GB2284898A - Nuclear quadrupole resonance spectroscopy - Google Patents

Abstract

A method of detecting the presence of a given quadrupolar nucleus in a sample, the resonance frequency and spin-lattice relaxation time of the nucleus varying with a given environmental parameter, comprises determining a range of the resonance frequency which corresponds to a selected range of the parameter, applying excitation to the sample to excite nuclear quadrupole resonance, and detecting the resonance response signal, the excitation being such that a detectable resonance response signal can be excited over the resonance frequency range, and being such as to produce a constant flip angle of less than 45 degrees actual. <IMAGE>

Description

METHOD OF AND APPARATUS FOR DETECTION, AND METHOD OF CONFIGURING SUCH APPARATUS This invention relates to a method of and apparatus for detecting the presence of quadrupolar nuclei in a sample having at least one Nuclear Quadrupole Resonance (NQR) property (most particularly spin-lattice relaxation time, T1) which varies with a given environmental parameter (for instance, temperature). It also relates to a method of configuring such apparatus.

As an example, the invention has application to the detection in the field of 14N quadrupole resonance signals from the explosives TNT, RDX, HMX or PETN concealed in parcels or luggage or on the person, or deployed in explosive devices.

As another example, it has application to the detection of concealed drugs, such as heroin and cocaine, for instance at airports. If the substance is determined to be present, an alarm may be sounded. It is to be noted that, in the detection of the presence of explosives or drugs at airports and the like, the temperature of the sample would be unknown, and may in fact vary widely. There may even be a significant temperature gradient within the sample.

NQR measurements have the advantage that they do not require the sample to be placed in a strong magnetic field, and therefore do not require the large, expensive and sample-size limiting magnet structures which are needed for nuclear magnetic resonance (NMR) measurements.

A method of detecting the presence of quadrupolar nuclei in a sample having at least one NQR property which varies with a given environmental parameter is known from international Patent Application No. PCT/GB92/00580 (British Technology Group Ltd.), whose disclosure is incorporated herein by reference. The method comprises in broad terms applying excitation to the sample to excite NQR resonance and detecting the NQR response signal, the excitation being such as to excite an NQR response signal which is detectable during the detection step over a predetermined range of the environmental parameter.

In this fashion, the presence of tSe relevant nuclei can be detected to the required sensitivity of detection. By "sensitivity of detection" is meant the number of selected nuclei (or amount of material) that can be detected at a given level of probability in a fixed volume of sample over a given test time. This sensitivity may be improved by Fourier transformation of the accumulated signals to give an absorption spectrum, the area under which is then measured by integration between appropriate limits.

The invention was based on the discovery that the NQR properties of a sample, such as resonance frequency (v) and spin-lattice relaxation time (T1), may vary considerably with a given environmental parameter, such as temperature, pressure or magnetic field, and that it was therefore necessary to take account of or compensate for these variations when carrying out NQR tests to cover a predetermined range of the environmental parameter.

Since the NQR testing would usually be carried out in the field (such as at an airport), the given predetermined range of the environmental parameter might typically be the range of that parameter likely to be encountered in the field (possibly the national or global average range or maximum range of that parameter). If the parameter is temperature, the predetermined range might be + 10 C globally (possibly between 5"C and 3,5"C) or + 20"C (possibly between -10 C and +30"C). The range might be as much as from -30 C (corresponding to arctic conditions) to +40"C (corresponding to desert conditions). In some industrial applications, considerably wider temperature ranges may be encountered. If the environmental parameter is pressure, the predetermined range may be, for example, + 1% (corresponding to a typical average daily pressure range) or + 5% (corresponding to a maximum range), or even higher if the technique is being used to test conditions inside a pressure vessel.

Hence, if the substance under suspicion were the explosive RDX in airport luggage, it might be subjected to a temperature variation of -25 to +15"C about a mean of 20"C (i.e. a variation from -5 to +3S"C). Different samples might be at different temperatures, and/or the same sample might have a non-uniform temperature.

Considering now the NQR property spin-lattice relaxation time (T1), it was also discovered that the effects on the spin-lattice relaxation time (Fi) of temperature (and also pressure) variations could be very significant. For example, with the RDX sample described above, T1 for the v+ frequencies of the ring 14N varies, surprisingly, by a factor of approximately 8 between 5 and 25"C.

We further discovered that one particularly important effect of the variation of T1 is on the ratio t/Tl, where X is the pulse spacing between consecutive excitation pulse repetitions (the pulse repetition time). As an example, reference is made to Figure 1 of the present specification, which has been derived from Figure 1 of PCT/GB92/00580, and in which the variation of signal strength with flip angle (a) is shown for different values of zTr,. Both these figures have in turn been derived from equations derived by Vega (J. Chem. Phys., 1974, ,1099, Eq. IV-29) for spin-l nuclei which predict the strength of the steady-state NQR signal in a powder as a function of T for a given T,. These figures are similar to a figure which appears in a paper by Buess et al. ("NQR Detection Using a Meanderline Surface Coil", J. Mag.

Res., Vol. 92, 348-362 (1991)). It should be noted that the plots shown in these figures are completely general, in the sense that they are not specific to any particular substance.

Present Figure 1 demonstrates that, at a given ratio of t/T, ( < 5), the signal is predicted to pass through a maximum as the flip angle is varied, this maximum moving to flip angles less than 119 degrees (am) and to lower signal strengths as the ratio of t/T, drops below 1. Suppose, at x/T,=5, the maximum Free Induction Decay (f.i.d.), corresponding to a flip angle am, has a relative strength of unity; at TfT1=0.1, it has shifted to 0.3am and has a relative signal strength of 0.5. For constant X and at a constant flip angle am, a variation in l/Tl between 1 and 0.1 (an increase in T, by a factor of ten, corresponding, for v+ for the ring '4N in RDX, to a temperature increase near ambient of roughly o0 C) gives rise to an approximately 70% loss in response signal strength. Such a loss might render the response signal undetectable if it were overwhelmed by noise.

Hence the invention disclosed in PCT/GB92/00580 included a technique of compensating for the variation of Tl. This technique involved applying the excitation as a series of excitation pulses, at a plurality of different pulse repetition rates and/or so as to produce a plurality of different flip angles, so that a detectable NQR resonance response signal could be excited over the range of the relaxation time corresponding to the range of the environmental parameter, and indeed preferably so that the response signal had a substantially constant strength over that range.

Although this technique has been successful, in that it affords a highly sensitive test, in certain applications it has the disadvantage that it is unnecessarily complicated, in so far as it involves the varying either of flip angle or pulse repetition time or both during the course of the test. This can add to the complexity of the detection apparatus and can affect the duration of the test.

The present invention seeks to solve these problems.

The invention is based on the realisation, pursuant to the present invention, that, provided the pulse repetition time and flip angle are selected carefully, they may be maintained at values which are constant for the particular test, without significant detriment to the sensitivity of the technique.

According to the present invention, there is provided a method of detecting the presence of a given quadrupolar nucleus in a sample, the resonance frequency and spin-lattice relaxation time of the nucleus varying with a given environmental parameter, comprising determining a range of the resonance frequency which corresponds to a selected range of the parameter, applying excitation to the sample to excite nuclear quadrupole resonance, and detecting the resonance response signal, the excitation being such that a detectable resonance response signal can be excited over the resonance frequency range, and being such as to produce a constant flip angle of less than 45 actual.

As used herein, the term "constant" is not to be interpreted in a particularly precise sense; rather, it is to be interpreted in the context of the relatively low degree of precision usually associated with methods of detecting the presence of quadrupolar nuclei in a sample. For example, a 25% or even greater (for example 50%) variation in the flip angle over time would usually be acceptable.

Further, references to flip angles of a certain number of degrees "actual" are to be contrasted with values of "effective" flip angle, the two being related by a Bessel function in a manner which is well-known (see, for example, the Vega reference referred to above). For example, for a powdered I = 1 quadrupolar spin system, a flip angle of 119 actual produces a maximum free induction decay, and so corresponds to a flip angle of 90" in Nuclear Magnetic Resonance, a flip angle of 119 actual thus corresponding to one of 90" effective.

