
approach should allow the indirect covariance method to be
applied to a diverse range of organic molecules including
molecules of considerable complexity.
One of our primary interests in indirect covariance trans-
formation is to use this form of processing to derive poten-
tially beneficial carbon–carbon connectivity information
from IDR-HSQC-TOCSY spectra for the data input matrix for
computer-assisted structure elucidation (CASE) programs
such as ACD/Structure Elucidator.19 – 22 Carbon–carbon con-
nectivity information removes any ambiguities that could
result from the extraction of such information from multiple
2D NMR spectra (direct and long-range heteronuclear cor-
relation experiments, for example). An array of multiple
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
9.0 8.5 8.0
F2 chemical shift (ppm)
124
126
128
130
132
134
136
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
134 132 130 128 126
F2 chemical shift (ppm)
125
126
127
128
129
130
131
132
133
134
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
Vicinal carbon–carbon
correlation (diagonally
symmetric)
Artifact responses
(asymmetric)
Figure 10. The IDR-HSQC-TOCSY spectrum of 2 is shown in the left panel. All of the responses in the data matrix except for those
responsible for the artifacts observed in Fig. 5 have been replaced by noise taken from a random location on the noise floor of the
spectrum. By subjecting the data matrix on the left to decomposition, as shown schematically in Fig. 7, followed by covariance
processing as illustrated schematically in Figs 8 and 9, the data matrix shown in the right panel is the indirect covariance result of
the manipulation. The asymmetric nature of the Type II artifact responses allows their removal by simple symmetrization of the type
used with homonuclear COSY data.
136 135 134 133 132 131 130 129 128 127 126 125
F2 chemical shift (ppm)
124
125
126
127
128
129
130
131
132
133
134
135
136
137
F 1
ch
em
ic
al
s
hi
ft
(pp
m)
C6–C5
C15–C14
C4–C3
Figure 11. When the full IDR-HSQC-TOCSY spectrum of 2
shown in Fig. 4 is subjected to the processing scheme
illustrated by Figs 7–9, the experimental result is shown above.
Correlations for three vicinal carbon–carbon correlations are
denoted by solid black lines. Two of these correspond to the
C5–C6 and C14–C15 correlations. The other arises from the
C3–C4 correlation and is shown because one of the
correlations from this diagonally symmetric pair overlaps the
Type II artifact response that was shown in the right panel of
Fig. 10. A 10-Hz broadened 13C reference spectrum is plotted
above the contour plot to provide a better representation of the
position of all of the carbon resonances in the spectrum of 2.
correlations and can only be differentiated from the latter
by the analysis of the indirect covariance spectrum in par-
allel with the IDR-HSQC-TOCSY data or through volume
integration in the particular case when all of the vicinal pro-
ton–proton homonuclear couplings are nearly identical as in
the case of a polynuclear aromatic or heteroaromatic such as
2. In the congested region of the 13C spectrum of IDR-HSQC-
TOCSY spectrum of 2 ranging from about 133–137 ppm,
the indirect covariance NMR spectrum (compare Figs 4 and
6) may offer some advantages over the IDR-HSQC-TOCSY
spectrum in terms of interpretability, although the issue of
Type II artifact responses must be considered.
In the second half of the present study, we have shown
that a modified or unsymmetrical covariance-processing
scheme applied to IDR-HSQC-TOCSY data can be used
in conjunction with symmetrization to remove artifacts
from indirect covariance NMR spectra, thereby eliminating
ambiguities that could arise in the interpretation of these
data when dealing with an unknown chemical structure. This
approach should allow the indirect covariance method to be
applied to a diverse range of organic molecules including
molecules of considerable complexity.
