JP2540472B2 - Multi-quantum NMR using frequency-modulated chirp pulse - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention provides a nuclear spin system to which a first π
/ 2 pulse, generated after the elapse of the defocusing time interval τ following the first π / 2 pulse and after the elapse of another time interval τ following the first π pulse Excitation of multiple quanta, especially two-quantum coherence, in an NMR pulse experiment by irradiating a first series of RF pulses comprising a fundamental sandwich consisting of a second π / 2 pulse The present invention relates to a method of reconverting coherence into longitudinal magnetization by irradiating the nuclear spin system with a second series of RF pulses to induce free induction decay.

A method of the kind described above is described by DP Weitekamp in Adv.Magn.Reson. 11 , 111 (198
3) Known from scientific publications.

Many modern NMR methods make use of the phenomenon of transverse magnetization, or more generally p-quantum coherence transfer from one spin (or set of spins) to another. Among the best known examples are correlation spectroscopy and multi-quantum NMR. Many of these experiments are based on the extensive use of said "hard" pulses, which are said to induce rotation of the density operator. In practice, such a pulse is not stiff enough to be truly offset-independent, and thus off-resonance.
ance) effect (tilted effective magnetic field) must be taken into account. In the simplest experiments, such as in Correlation Spectroscopy (COSY), the off-resonance effect simply results in a relatively harmless phase error that can be easily corrected by appropriate data processing.

However, in more sophisticated experiments, errors in successive pulses often have cumulative effects that can lead to complete failure of the desired coherence transfer process. Thus,
In the multi-quantum NMR quoted above, the coherence is determined by three monochromatic radio frequency pulses (π / 2-τ-π-τ-
It is excited by a sandwich of π / 2). If the effective magnetic field tilts, this sandwich can fail altogether. This is because Levitt and Ernst, who have proposed another series incorporating composite pulses.
and Ernst) in Mol. Phys. 50 , 1109 (1983). These series are based on the idea that some consecutive rotations can be designed to have better overall behavior than a single rotation.
The constraint of composite pulses is that they are typically RF only over a limited frequency window.
The point is that the performance is sufficiently exerted only over an offset range comparable to the amplitude.

In the following, a pulse that switches the magnetization from the initial state aligned with the magnetic field to a new state in which the magnetization is perpendicular to the magnetic field is called a "π / 2 pulse", while the transverse magnetization is
A pulse that is converted into transverse magnetization in a manner that causes a spin echo will be called a "π pulse".

Let's focus on two-quantum spectroscopy of weakly coupled pairs of spin-1 / 2 nuclei, such as Levitt and Ernst. In terms of spectroscopic recognition of molecular structure, one of the most promising experiments of this kind is J. Am. Chem. Soc.
102 , 4849 (1980) in Avax et al. (A. Bax et
al.) and is known under the name INADEQUATE. This method makes it possible to identify the total carbon skeleton of an organic molecule simply by examining the signal in the two-dimensional spectrum. There is no doubt that this method will gain greater popularity in the absence of difficulties in obtaining reliable excitation of biquantum coherence over arbitrary spectral widths.

The main object of the present invention is therefore to provide a method of the kind mentioned at the outset for efficiently exciting multiquantum coherence over a wide offset range with relatively small amplitude RF pulses.

According to one aspect of the invention, the object is that the three RF pulses of the basic sandwich are chirp pulses each having a duration t _P , during which the excitation
The frequency of the RF field is from the lower limit frequency ω _RFmin to the upper limit frequency ω
A chirp pulse swept monotonically with respect to _RFmax or vice versa, having a duration t _P / 2 in the basic sandwich and an amplitude of 2 to 4 times the amplitude of the first π / 2 pulse The fourth chirp pulse is achieved by the following method.

"Jounal of Magnetic resonance" Vol. 84, No. 1,
In the summer of 1989, pp. 191-197, co-inventors of the present invention, Gebodenhausen and Yacht-M. Beren, entitled "Reimaging with Chirp Pulses for Broadband Excitation without Phase Dispersion" with M. Ray. Published the article. This publication describes a method of exciting and observing single quantum coherence using chirp pulses. However, in this paper no method for exciting two-quantum or multi-quantum coherence is discussed, and the published article does not disclose anything similar to the correlation of a pair of nuclei, only the magnetization of one nuclide. It only discloses the excitation and observation of.

The multi-quantum coherence is generated by the action of three RF pulses of the basic sandwich on the spin system, but the fourth chirp pulse that follows is the phase of the multi-quantum coherence, which is almost offset-independent over a large phase bandwidth. Reimage as A series of chirp pulses can be viewed as an alternative to composite pulses that overcome the effects of large offsets and tilted effective magnetic fields. The chirp pulse sequence of the present invention satisfies the condition for re-imaging the magnetization vector, that is,
The phase distortions that should otherwise be taken into account can be kept negligible over a very large bandwidth, and thus the present invention combines the advantages of pulse spectroscopy techniques with those of CW spectroscopy. . Moreover, by using chirp pulses instead of monochromatic hard pulses, the pulse amplitude required for excitation of multiquantum coherence is significantly reduced, which is in-vivo for the method of the invention.
In the field of application, for example in NMR imaging or NMR-in-vivo-spectroscopy,
It has a particularly great advantage.

In a preferred embodiment of the method of the present invention, the basic sandwich of three RF pulses is immediately followed by said fourth re-imaging chirp pulse, thus minimizing the relaxation time, which is a multiquantum due to relaxation. It means less loss of coherence and therefore greater sensitivity of the method.

Although the method of the present invention is not very sensitive to the amplitude of the first chirp π pulse, in most practical cases the amplitude of the first π pulse will be the amplitude of the first π / 2 pulse. 2 to 4 times, preferably about 2.8 times, is covered. The amplitude of the second π / 2 pulse in the basic sandwich should be approximately the same as the amplitude of the first π / 2 pulse, and the amplitude of the fourth chirp pulse is the first π / 2 pulse in the basic sandwich. About the pulse amplitude of 3.
Should be tripled.