Also, although the excitation is such as to produce a constant flip angle, it will be understood that there may be circumstances in which the excitation produces flip angles which, whilst being constant (that is, not varying with time), may differ within the sample according to the the distance away from the excitation source.

Also, as used herein, the term "detectable" preferably connotes a signal/noise ratio which, within the measurement time of the detection step, is significantly above the actual or expected noise levels, the degree of significance being determined, for example, by standard statistical methods, such as the upper Student's t-distribution.

Also, as used herein, pulse repetition rate is simply the reciprocal of pulse repetition time.

For reasons explained more fully later, since the excitation is such that a detectable resonance response signal can be excited over the resonance frequency range, and since it is such as to produce a constant flip angle of less than 45 degrees actual, the technique of the present invention can detect relatively easily the presence of the relevant nuclei in circumstances where the given environmental parameter is unknown and may vary over a wide range, without there being a wide variation in the sensitivity of detection over the range of the environmental parameter. It can accomplish this by allowing in a simple fashion not only for the variation of the resonance frequency but also for the variation of the spin-lattice relaxation time with the given parameter.

Preferably, the excitation is pulsed excitation at a constant pulse repetition rate.

Preferably also, the method further includes selecting an expected extreme value of the environmental parameter, being that extreme value for which the relaxation time has its shortest value (as opposed to the opposite extreme value for which the relaxation time has its longest value), and determining a measure of the ratio of the pulse repetition time to the shortest value of the relaxation time, the pulse repetition rate being such that said measure of the ratio is no more than a selected value.

By operating at a constant pulse repetition rate and so as to produce a constant flip angle, the detection technique can be considerably simplified over that disclosed in International Patent Application No. PCI/GB92/00580. The purpose of the selection of an expected extreme value of the environmental parameter (such as the highest expected value of temperature), the determination of said measure of the ratio, and ensuring that the pulse repetition rate is such that said measure is no more than a selected value is that, by judicious choice of the selected value, the detection technique can be relatively rapid whilst at the same time not being overly susceptible to variations in the environmental parameter.

Hence, in a closely related aspect of the invention, there is provided a method of configuring apparatus for detecting the presence of a given quadrupolar nucleus in a sample, the spin-lattice relaxation time of the nucleus varying with a given environmental parameter, the apparatus including means for applying pulsed excitation to the sample to excite nuclear quadrupolar resonance, the method including selecting an expected extreme value of the environmental parameter, being that extreme value for which the relaxation time has its shortest value, arranging the excitation means to apply excitation at a constant pulse repetition rate and such as to produce a constant flip angle, and determining a measure of the ratio of the pulse repetition time to the shortest value of the relaxation time, the pulse repetition time being such that said measure of the ratio is no more than a selected value.

The invention extends to a method as directly aforesaid, for detecting the presence of the given nucleus in the sample, the apparatus further including means for detecting the resonance response signal, and including the step of operating the excitation and detection means to detect the presence of such nucleus.

The invention further extends to a method of detecting the presence of a given quadrupolar nucleus in a sample, the spin-lattice relaxation time of the nucleus varying with a given environmental parameter, including selecting an expected extreme value of the environmental parameter, being that extreme value for which the relaxation time has its shortest value, applying pulsed excitation to the sample to excite nuclear quadrupole resonance, and detecting the resonance response signal, wherein the excitation is at a constant pulse repetition rate and such as to produce a constant flip angle, a measure of the ratio of the pulse repetition time to the shortest value of the relaxation time is determined, and the pulse repetition time is such that said measure of the ratio is no more than a selected value.

Preferably, the value of said ratio (of the pulse repetition time to the shortest value of the relaxation time) is no more than 20, more preferably no more than 10, yet more preferably no more than 3 or 5 or even 1. The particular importance of values of this order of magnitude is that it has been found that, all other parameters being the same, signal strength and hence sensitivity does not increase markedly at higher values.

Thus tests conducted at higher values would be unnecessarily long. Therefore, for a rapid test, it is usually advantageous to utilise a value of this order of magnitude or lower. Even more preferably, such selected ratio value is no more than 0.2 or 0.3. If this is the case, theoretical analysis provided later indicates that, for sufficiently small flip angle, signal strength becomes more or less independent of the environmental parameter.

Preferably, the method includes further selecting an expected extreme value of the environmental parameter for which the relaxation time has its longest value, the ratio of, on the one hand, the repetition time, to, on the other hand, the relaxation time corresponding to a value of the environmental parameter centred between its expected extremes, being in the range 0.05 to 5, preferably 0.1 to 1. This is a further feature which can ensure a relatively rapid detection technique which is at the same time not overly susceptible to variations in the environmental parameter.

It is preferred that an alarm signal is produced from the response signal in dependence on whether a predetermined threshold of detection has been exceeded. In this case, the method preferably further includes selecting an expected extreme value of the environmental parameter for which the relaxation time has its longest value, determining a measure of the ratio of the pulse repetition time to the longest value of the relaxation time, and producing an alarm signal from the response signal in dependence on whether a predetermined threshold of detection has been exceeded, the threshold corresponding to a value of response signal strength which is no more than a value corresponding to said measure of the ratio of the pulse repetition time to such longest relaxation time. This can ensure that a detectable resonance response signal is produced over the entire selected range of the environmental parameter, that is, between its two extreme values.

Preferably the method includes further selecting an expected extreme value of the environmental parameter for which the relaxation time has its longest value, and determining a measure of the ratio of the pulse repetition time to the longest value of the relaxation time, the value of signal strength corresponding to the value of the ratio of the repetition time to the shortest relaxation time being less than a selected number of times, preferably less than five times, more preferably less than three times, even more preferably less than twice, the equivalent signal strength value for the longest relaxation time. By ensuring that the variation of signal strength over the range of the environmental parameter is maintained within acceptable limits, the sensitivity of detection can also be maintained within acceptable limits over this range.

In the preferred embodiment, given the selected range of the environmental parameter and a desired range of the response signal strength, the values of pulse repetition rate and flip angle are chosen by a process of iteration between these two quantities.

Preferably, the excitation is such as to produce a flip angle of less than 45" actual.

More preferably, the excitation is such as to produce a flip angle of less than 35 , even more preferably less than 25 , more preferably still less than 15 , yet more preferably less than 10 , yet more preferably still less than 5" actual. In fact, as will be explained later, the smaller the value of flip angle, the more effective the technique.

For simplicity, the excitation is preferably applied at a single excitation carrier frequency, although it may be applied at two or more such frequencies.

A particularly preferred feature is the use of so-called "shaped" pulses, as described in detail later, as well as in United Kingdom Patent Application No. 9319875 and International Patent Application No. PCT/GB94/02070. These can enable a broad frequency bandwidth to be excited at low power levels and using only a single excitation frequency. In broad terms, a shaped pulse may be thought of as one for which the time domain waveform of the excitation is either phase modulated or amplitude modulated, or both phase modulated and amplitude modulated.

As disclosed in United Kingdom Patent Application No. 9319875, apparatus for Nuclear Quadrupole Resonance testing a sample comprises means for applying excitation to the sample over a selected excitation frequency range to excite NQR resonance, and means for detecting the NQR response signal, the phase of the excitation varying generally non-linearly with the excitation frequency over the selected range.

Suitably, the excitation is pulsed excitation at a carrier frequency v0 and the phase of the excitation varies generally non-linearly with the frequency off-set (Av=v-v0).

The shaped pulses invention arises from the surprising discovery in the field of Nuclear Quadrupole Resonance that by applying excitation whose phase varies nonlinearly with the excitation frequency, the peak power of the excitation can be reduced very significantly. In one example, peak power has been reduced by more than an order of magnitude. A very significant reduction in peak radio-frequency magnetic field has also been achieved.

It is known from a paper entitled "Generation of optimal excitation pulses for two energy level systems using an inverse Fourier transform method" by Guan, S. (J.