One of our primary interests in indirect covariance trans-
formation is to use this form of processing to derive poten-
tially beneficial carbon–carbon connectivity information
from IDR-HSQC-TOCSY spectra for the data input matrix for
computer-assisted structure elucidation (CASE) programs
such as ACD/Structure Elucidator.19 – 22 Carbon–carbon con-
nectivity information removes any ambiguities that could
result from the extraction of such information from multiple
2D NMR spectra (direct and long-range heteronuclear cor-
relation experiments, for example). An array of multiple
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
Page 9
Elimination of artifacts in NMR spectra 1007
carbon–carbon connectivities, including partial substruc-
tures extracted from this approach, can reduce the challenge
of generating resultant structures from the CASE system,
which is presently being evaluated. The removal of the
artifacts using the unsymmetrical covariance–processing
scheme discussed in this work precludes using potentially
incorrect carbon–carbon connectivity information in the data
input matrix for a CASE program and further helps in
improving the utility of this approach to novel chemical
structure determination.
EXPERIMENTAL
The sample of cyclopentafuranone (1) was obtained from
Sigma–Aldrich. The sample of 2 used in this study was
synthetically prepared.11,12 Samples were prepared for NMR
data acquisition by dissolving ¾5 mg of each compound in
deuterochloroform in a 3-mm NMR tube.
NMR experiments performed on 1 used a Varian
Inova 500 MHz NMR; experiments performed on 2 used
a Varian Inova 600 MHz spectrometer. Both instruments
were equipped with 3 mm Nalorac micro inverse detection
gradient NMR probes. A mixing time of 18 ms was used
to acquire the IDR-HSQC-TOCSY spectra of both 1 and 2
shown in Figs 1 and 4, respectively. The IDR-HSQC-TOCSY
spectrum of 1 was acquired using 512 points in F2 with 96
increments of the evolution time, t1. Data were processed
by zero-filling to 1024 in t2 and linear predicting from 96
to 128 points followed by zero-filling to 256 points in the
second time domain. Gaussian multiplication was used in
both time domains. The IDR-HSQC-TOCSY spectrum of 2
was acquired as 512 points in t2 and again zero-filled to 1024
points; the experiment was digitized with 160 increments of
the evolution time, t1, in the second frequency domain; the
data were linear predicted to 256 points and zero-filled to
512 points in F1 during processing. Gaussian multiplication
was used in both time domains.
The indirect covariance NMR spectra were computed
with the method of Zhang and Bru¨schweiler,4 using the
2D NMR Manager module of ACD/SpecManager23 v8.2.
The processing was performed using a PC with a 2.8 GHz
Pentium IV processor with 1 Gbyte of RAM. The calculation
of the indirect covariance NMR spectra from the processed
IDR-HSQC-TOCSY data took approximately 4 s.
REFERENCES
1. Bru¨schweiler R, Zhang F. J. Chem. Phys. 2004; 120: 5253.
2. Bru¨schweiler R. J. Chem. Phys. 2004; 121: 409.
3. Trbovic N, Smirnov S, Zhang F, Bru¨schweiler R. J. Magn. Reson.
2004; 171: 277.
4. Zhang F, Bru¨schweiler R. J. Am. Chem. Soc. 2004; 126: 13 180.
5. Turner DL. Mol. Phys. 1981; 44: 1051.
6. Turner DL. J. Magn. Reson. 1982; 49: 175.
7. Turner DL. J. Magn. Reson. 1983; 53: 259.
8. Musmar MJ, Willcott MR III, Martin GE, Gampe RT Jr, Iwao M,
Lee ML, Hurd RE, Johnson LF, Castle RN. J. Heterocycl. Chem.
1983; 20: 1661.
9. Domke T. J. Magn. Reson. 1991; 95: 174.
10. Crouch RC, Davis AO, Martin GE. Magn. Reson. Chem. 1995; 33:
889.
11. Hadden CE, Martin GE, Luo J-K, Castle RN. J. Heterocycl. Chem.
1999; 36: 533.
12. Hadden CE, Martin GE, Luo J-K, Castle RN. J. Heterocycl. Chem.
2000; 37: 821.
13. Crouch RC, McFadyen RB, Daluge SM, Martin GE. Magn. Reson.
Chem. 1990; 28: 792.
14. Kawabata J, Fukushi E, Mizutani J. J. Am. Chem. Soc. 1992; 114:
1115.