Another object of the present invention is to recover the observable single quantum coherence (eg, of spins A and M in a two-spin system) from the unobservable multiquantum coherence. This means, according to another aspect of the invention, that the second sequence of RF pulses starts after a phase-dispersing time interval τ ₁ following a multi-quantum echo generated by the first sequence of RF pulses. And the second sequence of RF pulses is a chirp re-imaging π with a duration t _P / 2 in which the frequency of the RF field is swept monotonically with respect to time from the lower limit frequency ω _RFMIN to the upper limit frequency ω _RFMAX. Pulse and duration t _P
And a subsequent chirp π / 2 pulse of

In a preferred embodiment of the method of the invention, the second sequence of RF pulses is further chirp π with a duration τ _π = t _P / 2 applied after a delay τ ₂ following the chirp π / 2 pulse. A pulse, and the resulting spin echo is detected after a final delay time τ ₃ following the other chirp π pulse. This allows for the recovery of the pure phase spectrum, i.e. pure absorption or pure dispersion, in the ω ₂ dimension.

For the excitation of two quantum coherences between spin A and spin M, the delay time τ ₂ can be chosen to be τ ₂ 0 and the final delay time τ ₃ to be τ ₃ τ _π. it can. As a result, a tablet having an antiphase structure is obtained in the two-dimensional spectrum. If the delay time τ ₂ is chosen to be τ ₂ (2J _AM ) ^-1 , and the final delay time τ ₃ is chosen to be τ ₂ + τ _π + τ ₃ J _AM ^-1 , J _AM spin A and spin If we express the scalar coupling constant of the coupling with M, the amplitude of the signal as a function of the offset will be smoother, ie the unwanted frequency dependence of the signal will be smaller.

In order to obtain a tablet having an in-phase structure in a two-dimensional spectrum, the time delay τ ₂ is τ ₂ (4J _AM ) ^-1
And the final delay τ ₃ is τ ₂ + τ _π + τ ₃
It should be selected to be (2J _AM ) ^-1 .

Recovering one observable quantum coherence from multiple quantum coherences can also be achieved by another aspect of the method according to the invention, in which three RF pulses of the basic sandwich are each sustained. A chirp pulse having a time t _P , for excitation during the duration
The frequency of the RF field is from the lower limit frequency ω _RFmin to the upper limit frequency ω
A chirp pulse swept in a monotonic relationship with respect to _RFmax and vice versa, and after a deployment time t ₁ , a sandwich exactly time-reversed into the basic sandwich, i.e. each having a duration t _P , each of which is respectively _Sweeps from the upper limit frequency ω _RFmax to the lower limit frequency ω _RFmin or vice versa with a time interval τ. Π / 2, π and again π /
Three chirp pulses of two kinds follow.

The final conversion of Iz generated by a propagator V to observable single quantum coherence is the chirp π / 2 pulse followed by half the duration of the preceding chirp π / 2 pulse and 2 It may be achieved by an echo sequence consisting of a chirp π pulse with an amplitude of ˜4 times, preferably 2.8 times.

Also, the amplitude of the first π pulse in the basic sandwich should be selected to be 2 to 4 times, preferably about 2.8 times the amplitude of the first π / 2 pulse, and the amplitude of the second π pulse in the basic sandwich should be Π / 2 pulse amplitude is the first π / 2
It should be approximately the same as the amplitude of the pulse.

One-dimensional "INAD by selecting very short development time t ₁₀
EQUATE ”spectrum is obtained.

In order to refocus the two-quantum coherence (DQC) and the zero-quantum coherence (ZQC) excited by the basic sandwich, a solid re-imaging π pulse is applied at the beginning or end of the finite expansion time interval t _1. Can be applied.

The re-imaging of multi-quantum coherence also occurs during the expansion time t ₁ .
It is achievable when the basic sandwich is followed by a fourth chirp pulse having a duration t _P / 2 and an amplitude of 2 to 4 times the amplitude of the first π / 2 pulse of the basic sandwich. In the most practical case, the timing of the pulses is chosen such that the basic sandwich is immediately followed by a fourth chirp pulse, the amplitude of which is the first in the basic sandwich.
It is selected to be about 3.3 times the amplitude of the π / 2 pulse of.

In a preferred embodiment of the method according to the invention, the second sequence of RF pulses for reconverting multi-quantum coherence into observable single quantum coherence comprises a chirp π / 2 pulse followed by the preceding chirp π / It consists of a chirp π pulse with half the duration of two pulses and an amplitude of 2-4 times, preferably about 2.8 times.

The sampling of the free induction decay signal following the second series of RF pulses may start immediately at the falling edge of this pulse series or, in some cases, preferred, but at the maximum intensity of the spin echo signal, the latter of which In some cases, the spectrum obtained by the Fourier transform of the echo signal is essentially free of phase dispersion.

In the usual case, the frequency sweep of the RF excitation radiation is performed at a constant sweep rate b and is linear with time. However, in special cases, the limiting frequencies ω _RFmin and ω _RFmax
It may be useful to perform a non-linear but monotonic frequency sweep between.

Since the present invention is useful for various NMR pulse spectroscopy applications, such as two-dimensional transform spectroscopy, homonuclear and heteronuclear correlation spectroscopy, it has the above spectroscopic capabilities, An NMR pulse spectrometer equipped with an enabling device is also considered the subject of the present invention.

