Method for measuring NMR-data of a target sample in an NMR spectrometer and NMR spectrometer

Abstract

An NMR spectrometer includes a compensation system comprising at least one target sample coil and a lock sample coil positioned in a volume of interest between the main magnet poles of a main field magnet that generate a main magnetic field, at least one compensation coil for compensating a drift of the main magnetic field within the volume of interest, at least one target channel for generating RF-pulses with a target excitation frequency, and a lock data treatment system comprising a lock channel for generating RF-pulses with a lock excitation frequency, the lock data treatment system adapting a compensation current in the at least one compensation coil and correcting simultaneously the target frequency by applying a target frequency correction offset at the target channel. The spectrometer lock channel is improved, particularly for measurements where the lock coil and the target coil are positioned separately within the volume of interest.

Claims

1. A method for measuring NMR-data of a target sample in an NMR spectrometer comprising a main field magnet for generating a main magnetic field, wherein the main magnetic field shows main magnet variations, at least one compensation coil, at least one target sample coil, a lock sample coil, at least one target channel for generating RF-pulses with a target frequency and a lock data treatment system with a lock channel for generating RF-pulses with a lock excitation frequency and, the method comprising the following steps: i) providing a transfer function correlating a target resonance frequency of a resonance peak of the target sample and a lock resonance frequency of a resonance peak of a lock sample at a present compensation current I.sub.comp of the at least one compensation coil, coefficients of said transfer function being stored in the lock data treatment system; ii) measuring lock data FID.sub.lock by means of the lock sample coil; iii) determining an actual required compensation current I.sub.comp for compensating temporal changes of the main magnetic field, the compensation current I.sub.comp being determined by processing the lock data FID.sub.lock of step ii); iv) applying the compensation current I.sub.comp to the at least one compensation coil; v) determining a target frequency correction offset FCO for the required compensation current I.sub.comp by using the transfer function; and vi) acquiring NMR-data of the target sample by exciting nuclei of the target sample by applying an RF-pulse by means of the target sample coil with a target excitation frequency corrected by the target frequency correction offset FCO.

2. A method according to claim 1, wherein the transfer function is determined by acquiring a series of data points, each data point comprising lock data FID.sub.lock and target data FID.sub.tar, whereas for each data point, the compensation current I.sub.comp is modified and the lock resonance frequency and the target resonance frequency are recorded.

3. A method according to claim 2, wherein a range in which the compensation current I.sub.comp is modified corresponds to a range of the main magnet variations.

4. A method according to claim 2, wherein a difference Δf between the target resonance frequency and the lock resonance frequency at a specific compensation current I.sub.comp is determined for each data point.

5. A method according to claim 4, wherein the target resonance frequency and the lock resonance frequency are measured simultaneously.

6. A method according to claim 1, wherein the transfer function is a polynomial.

7. A method according to claim 1, wherein in step iii) the compensation current I.sub.comp is determined by comparing the lock resonance frequency determined from the lock data FID.sub.lock with a nominal lock resonance frequency.

8. An NMR spectrometer for carrying out the method of claim 1, the NMR spectrometer comprising a compensation system and a main field magnet with main magnet poles for generating a main magnetic field, the compensation system comprising: at least one target sample coil and a lock sample coil positioned in a volume of interest, the volume of interest being arranged between the main magnet poles, at least one compensation coil for compensating a drift of the main magnetic field within the volume of interest, at least one target channel for generating RF-pulses with a target excitation frequency, and a lock data treatment system comprising a lock channel for generating RF-pulses with a lock excitation frequency, wherein the lock data treatment system is configured to adapt a compensation current I.sub.comp in the at least one compensation coil and to correct simultaneously the target frequency by applying a target frequency correction offset FCO at the target channel.

9. An NMR spectrometer according to claim 8, wherein the target sample coil and the lock sample coil are arranged at different positions within the volume of interest.

10. An NMR spectrometer according to claim 8, wherein a communication pathway is provided between the lock data treatment system and each target channel for transferring the target frequency correction offset FCO from the lock data treatment system to the target channel.

11. An NMR spectrometer according to claim 10, wherein the lock channel, the target channel and the communication pathway are arranged on a common embedded logic device.