Chem. Phys., Vol. 96, No. 11, 1992, pp. 7959ff.) to use excitation, the phase of which varies non-linearly with the excitation frequency, in the different technical field of Nuclear Magnetic Resonance (NMR). Such excitation is employed to solve a different problem, namely that of NMR Solvent suppression. The excitation yields a notched spectrum in the frequency domain. The above paper follows on from an earlier paper by Guan, S. entitled "General phase modulation method for stored waveform inverse Fourier transform excitation for Fourier transform ion cyclotron mass spectrometry", J. Chem. Phys., Vol. 91, No. 2, 1989, pp. 775ff. (see also Section 4.2.3 of the book entitled "Fourier Transforms in NMR, Optical, and Mass Spectrometry, A Users' Handbook", by Marshall, A.G. and Verdun, F.R., published by Elsevier, 1990).

However, the technique proposed by Guan for using such excitation in NMR testing would not be expected to be applicable to NQR testing, because it relies on an assumption that the substance being tested is a two-level system which follows the Bloch equations. The Bloch equations, it should be noted, are applicable to NMR in liquids, ion-cyclotron resonance, FT mass spectroscopy and electronic laser spectroscopy. However, they are not applicable to substances which exhibit NQR effects. Hence Guan's technique would not be expected to work in NQR.

In fact, it is surprising that, and not fully understood why, the shaped pulses technique does actually work in NQR.

It is also to be noted that Guan's technique is deliberately limited to the time domain waveform of the excitation being only amplitude modulated, but not also otherwise phase modulated. (It will be understood that amplitude modulation by itself, when the amplitude varies between positive and negative values, involves modulation of the phase, say between 0 and 180 degrees.) Preferably, for the duration of the excitation the envelope of-the modulus of the time domain excitation waveform has no zero crossings. This can have the advantage of reducing the peak power (and radio-frequency magnetic field) of the excitation considerably.

In one preferred embodiment of shaped pulse, the excitation applying means is adapted to apply excitation the time domain waveform of which is both amplitude and phase modulated. As opposed to the case where the waveform is only amplitude modulated, this arrangement can be utilised to produce an excitation waveform whose envelope has no zero crossings.

In this embodiment, preferably the applying means includes means for producing two signals which are in relative quadrature, and means for combining the signals to form the amplitude and phase modulated excitation. Again, preferably the signal producing means includes means for generating a signal and for splitting the signal into two signals in relative quadrature. These two features can yield a simple way of putting the invention into practice.

It is to be noted that these features are known per se from a paper entitled "High Speed Phase and Amplitude Modulator", by Wittebort, R.J. et al. (J. Magn. Reson., Vol. 96, 1992, pp. 624ff.). The modulator described in this paper is designed to be operated at a carrier frequency of 30 MHz, far above the frequency range of, say, 0.5 to 6 or 8 MHz likely to be of interest in 14N NQR testing.

Two types of signal may be generated for subsequent splitting. The generating means may be adapted to generate a carrier signal at a fixed amplitude. Alternatively or additionally, the generating means may be adapted to generate a signal having a varying amplitude.

In an alternative preferred embodiment of shaped pulse, the applying means is preferably adapted to apply excitation the time domain waveform of which is amplitude modulated but not otherwise phase modulated. This embodiment may not permit such low peak excitation powers to be achieved as are achievable with other embodiments, but can be simpler to manufacture since there is no requirement to modulate the phase directly.

In a further preferred embodiment, the applying means is adapted to apply excitation the time domain waveform of which is modulated according to a chirp function. This is a particularly simple way of putting the invention into practice, since a chirp function is easy to implement. The chirp function may, for example, be a linear chirp function of the form cos(ot2); that is, the modulation is pure phase modulation, the phase being modulated linearly with respect to the frequency off-set.

In any embodiment of shaped pulse, preferably the excitation applying means includes means for generating a carrier signal and means for modulating the carrier signal to produce excitation the phase of which varies generally non-linearly with the excitation frequency over the selected excitation frequency range. This can be a particularly convenient way of putting the invention into practice. The modulating means suitably includes means for storing a representation of a modulating waveform for modulating the carrier signal.

The variation of the phase of the excitation with the excitation frequency may, for example, be symmetric or antisymmetric with respect to the carrier frequency v0. In either event, the phase variation is preferably quadratic (that is, proportional to (v to)2); this has been found to give rise to low peak excitation powers. Another possibility is to vary the phase with statistical noise, but this can result in a noisy spectrum. A further disadvantage may be that the time domain waveform has many zero crossings, so that time is not used efficiently.

If the phase variation is quadratic, then in one particularly preferred embodiment a plurality of excitation pulses are applied, in such a way as to generate at least one echo response signal. It may be advantageous to generate echo response signals, as opposed to free induction decay signals, especially in circumstances in which the spin-lattice relaxation time, Tl, of the substance being tested is relatively long.

This feature arises from the surprising discovery, made pursuant to the shaped pulses invention, that shaped pulses in which the phase variation is quadratic preserve the phase of the NQR response signal. Such phase preservation is necessary for the successful generation of echoes.

Preferably, the modulus of the frequency domain excitation spectrum is substantially constant over the selected excitation frequency range. Thus the excitation can be uniform over the frequency range.

The shaped pulses invention extends to a method of Nuclear Quadrupole Resonance testing a sample, comprising applying excitation to the sample over a selected frequency range to excite NQR resonance, and detecting the NQR response signal, the phase of the excitation varying generally non-linearly with the excitation frequency over the selected range.

A selected NQR resonance may be excited, and, if so, the duration of the excitation is preferably less than twice the free induction decay time, T2*, appropriate to that resonance. This is an important feature, which can ensure that unacceptable loss of the NQR response signal does not arise before the signal is detected. The duration is in fact preferably less than 100%, 75% or even 50% of T2*.

In a closely related aspect of the shaped pulses invention, there is provided apparatus for Nuclear Resonance testing a sample, comprising means for applying excitation to the sample over a selected excitation frequency range to excite nuclear resonance, and means for detecting the response signal, the applying means being adapted to apply excitation the time domain waveform of which is both amplitude and phase modulated.

In another closely related aspect, there is provided apparatus for Nuclear Quadrupole Resonance testing a sample, comprising means for applying excitation to the sample over a selected excitation frequency range to excite NQR resonance, and means for detecting the NQR response signal, the phase of the excitation varying generally nonlinearly with the excitation frequency over the selected range.

Analogous methods are also provided.

All the aforementioned features apply to these aspects of the shaped pulses invention.

In a further closely related aspect, the shaped pulses invention provides a method of extending the free induction decay time, T2*, of a sample, comprising applying excitation to the sample over a selected excitation frequency range to excite nuclear resonance, the phase of the excitation varying with the excitation frequency over the selected range.

The nuclear resonance is suitably Nuclear Quadrupole Resonance or Nuclear Magnetic Resonance. The variation of the phase of the excitation is preferably non-linear.

This aspect of the shaped pulses invention stems from the discovery pursuant to the present invention that if a nuclear signal is excited with a varying phase the length of the free induction decay may be greater compared with the case of simple excitation with constant phase. The reasons for this appear to be analogous to the reasons given below why the applied excitation itself is longer in such circumstances.

For a significant change in the f.i.d. time, T2t, to be observed, the applied excitation should have considerable phase variation within the bandwidth of the nuclear signal.

Hence an appropriate test of the technique is a broader nuclear resonance line (compared with the 5210 kHz v. line), such as the 3460 kHz (v NQR line of RDX with a bandwidth of 0.5 kHz.

The technique of this aspect of the shaped pulses invention can be particularly useful, for example, when T2* is less than T2, to obviate dead time problems due to probe ring down, or in circumstances where only small r.f. amplitudes and therefore small flip angles can be used.

Method features analogous to the various apparatus features of any aspect of the invention may also be provided, and vice versa.

Turning now away from the shaped pulses invention and considering other aspects of the present invention, preferably the environmental parameter is temperature. The expected upper extreme value may be less than 50 , preferably less than 40 or 30"C.