Further details, aspects and advantages of the invention will emerge from the following description based on the drawing, in which FIG. 1 shows the time dependence of the RF frequency (top), the instantaneous RF phase (middle) , And a schematic nuclear magnetic resonance spectrum (bottom) is shown; FIG. 2 a) RF amplitude γ B ₁ and RF frequency ω _RF , b for a fundamental sandwich of three chirp pulses designed for broadband multiquantum excitation. ) Shows the time dependence of the magnitude of the antiphase term | 2I ^A _Z I ^M _XY |, and c) the magnitude of the two-quantum term | DQC |; Fig. 3 shows the initial phase of the third pulse in the basic sandwich. The magnitude of the two-quantum coherence (solid line) and the zero-quantum coherence (broken line) excited by the basic sandwich of FIG. 2 as a function of the phase φ ₃ is shown, a).
All parameters are the same as in FIG. 2 b) the effective flip angle β ₃ of the third pulse is reduced to π / 4; FIG. 4 shows spin A as a function of phase φ ₃ . (Solid line) and spin M (dashed line) show contributions to a) two-quantum coherence and b) zero-quantum coherence.
5 has a flip angle of β = π / 2; FIG. 5 is a graph of the contribution to the two-quantum coherence generated from spins A and M; FIG. 6 is the basic three-pulse chirp sandwich of FIG. The magnitudes of a) the two-quantum coherence | DQC | and b) the zero-quantum coherence | ZQC | are shown below; FIG. 7 shows β ₃ = by a) numerical simulation and b) calculation based on pulse / expansion model Figure 8 shows the magnitude of the two-quantum coherence | DQC | excited by a fundamental sandwich with π / 2; Fig. 8 shows the RF amplitude γB in the case of the excitation sandwich followed by a chirp reimaging pulse acting on the two-quantum coherence. The time dependence of ₁ (top) and RF frequency ω _RF (bottom) is shown; FIG. 9 shows a) magnitude | DQC | and b) phase Φ of two quantum coherences as a function of relative offset of two spins. Indicates In the figure, the case of the sequence of Figure 2 indicated by solid lines,
The case of the sequence of FIG. 8 is shown by a broken line; FIG. 10 shows a) re-transformation sequence of chirp pulse and b)
Corresponding | 2T ^A _XY I ^M _Z | shows the time dependence of the claim; FIG. 11, as a function of the offset Omega _A and Omega _M, reverse phase one a quantum terms) size caused by the re-conversion from the two quantum coherence | 2T ^A _XY I ^M _Z | and | 2I ^A _Z I ^M _XY | and b)
FIG. 12 shows the phases φ _A and φ _M ; FIG. 12 shows the time dependence of the RF pulse sequence for excitation and reconversion to the Zeeman-terms of two-quantum coherence using time reversal. Fig. 13 shows a) RF frequency ω _RF of the same RF pulse sequence as in Fig. 12, and corresponding b) antiphase one-quantum coherence 2 I ^A _Z I ^M _XY and c) magnitude of two-quantum term DQC. FIG. 14 shows the Zeeman terms I ^A _Z and I ^M _Z retransformed from the two quantum coherences as a function of offset; and FIG. 15 shows the coupling J _AM = 5 Hz and 2 is an experimental two-quantum spectrum of a proton spin system with an enlarged region showing two 2-spin subsystems with _{JA'M '} = 15 Hz.

Consider a scalar coupled two-spin system under a frequency modulated pulse. Represent the time-dependent Hamiltonian in a rotating coordinate system synchronized with the instantaneous frequency of the transmitter: Where Ω _A (t) = ω ₀ ^A −ω _RF (t) is the offset of spin A with respect to the time-dependent transmitter frequency. In this coordinate system, the RF phase φ, which is time-independent, can be shifted from one pulse to the next in order to select the coherence travel path by phase cycling. The RF amplitude γB ₁ is considered to be constant during the frequency sweep. Hamiltonian
Is time-independent for a short time interval τ, the Liouville-von Neuman equation dσ (t) / dt = −i [H (t), σ (t)] (2) t ′, t ′ + τ] to obtain σ (t ′ + τ) = exp {−iH (t ′) τ} σ (t ′) exp {iH (t ′) τ} (3) be able to.

In this way, the time expansion of the density operator σ can be calculated by stepwise conversion under the effect of the instantaneous Hamiltonian H (t ′). Since H (t ') contains non-commutative contributions from RF emission, chemical shifts and scalar couplings, each step is numerically diagonalized.
ation). Alternatively,
The conitative transformation of equation (3) is exp {-iH ₀ τ} exp {-iH _RF τ} (4a) Therefore, it may be decomposed into continuous rotations for different parts of the Hamiltonian.

For small τ values, terms with a second order of τ and higher orders are negligible.

exp {-i (H ₀ + H _RF ) τ} exp {-iH ₀ τ} exp {-iH _RF τ} (4b) Thus, for each interval τ, the effects of H ₀ and H _RF are continuously calculated. Good. Further, by introducing a hypothetical weak bond may decrease the scalar coupling term in equation (1) to 2πJI _Z ^A I _Z ^M. This allows the density operator in the product of angular momentum operators to be expanded. We performed the simulation based on both the Cartesian product method and the explicit matrix notation of the density operator. Typically about 2000 intervals are needed to describe the effect of a pulse covering a sweep range of 20 kHz in 16 ms.

As an alternative to the accelerated rotational coordinate system used in equation (1), a normal coordinate system that rotates at a constant reference frequency ω _ref may be considered. In this coordinate system, φ _RF (t) =
Since ∫ω _ref (t) dt, a linear frequency sweep means that the RF transmitter phase changes quadratically as a function of time. In this coordinate system it is possible to shift the transmitter initial phase φ _RF (0) at the beginning of the sweep in order to select the coherence path of travel.

The conventional rotating coordinate system is a natural reference system for the approximate description of chirp pulses and serves to guide intuition. This description states that the spin is RF only when it exactly passes through the resonance.
It is assumed to be affected by the place. As a result, we have, in effect, instantaneous semi-selective pulses separated by pure expansion intervals. For simplicity, we will call this picture the "pulse / expansion model". Since the J-coupling is always very small compared to the RF amplitude γB ₁ , it may always be assumed that both components of the tablet are affected at the same time and thus the expression “semi-selective pulse”. This is Ferretti and Ernst
Although they must be distinguished from the method described by, they used a cascade of truly selective pulses that affect one transition at a time in their description of rapid passage.

FIG. 1 shows the relationship between frequency and phase. In the case of a linear frequency sweep, the phase φ _A of the semi-selective pulse acting on the spin A is secondarily in resonance at the passing instant t _A , as shown in FIG. Or equivalently to offset, Be associated. The sweep rate b is the ratio of the sweep range and the duration of the pulse, b = Δω _P / t _P , and ra
Expressed in d seconds ^-2 .

During a chirp pulse with a linear frequency sweep, the resonances of spins A and B are such that the RF frequencies (top) coincide with the chemical shifts Ω _A and Ω _B and the instantaneous RF phase (middle) is φ _A and φ.
_B is passed at times t _A and t _M.

In a system with isolated spins I = 1/2, the pulse / expansion model is quite sufficient to predict the formation of echoes by a sequence of two chirp pulses. However, in coupled spin systems, this model only gives a qualitative diagram. In the following, Hamiltonian stepwise diagonalization or operator product transposition of the analysis results obtained from the pulse / expansion description
It will be compared with the exact simulation obtained by the fomations).