12. An NMR spectrometer according to claim 8, wherein the main magnet is a permanent magnet.

13. NMR spectrometer according to claim 8, wherein the at least one compensation coil is a pair of compensation coils.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The invention is shown in the drawings as follows.

(2) FIG. 1 shows a preferred assembly of the inventive NMR spectrometer.

(3) FIG. 2 shows an embodiment of a compensation system of the inventive NMR spectrometer with several target channels.

(4) FIG. 3 shows the basic steps of the inventive method including lock routine and correction of the target excitation frequency.

(5) FIG. 4 shows a boot up routine for determining a transfer function.

(6) FIG. 5 shows a diagram showing the resonance frequency difference of target sample and lock sample in dependence of compensation coil current intensity without target frequency correction. A transfer function approximation is shown.

(7) FIG. 6 shows a diagram showing the resonance frequency difference of target sample and lock sample in dependence of compensation coil current intensity with correction of the target excitation frequency.

DETAILED DESCRIPTION

(8) FIG. 1 shows an NMR spectrometer 1 according to the present invention. The NMR spectrometer 1 comprises a main magnet 2 comprising two permanent magnets for generating a main magnetic field 3. In the example shown in FIG. 1 a permanent magnet is used as main magnet 2, but coil based magnets such as Helmholtz can also be used for generating the main magnetic field 3. A volume of interest 4 is located in an area where the main magnetic field 3 is as homogeneous as possible, i.e. between main magnet poles 2a, 2b of the main magnet 2.

(9) The NMR spectrometer 1 further comprises a compensation system 5, which comprises a target sample coil 6, a target signal generator 18, a lock sample coil 7, compensation coils 8, a lock data treatment unit 16, and a transfer function unit 22 for storing coefficients of a transfer function TF. The two compensation coils 8 are positioned on either side of the target sample coil 6 and the lock sample coil 7 next to the coils of the main magnet 2.

(10) The lock sample coil 7 and target sample coil 6 are geometrically separated within the volume of interest (air gap between the main magnet poles 2a, 2b). The magnetic field lines of the main magnetic field 3, which are illustrated schematically, show a different magnetic field environment for the target sample coil 6 and the lock sample coil 7.

(11) Compensation system 5 is shown in detail in FIG. 2. The compensation system 5 comprises from 1 to N target channels with at least one sample coil 6 if the nuclei have a similar Larmor frequency or, otherwise, several sample coils 6-1, 6-N, lock sample coil 7, compensation coil 8, N target channels 11-1, 11-N and the lock data treatment system 9. The components of the lock data treatment system 9 (lock channel 12, lock signal generator 14, lock pulse sequencer 15 and lock data treatment unit 16) and the target channels 11-1, 11-N are arranged on a common embedded logic 10. The lock channel 12 and the target channels 11-1, 11-N are running separately on the embedded logic device 10 and are connected via pathways 13-1, 13-N.

(12) The lock data treatment system 9 is used for adjusting a power supply (not shown) for the compensation coil 8 by determining a compensation current I.sub.comp (control signal) as well as target frequency correction offsets FCO-1, FCO-N for target channel frequency correction. In the embodiment shown in FIG. 2 several target channels 11-1, 11-N are provided, in particular for different nuclei, e.g. .sup.1H, .sup.13C, .sup.15N, .sup.31P etc. The frequency correction adjustment works in the same manner for all nuclei.

(13) A radio frequency (RF) pulse is generated with the lock signal generator 14 of the lock channel 12 to excite the lock nucleus of a lock sample (not shown) via the lock sample coil 7. As lock sample a fully fluorinated sample with a single NMR resonance can be used and thus the lock is done on .sup.19F instead of .sup.2H. However, this invention is not limited to any of these two nuclei and can be set to any other active NMR nucleus.

(14) The lock nucleus is used for observing a change of the static main magnetic field 3. The sequencer 15 controls the sequence of the pulsing. The free induction decay (FID) signal (lock data FID.sub.lock) of the lock sample is received and processed by the lock data treatment unit 16 to determine the corresponding correction for the compensation current I.sub.comp required for the compensation magnetic field which is applied by the compensation coils 8.