15. Martin GE, Spitzer TD, Crouch RC, Luo J-K, Castle RN. J.
Heterocycl. Chem. 1992; 29: 577.
16. Martin GE. Cryogenic NMR probes: applications. In Encyclopedia
of Nuclear Magnetic Resonance, Vol. 9: Advances in NMR,
Grant DM, Harris RK (eds). Wiley: Chichester, 2002; 33.
17. Martin GE. Applications of cryogenic NMR probe technology
for the identification of low-level impurities in pharmaceuticals.
In Handbook of Modern Magnetic Resonance, vol. 2, Craik DJ,
Webb GA (eds). Elsevier: Amsterdam, 2005.
18. Martin GE. Small volume and high sensitivity NMR probes. In
Ann. Report NMR Spectrosc., vol. 56, Webb GA (ed). Elsevier:
Amsterdam, 2005.
19. Blinov KA, Carlson D, Elyashberg ME, Martin GE, Mar-
tirosian ER, Molodtsov SG, Williams AJ. Magn. Reson. Chem.
2003; 41: 359.
20. Elyashberg ME, Blinov KA, Martirosian ER, Molodtsov SG,
Williams AJ, Martin GE. J. Heterocycl. Chem. 2003; 40: 1017.
21. Elyashberg ME, Blino KA, Molodtsov SG, Williams AJ,
Martin GE. J. Chem. Inf. Comput. Sci. 2004; 44: 771.
22. Molodtsov SG, Elyashberg ME, Blinov KA, Williams AJ,
Martin GE. J. Chem. Inf. Comput. Sci. 2004; 44: 1737.
23. ACD SpecManager, Advanced Chemistry Development
(ACD/Labs), 110 Yonge Street, 14th floor, Toronto M5C 1T4,
(http://www.acdlabs.com/products/spec lab/exp spectra/
2d nmr/).
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
carbon–carbon connectivities, including partial substruc-
tures extracted from this approach, can reduce the challenge
of generating resultant structures from the CASE system,
which is presently being evaluated. The removal of the
artifacts using the unsymmetrical covariance–processing
scheme discussed in this work precludes using potentially
incorrect carbon–carbon connectivity information in the data
input matrix for a CASE program and further helps in
improving the utility of this approach to novel chemical
structure determination.
EXPERIMENTAL
The sample of cyclopentafuranone (1) was obtained from
Sigma–Aldrich. The sample of 2 used in this study was
synthetically prepared.11,12 Samples were prepared for NMR
data acquisition by dissolving ¾5 mg of each compound in
deuterochloroform in a 3-mm NMR tube.
NMR experiments performed on 1 used a Varian
Inova 500 MHz NMR; experiments performed on 2 used
a Varian Inova 600 MHz spectrometer. Both instruments
were equipped with 3 mm Nalorac micro inverse detection
gradient NMR probes. A mixing time of 18 ms was used
to acquire the IDR-HSQC-TOCSY spectra of both 1 and 2
shown in Figs 1 and 4, respectively. The IDR-HSQC-TOCSY
spectrum of 1 was acquired using 512 points in F2 with 96
increments of the evolution time, t1. Data were processed
by zero-filling to 1024 in t2 and linear predicting from 96
to 128 points followed by zero-filling to 256 points in the
second time domain. Gaussian multiplication was used in
both time domains. The IDR-HSQC-TOCSY spectrum of 2
was acquired as 512 points in t2 and again zero-filled to 1024
points; the experiment was digitized with 160 increments of
the evolution time, t1, in the second frequency domain; the
data were linear predicted to 256 points and zero-filled to
512 points in F1 during processing. Gaussian multiplication
was used in both time domains.
The indirect covariance NMR spectra were computed
with the method of Zhang and Bru¨schweiler,4 using the
2D NMR Manager module of ACD/SpecManager23 v8.2.
The processing was performed using a PC with a 2.8 GHz
Pentium IV processor with 1 Gbyte of RAM. The calculation
of the indirect covariance NMR spectra from the processed
IDR-HSQC-TOCSY data took approximately 4 s.