The expected lower extreme value (for which the relaxation time for most substances has its longest value) may be greater than -30 C, preferably greater than -20 or 10 C. It is expected that in most substances T1 will decrease as the temperature rises; in cases where the reverse is true, the two extrema will need to be interchanged.

Preferably the pulse repetition time is greater than, more preferably 2 or 3 times greater than, the free induction decay time of the quadrupolar nuclei, in order to allow time for the f.i.d. to be detected.

The present invention extends to apparatus for detecting the presence of a given quadrupolar nucleus in a sample using pulsed excitation, comprising storage means in which is stored information for at least two types of nucleus having different NQR properties on suitable values of flip angle and pulse repetition rate to take account of expected variations in temperature, means responsive to the storage means for applying, for a given type of nucleus, pulsed excitation to the sample to excite nuclear quadrupole resonance, the excitation means being adapted to apply excitation at a constant pulse repetition rate and such as to produce a constant flip angle suitable for the particular substance, and means for detecting the resonance response signal.

It is to be noted that the requisite pulse repetition rate and flip angle selected may be dependent on the particular substance under test, so that different pulse repetition rates and flip angles may be appropriate to different substances.

Apparatus features analogous to the method features described above may be provided.

For example, the apparatus may further include means for producing an alarm signal from the detected response signal in dependence on whether a predetermined threshold of detection has been exceeded. Again, the excitation may be applied at a single excitation frequency, and may take the form of shaped pulses.

The theory underlying the present invention, as well as preferred features of the present invention and examples of its operation, are now described, by way of example only, with reference to the accompanying drawings, in which: Figure 1 (prior art) is a plot of the steady-state NQR response signal strength against flip angle for differing values of the ratio t; Figures 2 (prior art) show a time domain rectangular excitation pulse (Figure 2(a and its frequency domain equivalent (Figure 2(b)); Figure 3 is a block diagram of a preferred embodiment of NQR testing apparatus according to the present invention; Figures 4(a) to 4(e) illustrate a first example of NQR test using shaped pulses; Figures 5(a) to 5(f) illustrate a second example of NQR test using shaped pulses; Figure 6 is a block diagram of a variant of the preferred embodiment of NQR testing apparatus; Figures 7(a) to 7(f) illustrate a third example of NQR test, again using shaped pulses; and Figure 8 is a pulse timing diagram exemplifying the use of the preferred embodiment of the invention.

In overview, the preferred embodiment of the present invention involves the use of pulsed radio-frequency excitation to excite Nuclear Quadrupolar Resonance in the substance whose presence it is desired to detect. Any suitable form of pulse sequence can be used, most usually a regular sequence of uniform pulses.

The manner in which the excitation frequency, pulse repetition rate, flip angle and other parameters concerned with the excitation are chosen is discussed first. Details of the apparatus for putting the invention into effect, together with details of a particularly preferred type of pulse (a "shaped" pulse), are given later. Finally, an example of the operation of the invention is given.

In the preferred embodiment, the invention makes allowance for variations both in resonance frequency and in spin-lattice relaxation time (T1) of the given quadrupolar nuclei. Other NQR properties may also be taken into account, as taught in PCT/GB92/00580.

Specifically, frequency is compensated for in the manner taught in PCT/GB92/00580.

In the present preferred embodiment, only one excitation frequency is employed, the bandwidth of the excitation being sufficient, at that single frequency, to cover the entire frequency range of interest, that is, the range corresponding to the temperature range of interest.

It has been found that "shaped" pulses are particularly suitable for producing the requisite bandwidth at minimum radio-frequency power, and hence these are used in the preferred embodiment, although other types of pulse may also be suitable. Shaped pulses are described in more detail later.

Concerning now the manner in which the variation of T, with temperature is taken into account, the overall aim is to achieve this by keeping the pulse repetition time, t, and flip angle, a, constant, and by judicious choice of these parameters so as to minimise the variation in signal strength over the temperature range of interest. This is in contrast to the disclosure of International Patent Application No.

PCT/GB92/00580, which teaches to vary one or other of these parameters in order to achieve an appropriate degree of compensation. This latter technique would be expected to produce the most sensitive tests, but the technique of the preferred embodiment of the present invention would be expected to be adequate for most practical purposes, is easier to implement, and can operate at a single excitation frequency.

Referring now particularly to Figure 1, in the preferred embodiment, T and a are constant for the duration of the particular detection test being carried out, even if that test is being conducted at more than one excitation frequency. Of course, flip angle may be different for a further test with the same sample, or for a separate test with a different sample. If more than one excitation frequency were employed, as an alternative to the excitation being such as to produce a single constant flip angle, it would be possible to employ a different (constant) flip angle and/or pulse repetition time for each frequency.

Flip angle is also constant over substantially the entire region of interest in the sample, although this is in general terms not necessary provided flip angle is kept sufficiently low.

Flip angle is rendered constant over the duration of the test by rendering constant the radio-frequency magnetic field Bd produced by the excitation applying means.

Further, it may be rendered constant across the sample by the use of suitable types of the excitation coils employed to apply the excitation.

There will thus be unique values of the ratio T/T1 for the two extreme values of temperature to which the sample is expected to be subjected. Since signal intensity does not increase significantly for values of this ratio above 5 or so, the value of the pulse repetition time, T, is selected so that, for the shortest value of T, (corresponding for most substances to the higher extreme value of temperature), this ratio is no more than a selected value, say, 5. If it were higher, the test would be unnecessarily long, without there being any concomitant improvement in signal intensity and hence sensitivity.

Then flip angle is selected at a sufficiently low value that the variation of signal intensity over the temperature range lies within acceptable limits. For example, the value of signal strength corresponding to the lowest temperature extreme may acceptably be less than five times, preferably less than three times, more preferably less than twice, the equivalent signal strength value for the highest temperature extreme.

An examination of Figure 1 reveals that, at high values of flip angle, say, 119 actual, there is a wide variation of signal intensity with the ratio T/T1, and hence with temperature. At this value of flip angle, for example, signal intensity is roughly 7 times greater at a value of the ratio of 5 than it is at a value of the ratio of 0.1.

However, at lower values of flip angle, in regions of the plot where the signal intensity versus flip angle gradient is largely positive rather than negative, there is a relatively small variation of signal intensity with the ratio. For example, for a flip angle of 30 , signal intensity is less than 100% greater at a value of the ratio of 5 than it is at a value of the ratio of 0.1.

It will be understood that the above variations of signal intensity, taken as they are from the theoretical curves of Figure 1, may not be experienced in practice to the precise same extent. One major reason for this would be if the resonance frequency of the quadrupolar nuclei were to vary with temperature, and the excitation were not such as to excite all resonance frequencies corresponding to the temperature range of interest equally. However, within the tolerances experienced in this type of testing it would usually not matter whether the variation of signal intensity with the ratio X were determined from the plot of Figure 1 or by experiment.

Therefore, as stated above, the flip angle is chosen to be as small as possible. It may suitably be less than 45 , preferably less than 35 , more preferably less than 25", more preferably still less than 15", yet more preferably less than 10 , yet more preferably still less than 5 actual.

Finally, having selected values of pulse repetition time and flip angle, since both these parameters affect signal strength, it may be desirable iteratively to revise their values so as, for instance, to reduce the overall test time without prejudicing the sensitivity of the tests.

It will be appreciated that the pulse repetition time has an effect on the sensitivity of the tests in at least two different ways. Firstly, the longer the pulse repetition time (up to a certain limit), the greater is signal strength and hence sensitivity (see Figure 1).

Secondly, and on the other hand, the shorter the pulse repetition time, the greater the total number of pulses that can be generated for a given duration of test, and hence the greater the number of free induction decays that can be accumulated. Signal to noise ratio, and hence sensitivity, increases as the number of detected pulses increases.

It will also be appreciated that flip angle has a direct effect on the sensitivity of the tests, in that, within certain limits, the lower the flip angle, the lower the signal intensity. However, since a pulse at a lower flip angle is shorter than the equivalent pulse at a higher flip angle, more lower flip angle pulses can be utilised in a given test time; in other words, the pulse repetition time can be higher the shorter the pulse.