(15) The compensation coils 8 will induce additional field inhomogeneities as function of the flowing compensation current I.sub.comp. In order to correct this induced inhomogeneity, for each target channel 11-1, 11-N a transfer function is calculated by mapping the effect of different compensation current values with the frequency difference Δf between the frequencies measured in the respective target channel 6-1, 6-N and the lock channel 7. For this, the compensation current I.sub.comp sent to the compensation coils 8 is retrieved and evaluated in the transfer functions, thereby obtaining the frequency correction offsets FCO-1, FCO-N. The frequency correction offsets FCO-1, FCO-N are to be transferred via pathways 13-1, 13-N from the lock data treatment system 9 to the target channels 11-1, 11-N. The frequency correction offsets FCO-1, FCO-N are added in real time to target channel frequencies f-1, f-N set by a user 17 in order to obtain a corrected target excitation frequency f.sub.corr1, f.sub.corrN for each target channel 6-1, 6-N. This corrected target excitation frequency f.sub.corr1, f.sub.corrN will be sent to a target signal generator 18-1, 18-N, thus correcting the frequency of the target emitter path in real time. A target pulse sequencer 19-1, 19-N will control the target channel pulse routine. When a target channel acquisition is requested, a pulse with the corrected frequency is generated by target signal generator 18-1, 18-N (Direct Digital Synthesis sub-unit) and sent to the target sample coil 6-1, 6-N.

(16) FIG. 3 shows the basic steps of the inventive method including main magnet field compensation and correction of the target excitation frequency (lock routine) for one target channel 11.

(17) In step i) a transfer function TF that describes the relationship between the compensation current and the frequency deviation between lock data FID.sub.lock and target data FID.sub.tar provided. The transfer function TF is a mathematical function, which correlates a target frequency correction offset FCO as a function of the compensation current I.sub.comp (or the resulting magnetic field offset), wherein the compensation current I.sub.comp is a function of the variations of the lock resonance frequency.

(18) For calculating the compensation current I.sub.comp, lock data FID.sub.lock is acquired by measuring FID data FID.sub.lock of a lock probe (step ii)). In step iii) lock data is processed and the compensation current I.sub.comp is determined. For this, the resonance frequency of the excited lock nucleus is determined by processing the lock data FID.sub.lock. The lock data treatment unit 16 receives the lock data FID.sub.lock from the lock sample coil 7 through an analogue digital converter 20 (ADC) (see FIG. 2). The NMR lock signal can be decimated through a cascade integrator-comb filter (not shown) and a phase correction can be performed. A fast Fourier transform (FFT) of the down-sampled (decimated) lock data is applied to determine its resonance frequency. The dispersion component (imaginary part) of the FFT at the 0 Hz value is compared with a reference value (expected resonance frequency at a nominal main magnetic field). This compared value is fed to a proportional-integral-derivate controller (not shown), which will determine the compensation current I.sub.comp to be used by a controlled power supply (not shown) that feeds the compensation coils 8 through a digital analog converter 21 (DAC). However, other frequency estimation methods can be used to find the aforementioned resonance frequency, these well-known methods are, for instance, time domain based and rely on the zero crossing or the cancellation of the imaginary part of the FID NMR signal. The compensation current I.sub.comp that is required to change the main magnetic field in order to measure the lock resonance frequency according to the reference value is calculated and applied to the compensation coils 8 in order to change the magnetic field such that the resonance frequency corresponds to the reference value (step iv).

(19) In step v) the frequency correction offset FCO is determined with the compensation current I.sub.comp and the transfer function coefficients. By adding the frequency correction offset FCO to the target channel frequency f set by the user 17, the corrected target excitation frequency f.sub.corr is obtained for acquiring NMR-data of the target sample in step vi).

(20) FIG. 4 shows a boot up routine for determining the transfer function TF required for frequency correction.