The basic sandwich consisting of the three chirp pulses of Figure 2a is built on the same principle as the (π / 2-τ-π-τ-π / 2) sandwich commonly used to generate multiquantum coherence. All three chirp pulses must have the same duration and therefore the same sweep rate. Its amplitude must be adjusted to achieve either π / 2 pulse or π pulse analog. Chirp π / 2 pulse is relatively weak
Characterized by the RF amplitude, so the adiabatic condition is deliberately violated so that the magnetization does not remain locked in the effective magnetic field throughout the sweep. The chirp π-pulse, on the other hand, has an RF amplitude sufficient to allow adiabatic reversal or reimaging of the transverse magnetization as in this case. In adiabatic reversal, the magnetization remains locked along with the effective field, while due to reimaging, the magnetization remains perpendicular to the effective field during the sweep. In FIG. 2a, the first chirp π
The / 2 pulse causes transverse magnetization starting from thermal equilibrium, the second pulse reimages the effect of the offset (chemical shift) but not that of the nuclide scalar coupling, and the third pulse Converts antiphase-one-quantum magnetization into multiquantum coherence. Relative phase must be optimized for maximum creation of two-quantum or zero-quantum coherence. As in conventional sandwiches, the spacing between the beginning of the first and third pulses must be on the order of (2J _AM ) ^-1 .

2b and 2c show the time dependence of antiphase one-quantum term and two-quantum coherence in a two-spin system. The phase behavior is quite complex as it refers to the accelerated rotating coordinate system,
So we prefer to show the evolution of the sizes | 2I ^A _Z I ^M _XY | and | DQC | (see definition in Table 1). The antiphase term is
Envelope proportional to sin (πJ _AM t) which increases spontaneously under the influence of scalar coupling. During the third pulse of the sandwich, the antiphase term is partially converted to two-quantum coherence. Figure 2 shows how the | DQC | term begins with | 2I ^A _Z
2 due to contributions from I ^M _XY | and then | 2I ^A _XY I ^M _Z |
It is shown to increase in two successive stages. Whether positive interference of these two contributions is obtained is determined by the relative phase of the three pulses. The shift of the phase φ ₂ of the second pulse only induces a chain shift of 2φ ₂ of one quantum coherence in the reimaging period after the second pulse, so that the third pulse of the sandwich It can be shown that it is sufficient to consider the effect of the phase φ ₃ of.

The RF frequency matches the chemical shifts Ω _A / 2π = 7 kHz and Ω _M / 2π = 13 kHz at the time points indicated by the arrows. Simulation parameters: J _AM = 8Hz, t _P = 16ms,
τ = 15 ms (hence 2t _P +2 τ = 1 / 2J _AM ), sweep range 20 KHz, effective flip angles π / 2, π and π / 2 respectively
ΓB ₁ = 309Hz, 860Hz and 3 for 3 consecutive pulses with
It is 09Hz. The initial phase of the three pulses is φ ₁ = φ ₂ =
0 and φ ₃ = 120 °, the latter chosen for optimal excitation. Note the increase of the two quantum terms to almost the maximum amplitude in two successive steps. FIG. 3 shows that the optimal value of the two-quantum excitation corresponds to the phase φ ₃ 120 ° (± 180 °) if three pulsed RF amplitudes are selected as indicated above. In fact, the phase requirements and the amplitudes of the π and π / 2 pulses are interdependent. Various conditions can be found for sufficient excitation of multiquantum coherence.

In FIG. 3a, all parameters are the same as in FIG. 2, but in FIG. 3b the effective flip angle β ₃ of the third pulse has been reduced to π / 4. Note that in the case of β ₃ = π / 2, the zero quantum excitation is very inefficient, regardless of φ ₃ .

FIG. 4 shows how the two contributions originating from spins A and M add or cancel, depending on the phase φ ₃ and the effective flip angle β ₃ of the third pulse. With π / 2 pulses, the contributions to the zero quantum coherence always cancel each other out, as shown in Figure 3a. This is in direct analogy with the (π / 2−τ−π−τ−π / 2) series of hard pulses whose zero quantum coherence cannot be generated from antiphase magnetization by π / 2 pulses. However, if the RF amplitude is reduced and the effective flip angle of the third pulse is β ₃ = π / 4, the RF phase is φ ₃ =
In the vicinity of 0 or 180 °, equal amounts of two-quantum and zero-quantum coherence (see Fig. 3b) can be obtained. This behavior is again similar to classical experiments. The simple pulse / expansion model discussed above cannot predict the behavior of FIG. Because it would imply that the maximum two-quantum excitation occurs if all three pulses have the same phase. Direct comparison with accurate simulations is difficult.

Figure 5: U.Eggenber explaining the pulse / expansion model
ger and G. Bodenhausen in Angew. Chem. 102 , 392 (199
0) and Angew.Chem.Int.Ed.Engl. 29 , 374 (1990), using the conventions discussed in the publication, showing the "Cartesian product expansion graph". The two contributions to the two quantum coherences emanating from spins A and M are shown separately. The Zeeman order I ^A _Z of the spin A is first transformed into the in-phase one-quantum coherence I ^A _xy , which is the period 2τ.
Expanded for + 2t _P + t _AM and anti-phase one quantum coherence 2I ^A
It becomes _xy I ^M _Z , which is finally converted into two-quantum coherence. On the other hand, I ^M _Z is converted to I ^M _xy , which is period 2
It expands during τ + 2t _P −t _AM and becomes 2 I ^A _Z I ^M _xy before being converted into two quantum coherence. Theoretically, the two different periods 2τ + 2t _P ± t _AM shown in Fig. 5 are both (2J _AM ).
Should be approximately equal to ^-1 , which is clearly t _AM <<
(2J _AM ) Only ^-1 can be satisfied.

In FIG. 5, the resonant frequency is passed at the time indicated by the vertical dotted line. The circles above and below each graph correspond to spins A and M, respectively. The symbols Z, x / in these circles
y and l represent operators I _Z , I _x and I _y , and a unit operator, respectively.

In addition to the two main paths shown in FIG. 5, for a two-quantum coherence, if the coupling J _AM is large enough to cause some conversion from in-phase coherence to anti-phase coherence at time t _AM There may be an even smaller contribution. These passages have a t _AM (2
It is of utmost importance if it is comparable to J _AM ) ^-1 , where each chirp pulse taken in isolation can provide a significant amount of two-quantum coherence.

One-bond scalar coupling constants ¹ J _CC are usually all in the same range (about 40Hz), but carbon-13 (carbon-
The 13) spin chemical shifts may cover a very large range of frequencies, typically 200 ppm (30 kHz in a 600 MHz spectrometer with 150 MHz C ₁₃ -resonance frequency). These situations make it possible to limit the guidelines for the design of chirp pulse sequences and their performance testing by simulations and experiments.