It will further be appreciated that care must be taken not to use an excessively high value of flip angle. For example, for t/T, = 0.1 an increase in flip angle above, say, 40 will actually reduce rather than increase signal intensity.

Hence, because of this complicated interdependence between the pulse repetition rate and flip angle, as stated above, the final values of pulse repetition rate and flip angle may need to be chosen by a process of iteration between these two quantities.

As can be appreciated from the foregoing and a consideration of Figure 1, a mid-point value of the ratio TiT1 (that is, a value of the ratio corresponding to a value of temperature centred between its expected extremes) is likely to be in the range 0.05 to 5, preferably 0.1 to 1. In the context, it is not important what technique is used for evaluating such a centred value of temperature. It will be appreciated that a determination of the value of this ratio can be a useful check that the pulse repetition rate and flip angle have been chosen well.

In the light of the general principles explained above, a typical process for configuring the detection apparatus would be as follows: 1) Select upper and lower expected extreme values of temperature to which it is expected the sample will be subjected.

2) Select an acceptable range of variation for signal intensity over the temperature range.

3) Determine the longest and shortest values of the spin-lattice relaxation time T1 corresponding respectively to the lower and upper (or possibly upper and lower) expected extreme values of temperature. This needs to be done experimentally for each given quadrupolar nucleus of interest.

4) Select a value of the pulse repetition time, T, such that the ratio IT, is less than, say, 5 for the shortest value of T,.

5) Determine the corresponding values of the ratio TiT1 for the longest and shortest values of T,.

6) Select a value of flip angle, a, say, 30 actual, ensuring that, for the particular type of pulse used and the selected value of flip angle, T is longer than the duration of the pulse and the ensuing free induction decay (f.i.d.). The f.i.d. duration is usually assumed to be at least 2 or 3 times, maybe five times, the value of the free induction decay time (T, ).

7) Determine, from the plot of Figure 1 of this present specification or the plot of Figure 3 of the Buess et al. paper mentioned previously, whether the variation of signal intensity over the range of the ratio t for the selected value of flip angle is within its acceptable range.

8) If it is not, or, indeed, if the variation is unnecessarily small, choose revised values of T and/or a and repeat steps 4 to 7 above, iterating between T and a until an acceptable solution is reached.

9) Check that, for that temperature extreme with the longer T1, within the given measurement time Tm, the number Tm/T and strength of f.i.d.'s accumulated is sufficient to ensure that a given quantity of the material to be detected generates an accumulated signal larger than a given threshold.

10) If it is not, choose revised values of T and/or a and repeat steps 4 to 8 above, iterating between T and a until an acceptable solution is reached.

For example, the accumulated signal may be increased simply by increasing flip angle a.

Alternatively, the accumulated signal may be increased by decreasing the pulse repetition time T to a new value '. More f.i.d.'s are now collected in the given measurement time Tm, and the signal to noise ratio is improved by the factor ( I) .

However, the signal strength of each individual f.i.d. is reduced, and, to compensate for this, flip angle a may need to be increased, such that, for example, the signal strength for each individual f.i.d. reverts to its original value before T was reduced.

11) Store the final values of T, a and Tm, for a number of quadrupolar nuclei in different substances having different NQR properties, in the detection apparatus.

Theoretical justification for the above approach is apparent from an analysis of Vega's equations (loc. cit.) for the steady-state signal S, produced by a train of pulses of pulse repetition time T and flip angle a:

S m is the equilibrium steady-state signal observed at long values of T, the an are coefficients which depend only on a, and, for all 14N signals of RDX, which exhibit exponential spin-lattice relaxation, -TiT1 F( ) = e Vega shows that, for a < 30 , a1 - 0. Hence, neglecting the effect of higher order terms in Equation (1), we find that for small flip angles (a) Sa/S: = 1 - a2{ 1 - (2TiT1) + (2#/T1)2- ...} (2) It is clear that, provided a is sufficiently small and 2#/1 1, the observed signal s: becomes independent of TiT1, and therefore of all variations in the environmental parameter which affect T1. If the condition that 2#/T1 < < 1 is established for the shortest observed value of Tl, it will be true for all longer values.

In practice, the condition 2#/T1 < < 1 may be satisfied if #/T1 is less than 0.2 or 0.3 without the signal strength having significant dependence on the environmental parameter. It can thus be seen that there may be, for sufficiently low a, a significant range of X for which signal strength does not vary substantially. Hence, for an even wider range of TiT1, the variation of signal strength may lie within acceptable limits.

It is to be noted in passing that Vega also shows that, for a < 15 , a2 in the above Equation (1) is approximately proportional to a.

The above analysis can be applied to step 10 in the procedure described above. In mathematical terms, starting with Equation 2 given above, valid for 2TiT1 1, decreasing # makes the term in a2 larger, which diminishes s"s; increasing a increases S&alpha;# and also a2, and so changes S t. Provided that a is less than 30 , preferably less than 200, the change in : dominates and the signal : increases.

The basic theory underlying the "shaped" pulses invention rcferred to earlier is now discussed. Referring to Figures 2, which show the known rectangular excitation pulse, this pulse appears in the time domain as the waveform in Figure 2(a), and is equivalent, in the frequency domain, to the spectrum shown in Figure 2(b). The spectrum of Figure 2(b) includes a sharp central peak and sidebands which diminish in magnitude away from the central peak. It will be appreciated that the spectrum in Figure 2(b) is the Fourier transform of the waveform in Figure 2(a); this latter waveform is the inverse Fourier transform of this spectrum. It will also be appreciated that Figure 2(a) actually shows only the envelope of the pulse (or, more precisely still, one half of the envelope of the pulse), in that the pulse comprises many radiofrequency oscillations.

It can be shown that, for a rectangular pulse of duration T, half-height bandwidth Av and peak pulse power P, T equals 0.6/Av and P is proportional to 1M, provided that (in NQR) the product BlfT, and therefore the flip angle, remains constant. Bd is defined as the amplitude of the oscillating radio frequency field. Hence the known rectangular pulse has the drawback that peak power increases strongly (quadratically, in fact) as bandwidth increases.

It appears from an analysis of the rectangular pulse that the high power requirement is attributable to two main factors. Firstly, the sidelobes apparent in the frequency domain consume substantial power without contributing to the useful part of the spectrum (the central peak). Secondly, the range of excitation frequencies contained in the pulse are in phase at time zero. Hence there is a high coherency between the various frequencies at and very close to time zero, but destructive interference at other times. Thus the pulse is of short duration and high peak power.

In NQR, it has now been found that a solution to the first of these problems is to use a pulse which is shaped so as to yield a near-rectangular shaped frequency spectrum (that is, the modulus of the frequency domain excitation is substantially constant over the selected excitation frequency range). Shaping of the pulse is achieved by inverse Fourier transform of the near-rectangular frequency spectrum to yield the required shape of the time-domain excitation waveform.

The solution to the second of these problems (high coherency near time zero), it has now been found, is to use excitation the phase of which varies generally non-linearly with the frequency, so that the phases in the time domain are scrambled. Hence there is constructive interference between the various frequencies over a significantly longer duration than is the case for the simple rectangular pulse and at the same time less constructive interference at time zero. Thus the excitation pulse can be of longer duration and lower power. In fact, for a quadratic phase variation it has been found that at constant flip angle, the power P for such a pulse is more nearly proportional to Av than Av2 as is the case for the rectangular excitation pulse. This conclusion follows from the relationship P 0: AV/tp or B, 0: (AvAp)iA where tp is the pulse length; Av and tp are not dependent (they are for the simple rectangular pulse), so keeping t, constant we get B, (at)% and P 0: Av. Power has also been found to be inversely proportional to the duration of the pulse for a given excitation bandwidth Av. The duration is in practice only limited by the duration of the free induction decay time T2* of the particular substance in question. This is discussed in more detail later.