(21) In step a) after booting the spectrometer 1, or after a request by the user 17, the compensation current I.sub.comp is modified (ΔI.sub.comp). Then, an NMR signal acquisition of target data FID.sub.tar and lock data FID.sub.lock is carried out (step b). The resonance frequency deviation Δf of the lock data FID.sub.lock and the target data FID.sub.tar is mapped as a function of the compensation current I.sub.comp (step c)). In this particular setup, it is sufficient to sweep the compensation current between −20% up to 20% of the maximal current output for the compensation coil 8. However, the frequency sweep range can be extended or reduced, depending upon the NMR main magnet 2 characteristics in order to cover the complete magnetic field drift caused by its temperature stabilization. For highly stable permanent main magnets with a limited frequency oscillation over time, the sweep range for the compensation coil 8 can be reduced to +/−10% whereas for relatively instable setups the sweep range should cover up to +/−35%. The compensation coil setup is the preferred approach, since by sweeping the frequency in this fashion it is possible to cover the complete main magnetic field variations. Therefore, by sweeping the compensation current I.sub.comp in the compensation coil 8 and simultaneously measuring the corresponding NMR resonances of the lock nuclei and the target nuclei in the closed period possible give access to map the frequency deviations Δf between lock data and target data within the sweep limits. It is advantageous to acquire simultaneously the NMR data of the said lock and said target nuclei (lock data FID.sub.lock, and target data FID.sub.tar) in order to have a most accurate frequency offset correlation at a given time.

(22) The acquired NMR data FID.sub.lock, FID.sub.tar are stored and steps b) and c) are repeated until a predefined number # of data points are measured. The predefined number # of data points is a compromise between accuracy and efficiency. As a rule, the higher the number # of data points the more accurate is the calculated transfer function TF. Conversely, the time for mapping the frequency deviations Δf can be reduced by choosing fewer data points. The selection of the number # of data points strongly depends on the stability of the main magnet arrangement and it will differ from setup to setup. Hence, for a narrow sweep range, less data points for mapping are required and for a highly fluctuating magnet system, which requires a higher sweep range, more data points are needed to calculate a transfer function TF. In general, a range from 30 to 250 measured data points are sufficient to establish a transfer function.

(23) Once the predefined number # of data points is reached, the transfer function TF is calculated (step d)). The transfer function or its coefficients is/are stored and can then be used for frequency correction as described above.

(24) A special variant to determine the transfer function TF is described with respect to FIGS. 5 and 6.

(25) FIG. 5 shows the dependency of the frequency deviations of the compensation current I.sub.comp. The frequency differences Δf between the said lock and said target nuclei with respect to excitation frequency are calculated and then compared to each other. These frequencies are determined in ppm scale in order to eliminate the field dependency of the main magnetic field. In a perfect system without inhomogeneities, the frequency difference Δf would be constant independent of the compensation current I.sub.comp. However, current NMR systems in the described configuration show that there is a difference in the frequency deviations, as shown in FIG. 5.

(26) To determine the transfer function, the lock data FID.sub.lock are preferably retrieved through the acquisition embedded logic whereas the target data FID.sub.tar are acquired with a workstation (not shown in the figures). In the workstation, the spectral width (SW) of the Fourier transformed target data as well as its number of points (NP) for the target nucleus, are retrieved. Hence, the resolution or dwell time (DW) in Hertz of the Fourier transformed target data is:

(27) $\begin{array}{cc}\mathrm{DW}=\frac{\mathrm{SW}}{\mathrm{NP}}& \left(1\right)\end{array}$
For the lock data, its resolution or dwell time is determined with the acquisition frequency f.sub.acq and a decimation constant optimized in the used design by:

(28) $\begin{array}{cc}\mathrm{DW}=\frac{f_{acq}}{\mathrm{Decimation}*\mathrm{NP}}& \left(2\right)\end{array}$

(29) In order to compare the frequency variations of the target and lock nuclei, these frequencies are expressed in the absolute ppm scale. For a given magnetic field strength, the frequency of NMR active nuclei is scaled via the gyromagnetic ratios. The absolute frequencies of the target and lock nuclei are calculated by considering their resonance chemical shifts. The ratio of both emission frequencies is the following scaling factor K:

(30) $\begin{array}{cc}K=\frac{f_{\mathrm{lock}}}{f_{\mathrm{target}}}& \left(3\right)\end{array}$
Therefore, the correct target nucleus frequency can be calculated by the correction factor K and the corresponding frequency of the lock nucleus.

(31) Using the workflow depicted in FIG. 4, data are acquired and temporally stored for both target and lock nuclei. The frequency differences Δf are determined by mapping the magnetic field at the different sweeping selected values I.sub.comp. Each frequency is determined in the following way by performing a FFT and then a phase correction, and then the dispersion part of the FFT is used to establish the position of the zero crossing. Next, the target nucleus frequency is multiplied by the correction factor.