If all three pulses in the excitation sandwich of FIG. 2 have the same duration, then the magnitude of the multi-quantum coherence that occurs will be offset independent over most of the sweep range. FIG. 6 shows the offset dependence of the zero quantum and the two quantum terms generated by the 3-pulse excitation sequence.

The effective flip angle of the third pulse is set to either β ₃ = π / 2 (solid line) or β ₃ = π / 4 (broken line).
The offset Ω _{A of} spin A is invariant (Ω _A / 2π = 5kH
z), Ω _M moves towards the right end of the sweep, ie
Ω _M / 2π varies from 5kHz to 20kHz. The chirp pulse is swept from 0 kHz to 20 kHz in 16 milliseconds as in FIG.

FIG. 7 compares the exact calculations to the approximate analytical predictions derived from the pulse / expansion model discussed above. We consider a two-quantum coherence excitation with the sequence of Figure 2 using three different sweep rates. The tendency for the profile to decrease with increasing relative offset is because the two intervals highlighted in Figure 5 cannot be optimized simultaneously for all pairs of spins. Unlike the raw pulse / expansion description in Figure 7b,
The numerical simulation in Fig. 7a shows the so-called "edge effect" resulting from the finite range of the frequency sweep of the RF pulse.
accurately predicts "effects)".

The offset Ω _{A of the} spin A is fixed at Ω _A / 2π = 10 kHz, but Ω _M / 2π changes from 10 kHz to 40 kHz.
Chirp pulse from 0kHz to 40kHz at different sweep speeds b
Is swept up to. The duration t _P of the chirp pulse is shown. The RF amplitude γB ₁ was adjusted in proportion to √b. The reduced performance at lower sweep speeds can be explained by the increase in period t _AM shown in FIG. Note that the edge effects are not explained by the pulse / expansion model.

It may seem surprising that the range of offsets over which the sequence is valid can be extended by increasing the sweep speed. In fact, for fast sweep, period t _AM = | Ω _A −Ω _M | / b is period 2τ + 2t _P (see FIG. 5).
Is implied to be reduced to. The shorter t _AM , the closer to the ideal case that both periods 2τ + 2t _P ± t _AM can approach (2J _AM ) ^-1 regardless of the offset. On the other hand, the edge effects, which tend to worsen with increasing sweep speed, do not constitute too serious a drawback for sequences with fast sweeps.

For simple adiabatic inversion,
The edge effect can be damped by smoothing the profile of the RF amplitude γB ₁ at the beginning and end of the sweep, for example by shaping it with half-sine functions. But from the simulation,
It can be concluded that this type of smoothing does not improve the offset dependence of the multiquantum excitation using the sequence of FIG. this is,
Probably because the efficiency of the sequence is sensitive not only to the relative RF phase but also to the relative amplitudes of the three pulses.

The magnitude of the multi-quantum coherence produced by the excitation sandwich of FIG. 2 is thus largely uniform for all offsets, but the phase is strongly frequency dependent.

According to the invention, the phase of multi-quantum coherence can be re-imaged by another chirp pulse.

FIG. 8 shows that a fourth chirp pulse with twice the sweep rate and half the duration of the preceding pulse was inserted immediately after the two quantum echoes were formed by the excitation sandwich after the reimaging delay time t _P. The sequence is shown. The phase of the two-quantum coherence at the time of the echo is largely independent of the offset of the two contained spins.

Phase of this pulse that determines the phase of two quantum echoes φ ₄
Can be freely selected. Relative RF of four pulses in Fig. 8
The amplitude γ B ₁ was optimized for the ratio 1: 2.8: 1: 3.3.

This sequence makes it possible in fact to obtain multi-quantum coherence with a uniform phase for all offsets. By the broken line in the lower part of FIG. 9, two quantum phases Φ = arct
It is confirmed that the offset dependence of an ({DQC} _y / {DQC} _x ) is almost flat.

In Fig. 9, offset Ω _A / 2π = 10kHz is 20kH.
While kept constant in the middle of the z-sweep range, Ω _M / 2π is increased from 10kHz to 20kHz towards the right edge. The solid line in FIG. 9 is immediately after the simple 3-pulse sandwich in FIG.
That is, the magnitude and the phase without re-imaging are shown, while the broken line shows the magnitude and the phase at the time of echo in the sequence having the two-quantum imaging pulse in FIG.

Once generated, re-imaged, and made expandable for the period t ₁ , the multi-quantum coherence cannot simply be converted into a single quantum coherence observable by a single π / 2 chirp pulse. From such a pulse, antiphase one quantum terms, it will offset dependent mixtures residual multiple-quantum coherence and longitudinal 2 spin order 2I ^A _Z I ^M _Z is the result obtained. In view of this difficulty, two methods seem to be practical for obtaining the in-phase one-quantum coherence at the beginning of the detection period t ₂ of the two-dimensional experiment. The first involves creating a specific phase distribution of multi-quantum coherence as a function of offset before conversion to antiphase magnetization. The second method relies on the concept of symmetric excitation and detection and is aligned so that the preparation and detection part of the multi-quantum experiment achieves an apparent time-reversal.

The sequence of FIG. 8 makes it possible to reimage the two quantum coherences that occur at different times during the third pulse, and, conversely, the two quantum coherences that precede the reconversion to one quantum coherence. Intentionally expand the phase (sp
read out) is also good. In this way, when the RF field sweeps through the resonances of the spins A and M and passes between the instantaneous RF phase at the time points t _A and t _M (see FIG. 1) and the phase of the two quantum coherence, You can be sure that you have a relationship. As shown in Figure 10, this is pure {DQ
The post-echo phase dispersion of the two-quantum coherence, which is assumed to consist of C} _x , and an effective flip angle π of duration t _P / 2,
That is, it can be achieved by applying a chirp re-imaging pulse that has a sweep rate twice that of a pulse of duration t _P. A subsequent chirp π / 2 pulse of duration t _P results in a conversion to (anti-phase) single quantum coherence. The initial RF phase of this pulse must be selected for optimal movement. The amplitude (rather than the phase) of the observable magnetization thus recovered is determined by the resonance frequency Ω.
It is uniform for all pairs of _A and Ω _M. Roughly stated, the whole process is the reverse of what happens between the third pulse and the echo of FIG.