The variation of phase with frequency should be non-linear in order to ensure a proper scrambling of the phases. A linear phase variation merely has the effect of producing a time-shift, but does not have the effect of scrambling phases. It has been found, in fact, that for NQR testing a quadratic phase variation may be optimal.

A preferred embodiment of NQR testing apparatus is now described with reference to Figure 3. This embodiment is particularly suited to the detection of the presence of a particular substance in a sample (such as a suitcase or the like). The apparatus comprises in general terms a control computer 100, means 102 for applying one, or more usually several, continuous radio-frequency excitation pulses to the sample covering a selected excitation frequency range and for a given duration, means 104 for generating shaped pulses for passing to the applying means 102, means 106 for detecting the NQR response signal, and an audio or visual alarm 108 which alerts the operator to the presence of the substance under test. Although not illustrated, the apparatus would normally include some means, such as a conveyor, for transporting the sample relative to the apparatus, so that a series of samples can be tested "on the fly".

With the technique of the preferred embodiment, each individual excitation pulse is shaped such that phase varies during the pulse, and preferably over a substantial proportion of the duration of the pulse, say, over at least 50%, more preferably at least 75 or 90%, most preferably over the entirety of the pulse. The phase modulation is preferably continuous in time, the excitation pulse itself being continuous in time (that is, there is no period during the pulse when the excitation is off, although the excitation may instantaneously pass through zero).

In more detail concerning the preferred embodiment, the excitation pulse application means 102 includes a radio-frequency power amplifier 110 whose output is connected to an r.f. probe 112 which contains one or more r.f. coils disposed about the sample to be tested (not shown).

The r.f. probe 112 also forms part of the detecting means 106 which also includes r.f.

receiver and detection circuitry 120.

The shaped pulse generating means 104 comprises a pulse programmer 130 for producing trigger signals to time the pulse, a spectrometer 132, manufactured by SMIS, United Kingdom, for generating a radio-frequency carrier signal at a known carrier reference frequency and fixed amplitude, the signal being gated by the trigger signals, a function generator 134, manufactured by Farnell, United Kingdom (Model No. SFG 25), for generating from a stored representation a waveform to modulate the amplitude of the carrier signal, the output of the function generator also being initiated by the trigger signals, and a double balanced mixer 136 for mixing the modulating waveform and the carrier signal and passing the mixed signal to the r.f. power amplifier 110. It will thus be appreciated that the shaped pulse generating means 104 is capable of applying to the sample a time-domain excitation waveform which is amplitude but not otherwise phase modulated.

The computer 100 ultimately controls all pulses, their radio-frequency, time, width, amplitude and phase. It is arranged to time the application of the excitation pulses substantially simultaneously with the arrival of a particular sample adjacent the probe 112. It also acts to receive the detected NQR response signal from the detecting means 106 and to process it, carry out signal addition or subtraction, and finally trigger the alarm 108 if appropriate.

The alarm 108 is arranged to be controlled by the computer 100 in such a way that an alarm is sounded or indicated in dependence on whether a predetermined threshold of detection has been exceeded, the threshold corresponding to a value of response signal strength which is no more than the value corresponding to the value of the ratio of the pulse repetition time (ç) to the extreme longest value of the spin-lattice relaxation time (T1). (This extreme longest value is usually that corresponding to the lower temperature extreme.) This can ensure that a detectible resonance response signal is produced over the entire selected temperature range, that is, between the two extreme temperature values.

As mentioned previously, the invention is based on the phase of the excitation varying generally non-linearly with excitation frequency. Thus the excitation can be represented as a complex function in the frequency domain (of the form E(o) = Ereai (o) + i > E] (cho)). In general, the excitation will also be a complex function in the time domain (of the form E(t) = Erect (t) + .Eaginy (t)). Hence a purely amplitude modulated time domain excitation waveform will in general not be adequate for the purposes of the present invention, so that the apparatus of Figure 3 would generally be inappropriate. However, there are special cases for which this apparatus can be used.

Two such cases are now exemplified with reference to Figures 4 and 5 respectively.

In both examples, the functions in the time domain had no fixed length, so that it was necessary to truncate them symmetrically at both ends where they tended to zero. The result of the truncation was to produce various artefacts in the frequency domain.

These could be reduced (but not avoided) by using some form of apodisation if desired.

In the first example (Figures 4), a rectangular shape was selected for the frequency domain excitation spectrum. The phase variation over the selected excitation frequency range was quadratic with respect to the frequency off-set (lav= v-v0), and anti-symmetric about the centre (carrier) frequency (v0). That is, in the lower frequency half-range phase varied quadratically and was negative, at the centre frequency phase was zero, and in the upper frequency half-range phase varied quadratically and was positive.

The time domain waveform obtained by inverse Fourier transformation of this spectrum is shown in Figure 4(a) (the real part of the waveform) and Figure 4(b) (the imaginary part). The waveform was truncated to a duration of lems. Note that in this example the imaginary part should have been zero in theory, and only appears because the centre of the spectrum was taken at point 512 of 1024 points instead of between 512 and 513.

It will be understood that in Figures 4(a) and 4(b) the radio-frequency carrier signal has been omitted for clarity. In fact, Figures 4(a) and 4(b) represent the modulating waveform (shaping function) which is mixed with the carrier signal. This convention is utilised from henceforth.

Figure 4(c) shows the modulus of the Fourier transform of the waveform of Figures 4(a) and 4(b). The deviation from a pure rectangular shape is caused by the truncation of the waveform.

Figure 4(d) is the same as Figure 4(c), except that the imaginary part of the time domain waveform (see Figure 4(b)) was set to zero. It can be seen that Figure 4(d) does not differ excessively from Figure 4(c). This is because the imaginary part of the waveform was in any case small. Hence it is acceptable to ignore the imaginary part of the waveform and utilise as the modulating waveform purely the real part shown in Figure 4(a).

Thus in the first example, in practice only the waveform shown in Figure 4(a) was utilised as the modulating waveform in the first embodiment of the apparatus (Figure 3). This waveform amplitude modulates, but does not otherwise phase modulate, the carrier signal. It will of course be understood that even "pure" amplitude modulation (achieved, for example, by multiplication of two signals with each other) gives rise to a change of phase (between 0 and 1800).

Figure 4(e) shows the result of an experiment with the explosive RDX as the substance of interest. The NQR testing apparatus was used to excite one of the '4N Vf resonances of RDX (the one near 5192 kHz having a free induction decay time, T2*, of roughly 1.4 ms at room temperature and a line width of about 200Hz). The variation of the integrated area under the NQR signal detected by the detection means with respect to the carrier frequency of the spectrometer is presented. The circular data points represent an excitation pulse having a bandwidth (the selected excitation frequency range) of 20 kHz; the bandwidth for the square data points was 10 kHz.

It can be seen that the two sets of experimental data points agree well. For the 20 kHz bandwidth the peak power (P) for the pulse was 208W for a 900effediye pulse. This compares with a peak power of 2000W for the corresponding rectangular excitation pulse of the same excitation bandwidth.

In the second example (Figures 5), again a rectangular shape, symmetric with respect to the carrier frequency, was selected for the frequency domain excitation spectrum, but only the upper half of the desired rectangular spectrum was selected. The phase variation over the selected excitation frequency range was quadratic. The example relies on the fact that if a carrier frequency vO is multiplied (mixed, amplitude modulated) with a linear modulating signal Vm the resultant is a pair of frequencies symmetric about the original carrier, with the carrier itself gone: vl2 = vO | vm.

Therefore if a complex waveform is derived which covers one half of the desired bandwidth on one side of the carrier frequency, and then only the real part of that waveform is taken, the complete bandwidth is covered. This is illustrated in Figures 5.

Figure 5(a) shows the modulus of the time domain waveform obtained by inverse Fourier transformation of the upper half-bandwidth. The dotted lines indicate where the waveform was truncated to create a lms pulse duration. Figure 5(b) shows the truncated real part of this waveform. This real part is the modulating waveform which was used to modulate the carrier signal. Figure 5(c) shows the real part of the Fourier transform of the truncated waveform of Figure 5(b), whilst Figure 5(d) shows the modulus of this transformation. The deviation from a rectangular shape is again due only to the truncation. In Figure 5(d) the carrier frequency is shown as a dotted line.