REFERENCES
1. Bru¨schweiler R, Zhang F. J. Chem. Phys. 2004; 120: 5253.
2. Bru¨schweiler R. J. Chem. Phys. 2004; 121: 409.
3. Trbovic N, Smirnov S, Zhang F, Bru¨schweiler R. J. Magn. Reson.
2004; 171: 277.
4. Zhang F, Bru¨schweiler R. J. Am. Chem. Soc. 2004; 126: 13 180.
5. Turner DL. Mol. Phys. 1981; 44: 1051.
6. Turner DL. J. Magn. Reson. 1982; 49: 175.
7. Turner DL. J. Magn. Reson. 1983; 53: 259.
8. Musmar MJ, Willcott MR III, Martin GE, Gampe RT Jr, Iwao M,
Lee ML, Hurd RE, Johnson LF, Castle RN. J. Heterocycl. Chem.
1983; 20: 1661.
9. Domke T. J. Magn. Reson. 1991; 95: 174.
10. Crouch RC, Davis AO, Martin GE. Magn. Reson. Chem. 1995; 33:
889.
11. Hadden CE, Martin GE, Luo J-K, Castle RN. J. Heterocycl. Chem.
1999; 36: 533.
12. Hadden CE, Martin GE, Luo J-K, Castle RN. J. Heterocycl. Chem.
2000; 37: 821.
13. Crouch RC, McFadyen RB, Daluge SM, Martin GE. Magn. Reson.
Chem. 1990; 28: 792.
14. Kawabata J, Fukushi E, Mizutani J. J. Am. Chem. Soc. 1992; 114:
1115.
15. Martin GE, Spitzer TD, Crouch RC, Luo J-K, Castle RN. J.
Heterocycl. Chem. 1992; 29: 577.
16. Martin GE. Cryogenic NMR probes: applications. In Encyclopedia
of Nuclear Magnetic Resonance, Vol. 9: Advances in NMR,
Grant DM, Harris RK (eds). Wiley: Chichester, 2002; 33.
17. Martin GE. Applications of cryogenic NMR probe technology
for the identification of low-level impurities in pharmaceuticals.
In Handbook of Modern Magnetic Resonance, vol. 2, Craik DJ,
Webb GA (eds). Elsevier: Amsterdam, 2005.
18. Martin GE. Small volume and high sensitivity NMR probes. In
Ann. Report NMR Spectrosc., vol. 56, Webb GA (ed). Elsevier:
Amsterdam, 2005.
19. Blinov KA, Carlson D, Elyashberg ME, Martin GE, Mar-
tirosian ER, Molodtsov SG, Williams AJ. Magn. Reson. Chem.
2003; 41: 359.
20. Elyashberg ME, Blinov KA, Martirosian ER, Molodtsov SG,
Williams AJ, Martin GE. J. Heterocycl. Chem. 2003; 40: 1017.
21. Elyashberg ME, Blino KA, Molodtsov SG, Williams AJ,
Martin GE. J. Chem. Inf. Comput. Sci. 2004; 44: 771.
22. Molodtsov SG, Elyashberg ME, Blinov KA, Williams AJ,
Martin GE. J. Chem. Inf. Comput. Sci. 2004; 44: 1737.
23. ACD SpecManager, Advanced Chemistry Development
(ACD/Labs), 110 Yonge Street, 14th floor, Toronto M5C 1T4,
(http://www.acdlabs.com/products/spec lab/exp spectra/
2d nmr/).
Copyright 2005 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2005; 43: 999–1007
Sign up today - FREE
Mendeley saves you time finding and organizing research. Learn more
- All your research in one place
- Add and import papers easily
- Access it anywhere, anytime
Start using Mendeley in seconds!
Readership Statistics
9 Readers on Mendeley
by Discipline
89% Chemistry
by Academic Status
33% Researcher (at a non-Academic Institution)
22% Student (Master)
22% Ph.D. Student
by Country
33% United Kingdom
22% United States
11% Russia