In FIG. 10, when the second pulse passes through the resonance frequency, the relationship between the two quantum phases and the RF phase is converted to 2I ^A _XY (and to 2I ^A _Z I ^M _XY not shown). It is supposed to happen. These anti-phase one-quantum terms are then allowed to change into in-phase terms I ^A _XY and I ^M _XY (also not shown), which are re-imaged by the chirp re-imaging pulse and τ ₃ At the end of the period, an antiphase-one-quantum coherence echo is formed. All pulses are swept over 20kHz. The π and π / 2 pulses have durations of 8 and 16 ms and RF amplitudes of 1020 Hz and 309 Hz, respectively.
And the initial RF phase of the second pulse is shifted by 246 ° with respect to the first pulse. Offset is Ω _A = 8
kHz and Ω _M = 12 kHz. The delay times are τ ₁ = 8 msec, τ ₂ = 54.5 msec, τ ₃ = 62.5 msec, where τ ₃ = τ ₂ + τ _π . The latter two are τ ₂ + τ _π +
It was adjusted so that τ ₃ = 1 / J _AM , at which time J _AM = 8 Hz.

FIG. 11 shows the efficiency of the reconversion series of FIG. 10 for spin pairs with different offsets. For all spin pairs we start with the pure {DQC} _x at the time of the echo (see pulse sequence in Figure 10). The offset, in contrast, was varied from 0 kHz up to 10 kHz for Ω _A / 2π and 20 kHz down to 10 kHz for Ω _M / 2π, but all chirp pulses were swept from 0 to 20 kHz. The pulse duration, RF amplitude, RF phase and interval are the same as in FIG.

Thus, starting from two quantum coherences aligned along the X-axis for all pairs of offsets Ω _A and Ω _M , it is practically possible to obtain uniform magnitude antiphase single quantum terms over a wide offset range. Becomes However, recording the (anti-phase) signal immediately after the second pulse in the sequence of Figure 10 would result in a strong offset-dependent phase dispersion. So by experiment, in ω ₁ dimension, but not in ω ₂ dimension of the obtained frequency domain,
It would be possible to obtain a pure absorption line shape.

The second method of detection is the t ₁ populations.
It is based on the symmetric excitation and reconversion to and the idea that it is then transferable to the observable magnetization.

For example, if a spin system is prepared by a chain of transformations in a desired state consisting of multiple quantum coherences, under certain conditions it is possible to undo all of the previous transformations and restore the initial state of equilibrium magnetization.

According to this concept, we will relate the preparation and reconversion part of the sequence of FIG. 12 to the propagators U and V. By inserting a development period t ₁ between the preparation and reconversion part, a longitudinal Zeeman magnetization modulated as cos (Ω _A + Ω _M ) t ₁ is obtained. This may then be converted to a transverse magnetization observable by a hard π / 2 pulse or, preferably, an echo sequence consisting of a chirp π / 2 pulse immediately followed by a chirp π pulse of double sweep rate.

Each of the propagators U and V has a total duration τ of 3
Represents a sandwich of two chirp pulses. If the evolution period is t ₁ = 0, and if a nonselective hard π pulse is inserted between U and V (vertical arrow), the effect of V is to offset the two-quantum coherence generated by U by Ω. _To reconvert back to the Zeeman terms I ^A _Z and I ^M _Z of the two spins A and M over a wide range of _A and Ω _M.

The effect of the preparatory propagator U may be “reverted” (or “reversed in time”) by the action of the propagator V if the condition V = U ⁻¹ can be fulfilled. This corresponds to the requirement H _V = −H _{U for} the corresponding (Hermitian) Hamiltonian operator. These relationships can be easily understood by examining the expected value of the Z component of the magnetization I _Z = I ^A _Z + I ^M _Z immediately after reconversion in the two-dimensional experiment of FIG. 12 having the expansion time t ₁ . : <I _Z > = tr {I _Z [Vexp (−iH ₀ t ₁ ) UI _Z U ^-1 exp (iH ₀ t ₁ ) V ^-1 ]},
(5) In the formula, U = exp (−i∫H _U (t) dt, and H ₀ is the formula (1).
This is the Hamiltonian of free precession. If t ₁ = 0, then S = tr {(V ^-1 I _Z V) (UI _Z U ^-1 ) (6) is obtained, so if U ^-1 = V (ie if H _V = −H _U ), the maximum possible signal tr {I _Z ² } is obtained. Since the sign of a perfect Hamiltonian cannot simply be inverted, one has to explore how changing the amount of RF amplitude, phase and frequency etc. can have a similar effect as an exact time reversal. First, the offset Ω _A (t) = ω _A0 related to the time-dependent transmitter frequency
Consider the Hamiltonian corresponding to a chirp pulse acting on an isolated spin with −ω _RF (t): Based on FIG. 12, contrasting times τ−t ′ and τ +
The same offset and the same RF amplitude must be obtained at t ′, where τ corresponds to the end of the preparation period and the beginning of the reconversion period: Ω _A (τ−t ′) = Ω _A (τ + t ′). ) Or ω _RF (τ−t ′) =
ω _RF (τ + t ′) and γB ₁ (τ−t ′) = γB ₁ (τ + t ′) (8) This means that the sweep direction must be changed. Now, if we choose the phase so that the Hamiltonian H _V associated with the propagator V is the complex conjugate of H _U , that is, if U ^* = V ⁻¹ , then If, in the case of perfect time reversal _{^{tr {| U -1 I Z U}} | 2} signal tr instead of it can be shown that _{^{{(U -1 I Z U)}} 2} is obtained. For matrix elements, we have S =
Σ _ij ρ _ij ² and S = Σ _ij ρ _ij ρ _ij ^* are respectively obtained,
At this time, ρ _ij = (U ^-1 I _Z U) _ij . The two equations differ by a phase factor, and for the first one can only say that the maximum average signal strength over t ₁ will be achieved. If a precession motion associated with the so-called spin-inversion transition is obtained during the expansion interval,
In-phase signals can actually be obtained. Such a transition | i ＞ ＜ j |
Can be converted to its complex conjugate | j><i | by a π _x pulse, and if such a pulse is applied during expansion,
The exact time reversal equivalent is actually obtained. In the two-spin system, both the two-quantum coherence | αα><ββ | and the zero-quantum coherence | αβ><αβ | correspond to the spin inversion transition.

If now as in equation (1), the scalar coupling term H _J = 2
It must be noted that if we introduce πJ I _A · I _M , there is no obvious way to invert the sign of this term, except perhaps by using the select pulse we want to avoid. One possible solution to this problem is to match the pulse sequence to the time dependence resulting from the coupling term. This is
The end of the preparatory period can be achieved if the coupling coincides with the point when the in-phase magnetization has completely converted to anti-phase terms.
This condition is quite naturally satisfied because the 3-pulse sandwich sequence of FIG. 2 relies on creating the maximum antiphase term anyway. In the t ₁ interval, only two quantum and zero quantum coherences are obtained, which correspond to spin-inversion transitions and may therefore be re-imaged with a hard π pulse.