It is clear from Figure 5(d) that there is excitation of both sides of the carrier frequency.

Figure 5(e) shows the result of a similar experiment to that described in relation to Figure 4(e), but using the modulating waveform of Figure 5(b) rather than that of Figure 4(a). The bandwidth of the excitation pulse was 20kHz. The plot in Figure 5(e) matches closely that in Figure 4(e). The peak power (P) for the pulse was 133W for a 900effea,ve pulse.

Figure 5(f) is a similar plot to that shown in Figure 5(e), except that it was derived using a modulating waveform obtained by inverse Fourier transformation of the lower half-bandwidth. This modulating waveform is in fact the time reversal of that shown in Figure 5(b). Figures 5(e) and 5(f) differ only in that the former shows a peak in the NQR signal near 5192 kHz, whilst the latter shows a trough in the same region.

If the NQR response signal were to follow the excitation closely, the region would be expected to be nearly flat (as shown in Figure 5(d)).

The explanation for this difference is as follows. For the experiment of Figure 5(e) frequencies far 'from the carrier frequency were applied at the beginning of the pulse, whereas frequencies near the carrier frequency were applied at the end, as can be seen in Figure 5(b). Thus different NQR resonances were excited at different instants.

The free induction decay time T2* of 1.4ms for the particular resonance frequency of RDX under test is not much greater than the pulse length of Ims. Therefore the resonances far from the centre frequency excited at the start of the pulse have already partly decayed during the pulse and are hence weaker by the time detection takes place after the end of the pulse than those closer to the centre excited only at the end of the pulse. In Figure 5(f) the reverse is true, as a time reversed function was used, which explains the reversal of the peak of Figure 5(e).

Hence it is important when carrying out NQR tests using the present invention that the pulse duration is carefully controlled in relation to the T2* of the substance being tested, in order to prevent unacceptable loss of NQR response signal before detection.

Hence the pulse duration is preferably less than twice T2*, more preferably only 100%, 75% or even 50% of T2*. In the present case it is lms/1.4ms = 70%.

One way of maximising the pulse duration with respect to T2* would be to arrange for the NQR testing apparatus to generate a given number of pulses of the type described in relation to Figure 5(e), followed by an equal number of pulses of the type described in relation to Figure S(f), or vice versa. Addition of the two sets of responses would then yield a substantially flat response over the excitation bandwidth.

This technique may be important when testing the explosives TNT and PETN, since T2* is expected to be shorter for such explosives than it is for RDX.

The particular v+ line of RDX mentioned above as having a T2* of roughly 1.4ms is an especially advantageous line to investigate. The other v, lines have T2*,s of less than lms, and hence would necessitate the use of a shorter pulse. This would have the disadvantage of raising the excitation pulse peak power.

A variant of the preferred embodiment of NQR testing apparatus is now described with reference to Figure 6. Only the shaped pulse generating means 204 is illustrated; the remaining components are identical to those of the embodiment described with reference to Figure 3. In broad terms, the generating means 204 of the variant is similar to the generating means 104, except that it is a double channel rather than a single channel arrangement. Hence it can produce a waveform which is phase as well as amplitude modulated.

The variant again includes a pulse programmer 230 and spectrometer 232. However, two function generators 234a and 234b, and two double balanced mixers 236 are provided. In addition, a quadrature hybrid 0-90 splitter 238, a combiner 240 and resistors 242 and 244 are provided. In this embodiment, the splitter 238 is a SMHz splitter made by Mini Circuits (U.S.A.) and bearing Model Number PSCQ 2-5-1; the combiner 240 and mixers 236 are both made by Hatfield (U.K) and bear the Model Numbers DP102 and MC 291 respectively. The resistors 242 are 56Q whilst resistors 244 are 560Q. The whole network of resistors 242 together with resistors 244 results in a resistance of 50Q seen by function generators 234.

The variant functions as follows. Spectrometer 232 is gated and the outputs of the function generators 234 are initiated by the pulse programmer 230 as described in relation to the first embodiment. The splitter 238 produces from the radio-frequency carrier signal two radio frequency signals in relative quadrature. The function generators 234a and 234b generate the real and imaginary parts respectively of the modulating waveform. The resistors 242 provide impedance matching with the cables from the function generators, whilst the resistors 244 convert the voltage output of the function generators to a current output for passing to the mixers 236. After mixing of the relevant modulating waveforms and carrier signals in the mixers 236, the two resultant waveforms are combined in the combiner 240 to form an amplitude and phase modulated signal for passing to the radio-frequency power amplifier 110.

In a modification of the variant, a single function generator could be provided. The output of the generator would be passed through a further quadrature hybrid, the two outputs of which would be passed to the respective mixers 236. This modification would produce the type of modulation known in the field of telecommunications as single side-band modulation with suppressed carrier. The modification has the possible disadvantage that the quadrature hybrid would work at very low frequencies.

A third example of the use of the present invention is now described in relation to Figures 7. The variant of the apparatus described with refcrence to Figure 6 was used, with the excitation spectrum nominally rectangular and a quadratic phase variation.

In Figure 7(a) the truncated modulus of the modulating waveform derived by inverse Fourier transformation of the desired rectangular frequency domain excitation spectrum is shown. The real and imaginary parts of this waveform are shown respectively in Figures 7(b) and 7(c). The modulus of the Fourier transform of this waveform is shown in Figure 7(d), whilst the real part of this spectrum is shown in Figure 7(e). The waveform shown in Figure 7(d) is not perfectly rectangular because the waveform of Figure 7(a) has been truncated. Finally, in Figure 7(f) an oscillogram of the time domain excitation pulse is shown. This shows the effects of the modulating waveform on the radio frequency carrier signal.

The peak power of the pulse was 1.44W. The peak magnetic field (BS for the excitation pulse was no greater than 0.16 mT at a flip angle of 300effect,ve and for a 15kHz excitation bandwidth. These are the lowest values of peak power and B, of any of the examples. These low values were achieved mainly because the modulus of the time domain waveform had no zero crossing (compare Figure 7(a) with, for example, Figure 4(a)), but rather was relatively flat, so that power was distributed as evenly as possible throughout the duration of the excitation.

In a fourth example of NQR test, it was found that echo responses could be generated from the NQR resonance of RDX at 5.19MHz using shaped pulses of the type described above. For the experiment, a pair of 900effective shaped pulses was used.

Aside from the greater flip angle, the individual pulses were identical to those described in relation to the third example. In particular, a quadratic phase variation was employed. The echo signal amplitude was found to be similar to that obtained using a pair of rectangular 900eff pulses. The value of Blf in this fourth example was 0.47 mT.

It is to be noted that, in any of the embodiments or examples described above, it should be possible to reverse the phase distortion in the final spectrum of the excitation by data manipulation based on the known desired phase dependence with frequency. The absorption mode signal can then be used, rather than the modulus, for detection purposes.

It has been found that the shaped pulses described herein may be most satisfactorily employed in experiments with substances having signals exhibiting relatively small line widths. A small line width is preferable because it enables the excitation of large frequency ranges with low Barf field and low power. If only relatively broad lines are available, higher power may need to be used. For RDX, a good line is that at 5190 kHz, which has a line width of only about 200 Hz at room temperature.

Finally, with reference to Figure 8, an example of the method by which the detection apparatus of the present invention can be configured is now described. In the example, the apparatus was configured to detect the presence of the explosive RDX. The means for applying excitation was arranged to apply a sequence of regularly repeating pulses, as shown in the pulse timing diagram of Figure 8. The detection means was arranged to detect the free induction decays following each pulse. The pulses were shaped as described in relation to Figures 7 above, and were at a frequency, f, of roughly 5.2 MHz, so as to be able to excite the RDX resonance at 5.191 MHz at room temperature. They were of sufficient bandwidth (roughly 20kHz) to accommodate a temperature range of -S to +35"C. They each had a duration (t3 of roughly lms.