(32) As an example, in FIG. 5 the compensation-coil current I.sub.comp was set from −20% to 20% with a total of 64 sampling data points. The frequency difference curve is fitted using the mathematical operation that fits the best, in the given example a second order polynomial is used, but it is not limited to any specific function or polynomial order.

(33) This workflow may be repeated several times in order to get as many as possible coefficients, which then will be averaged. The mean values provide the most accurate attainable coefficients due to its robustness. As an example, the data plotted in FIG. 5 were obtained with the polynomial coefficients as given by:
f(x)=c(0)+c(1)*CS+c(2)*CS.sup.2+ . . . +c(m)*CS.sup.m
where c(i) (i=0, 1, 2, . . . , m) are the polynomial coefficients and CS is the control signal, which is the corresponding current percentage value that has been sent to the compensation coil via a controlled power supply.

(34) The workflow for this fitting uses the x-axis data (control signal) to construct a Vandermonde matrix V with n+1 columns (where n is the order of the polynomial coefficient) and m rows equal to the number points, resulting in the following linear system:

(35) $\left(\begin{array}{cccc}{x}_1^n& x_1^{n-1}& \mathrm{.Math.}& 1\\ x_2^n& x_2^{n-1}& \mathrm{.Math.}& 1\\ \mathrm{.Math.}& \mathrm{.Math.}& \ddots & \mathrm{.Math.}\\ x_m^n& x_m^{n-1}& \mathrm{.Math.}& 1\end{array}\right)\left(\begin{array}{c}C\left(1\right)\\ C\left(2\right)\\ \mathrm{.Math.}\\ C\left(m\right)\end{array}\right)=\left(\begin{array}{c}{y}_1\\ y_2\\ \mathrm{.Math.}\\ y_m\end{array}\right)$
Here γ.sub.i (i=0, 1, 2, . . . , m) is the frequency difference for the given control signal (current percentage). This linear equation is solved ensuring the least square error.

(36) After applying the found coefficients from the Vandermonde matrix, and in order to illustrate its correction effect, new NMR data FID.sub.lock, FID.sub.tar for both target and lock nuclei were acquired. The workflow is used once again to determine new frequency differences Δf. The results after the correction can be seen in FIG. 6. The frequency differences Δf have been considerably reduced proving that the combination of a compensation coil I.sub.comp and the frequency correction offsets FCO is of great help in order to compensate for frequency shifts due to magnetic field inhomogeneities.

(37) With the applied coefficients the traditional lock sequence is applied where the lock sample is measured periodically, lock data FID.sub.lock is treated and the compensation coil current I.sub.comp is updated. In addition, the target channel will be updated for each cycle through the frequency correction function TF, as shown in FIG. 3.

(38) The implementation of the inventive method for frequency compensation in benchtop magnets has been proved. This methodology includes novel developments at the hardware level along with the software implementations to minimize parasitic magnetic field gradient distributions. It also helps to reduce the complexity of the compensation coil design that is required in such setups. Due to the given geometry, the lock-coil and the target-coil are set at different positions requiring that both must be at the best achievable magnetic field homogeneity. This is a rather complex task; without the aid of the inventive frequency compensation methodology, it would be almost impossible to achieve. The development of this concept is most efficient if both the lock channel and the target channel are implemented in the same embedded logic. Therefore, the communication and control among the necessary sub-units that are needed becomes optimally feasible. It is important to take into consideration that more independent designs on the same embedded logic can be realized because the density of a FPGA as embedded logic allows these connections. This allows independent systems to interact with each other straightforward.

(39) Summarizing the invention, the value of the compensation current is used to compensate the frequency difference caused by the magnetic field inhomogeneities between the lock resonance frequency and that the target resonance frequency, i.e., in the lock channel, the corresponding control signal is used to drive a current source, which feeds the compensation coil, and one can add a frequency offset to consider the frequency difference of the target sample. The frequency offset is determined via a transfer function, the coefficients of which are provided with the lock data treatment system. Via a communication pathway, which connects the lock channel and the lock data treatment system the target frequency correction offset can be transferred from the lock channel to the corresponding target channel in order to correct an uncorrected frequency chosen by a user.