Of course, this π-pulse might otherwise appear unorthodox in a sequence consisting only of chirp pulses. As a rule, this pulse is transmitted by the propagator U
If the multi-quantum coherence generated under the action of has phase φ = 0 for all offsets, ie ρ _ij = (U ⁻¹ I _Z U) _ij is purely real You can omit it. For this purpose, in principle, the re-imaging sequences discussed above can be used. Because it produces two-quantum and zero-quantum coherence with a common phase for all offset pairs. However, specific values for the phase and amplitude of all pulses are needed to obtain multiple quantum coherence along the X axis.

FIG. 13 shows the time evolution of the magnitude of the antiphase one-quantum term and two-quantum coherence between the proposed time-reversed sequences, shown for the case of t ₁ = 0. The phases in the reconversion part are sign-changed with respect to those in the preparation period in order to satisfy the condition of equation (9).

A hard π-pulse (not shown) is applied during excitation and reconversion. The RF frequency of the chirp pulse is similar to that in FIG. 2, in the case of π / 2 and π pulse, the RF amplitude γB ₁ =
Swept up and down between 0kHz and 20kHz in 16 milliseconds at 309Hz and 860Hz. Resonance frequency is Ω _A / 2π = 9kHz and Ω _M / 2
It is fixed at π = 11 kHz and J _AM = 8 Hz. Note that almost all the initial Zeeman orders are converted to two-quantum coherence.

FIG. 14 shows that the longitudinal magnetization profile obtained at the end of the series is largely offset independent. Performance tends to improve with increasing sweep speed, as in FIG.

Zeeman term I ^A _Z retransformed from two-quantum coherence
And I ^M _Z is 0 kHz up to 20 kHz for Ω _A / 2π,
The case of Ω _M / 2π is shown as a function of offset varied symmetrically from 40 kHz down to 20 kHz. The solid lines are the RF amplitude γ for π / 2 pulse and π pulse, respectively.
It shows the response to a chirp pulse swept between 0-40 kHz in 8 ms with B ₁ = 618 Hz and 1720 Hz and phase φ ₃ = 122 °. The broken line indicates γB ₁ = 440Hz and 1220Hz and φ ₃
Corresponds to a 16 msec 40 kHz sweep at = 52 °.

The final conversion of I _Z generated by the propagator V into observable single quantum coherence is the chirp π / 2 pulse followed by half the duration of the chirp π / 2 pulse and preferably 2 to 4 times, preferably It may be achieved by an echo sequence consisting of a chirp π pulse with an amplitude of 2.8.

Experiments were performed on a Bruker AM-400 spectrometer equipped with a selective excitation device from Oxford Research Systems. The chirp pulse has a phase that is a quadratic function of time, that is, φ = 1 / 2b (t
-T ₀ ) ² (where t ₀ corresponds to the center of the pulse and b is the sweep speed). Aspect 3000
The files created by the Pascal program running on the computer were converted with the Bruker SHAPE package to create the data for the waveform memory.

Two-quantum experiments were conducted on the proton system. These experiments combine the reimaged two-quantum excitation of Figure 8 with a "hard" π / 2 monitoring pulse during echo. FIG. 15 shows two subsystems (su with two scalar-bound protons with J _AM = 5 Hz and J _{A'M '} = 15 Hz).
bsystems) (1RS, 2SR, 4RS, 5RS, 6RS) -exo- (2-nitrophenylthio) spiro [bicyclo [2.
2.2] shows the spectrum of octane-2,2'-oxirane] -5-endo-yl. Since these couplings are multiples of each other, the delay in the experiment can be optimized for both subsystems simultaneously. The chirp pulse has a duration t _P = 8 msec and a τ delay of 42 msec, so the total duration of the excitation sandwich is 100 msec (2
J _AM ) ^-1 3 (2J _{A'M '} ) ^-1 .

Due to storage capacity limitations, only 256 time intervals can be defined for each pulse, and the amount of phase increase is chosen so that the frequency sweep is limited to a ± 5kHz window on either side of the carrier frequency. It was For the first and last time intervals within each pulse, the amount of phase increase was 0.98 rad. Simulations were used to optimize RF amplitude and phase under these relatively rough digitized conditions. 4 in FIG.
The optimal amplitude γB ₁ of the two pulses is 309Hz and 860H, respectively.
It was found to be z, 309 Hz and 1020 Hz. Last two
The one pulse phase was shifted by 150 ° relative to the first two. In the experiments, the best results were to calibrate the amplitude γB ₁ of a single chirp pulse for the maximum occurrence of transverse magnetization, and proportionally to the optimized value, i.e. other values at 1: 2.8: 1: 3.3. Obtained by setting the pulse amplitude. By cycling the phase of the hard retransform pulse and the receiver, we were able to suppress the undesired signal due to zero and one quantum coherence.

The spectrum of FIG. 15 has a frequency ω ₁ = (Ω _A + Ω _M ).
And ω ₂ = Ω _A and Ω _M , show pairs of multiplets symmetrically arranged with respect to the skew two quantum diagonal ω ₁ = 2ω ₂ . The cross section reveals the antiphase up / down characteristics of the tablet in each of the two subsystems.

Selected cross sections taken parallel to the ω ₂ axis show the antiphase structure of the tablet. Spectra were obtained using a four-pulse chirp sequence with two-quantum reimaging in Figure 8, followed by a nonselective "hard" π / 2 pulse during reconversion to antiphase one-quantum coherence. . Spectral width is ω
10 Hz and ± 5 Hz at ₁ and ω ₂ , the data matrix is 512 × 4K before zero filling and 1K after that
It was × 1K. The chirp pulse was generated by phase modulation of the RF carrier.

The performance of the method according to the invention, namely the device enabling the excitation of the other quantum coherence by the first series of chirp RF pulses and the reconversion of the multiquantum coherence to the transverse magnetization by the second series of chirp RF pulses, is Offset-sensitive two-dimensional transform spectroscopy (“NOESY”), correlation spectroscopy (“COSY”), heteronuclear and homonuclear correlation spectroscopy, as well as NMR-imaging Experiment, especially N
It would be advantageous in MR-in-vivo spectroscopy. Such devices also belong to the subject matter of the invention and are intended to be covered by the appended claims.