Over this temperature range, spin-lattice relaxation time, T1, varies by a factor of 100, between 500ms at the lower extreme of temperature and Sms at the higher extreme.

These being the circumstances, pulse repetition time, T, was set to 25ms, giving a range of the ratio TiT1 of 0.05 to S. Flip angle, a, was set to 30 actual. With these values, it can be seen from Figure 3 of the Buess et al. paper that signal strength was (at least in theory) only 2.5 times greater at a temperature of +35 C than it was at a temperature of -5 C. This was considered to be an acceptable variation. It is to be noted that T was considerably greater than the sum of tf (which was lms) and the duration of the free induction decay appropriate to the substance and the selected frequency. The duration was approximately 2 or 3 times the free induction decay time, T. of 1.4 ms.

It will of course be understood that the present invention has been described above purely by way of example, and modifications of detail can be made within the scope of the invention.

Each feature disclosed in the description, and (where appropriate) the claims and drawings may be provided independently or in any appropriate combination.

Claims (29)

1. A method of detecting the presence of a given quadrupolar nucleus in a sample, the resonance frequency and spin-lattice relaxation time of the nucleus varying with a given environmental parameter, comprising determining a range of the resonance frequency which corresponds to a selected range of the parameter, applying excitation to the sample to excite nuclear quadrupole resonance, and detecting the resonance response signal, the excitation being such that a detectable resonance response signal can be excited over the resonance frequency range, and being such as to produce a constant flip angle of less than 45 degrees actual.

2. A method according to Claim 1 wherein the excitation is pulsed excitation at a constant pulse repetition rate.

3. A method according to Claim 2 further including selecting an expected extreme value of the environmental parameter, being that extreme value for which the relaxation time has its shortest value, and determining a measure of the ratio of the pulse repetition time to the shortest value of the relaxation time, the pulse repetition rate being such that said measure of the ratio is no more than a selected value.

4. A method of configuring apparatus for detecting the presence of a given quadrupolar nucleus in a sample, the spin-lattice relaxation time of the nucleus varying with a given environmental parameter, the apparatus including means for applying pulsed excitation to the sample to excite nuclear quadrupolar resonance, the method including selecting an expected extreme value of the environmental parameter, being that extreme value for which the relaxation time has its shortest value, arranging the excitation means to apply excitation at a constant pulse repetition rate and such as to produce a constant flip angle, and determining a measure of the ratio of the pulse repetition time to the shortest value of the relaxation time, the pulse repetition time being such that said measure of the ratio is no more than a selected value.

S. A method according to Claim 4, for detecting the presence of the given nucleus in the sample, the apparatus further including means for detecting the resonance response signal, and including the step of operating the excitation and detection means to detect the presence of such nucleus.

6. A method of detecting the presence of a given quadrupolar nucleus in a sample, the spin-lattice relaxation time of the nucleus varying with a given environmental parameter, including selecting an expected extreme value of the environmental parameter, being that extreme value for which the relaxation time has its shortest value, applying pulsed excitation to the sample to excite nuclear quadrupole resonance, and detecting the resonance response signal, wherein the excitation is at a constant pulse repetition rate and such as to produce a constant flip angle, a measure of the ratio of the pulse repetition time to the shortest value of the relaxation time is determined, and the pulse repetition time is such that said measure of the ratio is no more than a selected value.

7. A method according to Claim 3, 4, 5 or 6 wherein the value of said ratio is no more than 20, preferably no more than 10, more preferably no more than 5, yet more preferably no more than 1.

8. A method according to any of Claims 3 to 7 including further selecting an expected extreme value of the environmental parameter for which the relaxation time has its longest value, the ratio of, on the one hand, the repetition time, to, on the other hand, the relaxation time corresponding to a value of the environmental parameter centred between its expected extremes, being in the range 0.05 to 5, preferably 0.1 to 1.

9. A method according to any of Claims 1 to 3 or S to 8 wherein an alarm signal is produced from the response signal in dependence on whether a predetermined threshold of detection has been exceeded.

10. A method according to any of Claims 2, 3 or S to 7 including further selecting an expected extreme value of the environmental parameter for which the relaxation time has its longest value, determining a measure of the ratio of the pulse repetition time to the longest value of the relaxation time, and producing an alarm signal from the response signal in dependence on whether a predetermined threshold of detection has been exceeded, the threshold corresponding to a value of response signal strength which is no more than a value corresponding to said measure of the ratio of the pulse repetition time to such longest relaxation time.

11. A method according to any of Claims 3 to 7 including further selecting an expected extreme value of the environmental parameter for which the relaxation time has its longest value, and determining a measure of the ratio of the pulse repetition time to the longest value of the relaxation time, the value of signal strength corresponding to the value of the ratio of the repetition time to the shortest relaxation time being less than a selected number of times, preferably less than five times, more preferably less than three times, even more preferably less than twice, the equivalent signal strength value for the longest relaxation time.

12. A method according to Claim 11 wherein, given the selected range of the environmental parameter and a desired range of the response signal strength, the values of pulse repetition rate and flip angle are chosen by a process of iteration between these two quantities.

13. A method according to any of Claims 4 to 6 wherein the excitation is such as to produce a flip angle of less than 45 degrees actual.

14. A method according to Claim 1, 2, 3 or 13 wherein the excitation is such as to produce a flip angle of less than 35 degrees, preferably less than 25 degrees, more preferably less than 15 degrees, more preferably still less than 10 degrees, yet more preferably less than S degrees actual.

15. A method according to any of Claims 1 to 3 and 5 to 14 wherein the excitation is applied at a single excitation frequency.

16. A method according to any of Claims 1 to 3 and S to 15 wherein the excitation is applied to the sample over a given frequency range, the time domain waveform of the excitation being either phase modulated or amplitude modulated, or both phase modulated and amplitude modulated.

17. A method according to Claim 16 wherein the phase of the excitation varies generally non-linearly with the excitation frequency over the selected range.

18. A method according to Claim 16 or 17 wherein the excitation is pulsed and the phase of the excitation varies generally quadratically with the frequency off-set.

19. A method according to Claim 16, 17 or 18 wherein a selected NQR resonance is excited with at least one excitation pulse, and the duration of the or each excitation pulse is less than twice the free induction decay time appropriate to that resonance.

20. A method according to any of the preceding claims wherein the environmental parameter is temperature.

21. A method according to any of Claims 3 to 7 wherein the environmental parameter is temperature and the higher expected extreme value is less than 50 C, preferably less than 40 or 30"C.

22. A method according to Claim 8, 10, 11 or 12 wherein the environmental parameter is temperature and the lower expected extreme value is greater than -30 C, preferably greater than -20 or -10"C.

23. A method according to any of Claims 2 to 8, 10 to 13, 21 or 22 wherein the pulse repetition time is greater than, preferably 3 times greater than, the free induction decay time T2 of the quadrupolar nuclei.

24. Apparatus for detecting the presence of a given quadrupolar nucleus in a sample using pulsed excitation, comprising storage means in which is stored information for at least two types of nucleus having different NQR properties on suitable values of flip angle and pulse repetition rate to take account of expected variations in temperature, means responsive to the storage means for applying, for a given type of nucleus, pulsed excitation to the sample to excite nuclear quadrupole resonance, the excitation means being adapted to apply excitation at a constant pulse repetition rate and such as to produce a constant flip angle suitable for the particular substance, and means for detecting the resonance response signal.

25. Apparatus according to Claim 24 including means for producing an alarm signal from the detected response signal in dependence on whether a predetermined threshold of detection has been exceeded.

26. Apparatus for detecting the presence of a given quadrupolar nucleus in a sample when configured according to the method of Claim 4 or S.

27. A method of detecting the presence of a given quadrupolar nucleus in a sample substantially as herein described.

28. A method of configuring apparatus for detecting the presence of a given quadrupolar nucleus in a sample substantially as herein described.

29. Apparatus for detecting the presence of a given quadrupolar nucleus in a sample substantially as herein described with reference to, and as illustrated in, Figures 3, 6 or 8 of the accompanying drawings.