The frequency-modulated chirp pulse sequence proposed by the present invention holds considerable promise for extending two-quantum spectroscopy to systems with a wide spectral range, such as occurs in the case of high-field carbon-13 spin pairs. It is understood that the method of the present invention is applicable to all types of nuclear magnetic systems.

Substantial advances can be made by systematically using frequency modulation. The chirp pulse sequence according to the present invention is
It is applicable to a wide variety of one-dimensional and two-dimensional experiments involving coherence transfer, systems with quadrupole coupling in anisotropic phases, and dipole coupling systems.

Table 1 Table 1: Definition of shorthand notation used for single and double quantum coherence magnitudes. The symbols refer to the (actual) expected value of the corresponding operator.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Behren Jean-Marc Lausanne Tse Her 1005 Marterei, Switzerland 34 (72) Inventor Burghart Irene Swiss Reneus Tse Her 1020 Ludleman 4 (56) References JP-A-50-110384 (JP, A) JP-B 2-61251 (JP, B2) JP-B 7-36007 (JP, B2)

Claims (9)

(57) [Claims] 1. A first π is applied to a nuclear spin system to which a magnetic field is applied.
/ 2 pulse, generated after the elapse of the defocusing time interval τ following the first π / 2 pulse and after the elapse of another time interval τ following the first π pulse Excitation of multiple quanta, especially two-quantum coherence, in an NMR pulse experiment by irradiating a first series of RF pulses comprising a fundamental sandwich consisting of a second π / 2 pulse In the method of reconverting coherence into longitudinal magnetization by irradiating the nuclear spin system with a second series of RF pulses, wherein three RF pulses of the basic sandwich each have a duration of a chirped pulse having a t _P, between said duration, chirp pulse der swept in a monotonous relation with respect to the lower limit frequency omega _RFmin upper limit frequency omega _RFmax or time vice versa , Fourth with 2-4 times the amplitude of the amplitude of the duration in the basic sandwich t _P / 2 and the first [pi / 2 pulses
A method characterized by being followed by a chirp pulse of. 2. The method of claim 1 wherein said sandwich is immediately followed by said fourth chirp pulse. 3. The amplitude of the first π pulse in the basic sandwich is 2 to 4 of the amplitude of the first π / 2 pulse.
2. The amplitude of the second π / 2 pulse of the basic sandwich is approximately double, preferably 2.8 times, and is substantially the same as the amplitude of the first π / 2 pulse. Alternatively, the method according to item 2. 4. The amplitude of the fourth chirp pulse is about 3.3 times the amplitude of the first π / 2 pulse in the basic sandwich, as claimed in any one of the preceding claims. The method described in the section. 5. A nuclear spin system subjected to a magnetic field of high magnetic field strength, which is generated after a lapse of a first π / 2 pulse and another time interval τ following the first π / 2 pulse. First
A first series of RF pulses comprising a basic sandwich consisting of a π pulse and a second π / 2 pulse generated after the elapse of another time interval τ following the first π pulse. In the NMR pulse experiment by irradiating the first series according to any one of the preceding paragraphs, and the excited multiquantum coherence is applied to the nuclear spin system. In the method of re-converting to longitudinal magnetization by irradiating a second series of RF pulses and inducing free induction decay, a phase dispersion time interval following spin echo generated by the first series of RF pulses. A second sequence of RF pulses is started after τ ₁ and the second sequence of RF pulses is RF
Field frequency is lower limit frequency ω _RFMIN to upper limit frequency ω _RFMAX
Comprising a chirp re-imaging π pulse having a duration t _P / 2 swept in a monotonic relationship with time and vice versa and a subsequent chirp π / 2 pulse of duration t _P. How to characterize. 6. The second series of RF pulses further comprises another chirp π pulse of duration τ _π = t _P / 2 applied after a delay time τ ₂ following the subsequent chirp π / 2 pulse. A method according to claim 5, characterized in that the resulting spin echo is detected after a final delay time τ ₃ following the another chirp π pulse. 7. A nuclear spin system subjected to a magnetic field of high magnetic field intensity, which is generated after a lapse of a first π / 2 pulse and a defocusing time interval τ following the first π / 2 pulse. Irradiating a first series of RF pulses comprising a basic sandwich consisting of one π pulse and a second π / 2 pulse generated after another time interval τ following the first π pulse. To excite multiquantum, especially two-quantum coherence in an NMR pulse experiment, and re-convert the excited multiquantum coherence into longitudinal magnetization by irradiating the nuclear spin system with a second series of RF pulses. ,
In the method for inducing free induction decay, the three RF pulses of the basic sandwich are chirp pulses each having a duration t _P , during which the lower limit frequency ω _RFmin to the upper limit frequency ω _RFmax or vice versa. a chirped pulse is swept in a monotonous relation with respect to time, and, after deployment time t _1, the basic sandwich, accurate time reversed sandwich, i.e. each have a duration t _P, by a distance τ respectively time together Pause,
A method characterized in that three chirp pulses of π / 2, π and again π / 2 are swept from the upper limit frequency ω _RFmax to the lower limit frequency ω _RFmin or vice versa. 8. The second sequence of RF pulses is a chirp π.
7. A half-pulse followed by a half-duration of the preceding chirp π / 2 pulse and a chirp π pulse having an amplitude of 2 to 4 times, preferably about 2.8 times. The method described in the section. 9. An NMR spectrometer, in particular an NMR spectrometer implementing the method according to any one of the preceding claims, which is used for exciting transverse magnetization in an NMR experiment. A frequency modulator designed to enable a fast frequency sweep of a field, the frequency swept excitation field comprising a sweep rate, an RF amplitude, and a sweep repeti
and a frequency modulator that is adjustable with respect to the carrier frequency for which the excitation field frequency range is defined by frequency modulation, and is further designed to allow adjustment of the phase shift of the excitation field. A phase controller, a phase shifter that allows adjustment of the phase shift between the input signal of the phase-sensitive receiver and a receiver reference signal that is held in a coherent phase relationship with respect to the RF carrier frequency, and for multiple quantum coherence excitation. Of the RF pulse for reconverting the excited multi-quantum coherence into transverse magnetization, in particular the second sequence of chirp pulses, and in particular the second sequence of chirp pulses An NMR spectrometer comprising a time control device adapted to start the collection of the induced decay signal.