The invention relates to a method for varying adaptively pulse interval in NMR- based water content measurement according to the preamble of claim 1 ,

The invention also relates to an apparatus for adaptive pulse interval adjustment in NMR-based water content measurement.

NMR- technology (Nuclear Magnetic Resonance) has been used for determining moisture content of materials. For example FR 2786567 describes this kind of a system. The present systems are clumsy and expensive and therefore used rarely in commercial application.

It is an object of the present invention to provide a novel type of NMR-based water content measurement capable of overcoming at least some problems of the prior-art technology described in the foregoing. The invention is based on the concept of using such pulse sequences, where the rate of pulses is optimized for different humidity levels of the sample to be measured. The said optimization is advantageously based on estimating the so-called spin- lattice relaxation time constant. Furthermore, also the measuring equipment is characterised by using low energy magnetic field and a weighing apparatus.

More specifically, the method according to the invention is characterized by what is stated in the characterizing part of claim 1.

Furthermore, the apparatus according to the invention is characterized by what is stated in the characterizing part of claim 11. The invention offers significant benefits.

Firstly, the measurement time may be minimized for all humidities and sample materials.

Secondly, the measurement equipment is light weight and inexpensive without compromising the measurement accuracy. In the following, the invention will be examined with the help of exemplifying embodiments illustrated in the appended drawings in which

Figure 1 presents schematically the basic conception of the NMR-humidity measurement equipment suitable for the invention.

Figure 2 presents graphically typical NMR signals with their relaxation times.

Figure 3 presents a typical pulse sequence in accordance with the invention. Figure 4 shows graphically an example of the effect of the pulse interval on the amplitude sum over a large number of pulses.

In accordance with figure 1 in NMR -humidity measurement a homogeneous DC- Magnetic field is generated by a magnet 2 into the a sample 1 to be measured, then interaction of the magnetic field with the hydrogen in the sample 1 causes a small magnetization to develop in the sample 1. Next, the sample 1 is exposed to a short intense radio frequency (RF) excitation pulse 3 (Figures 2 and 3) by a transmitter/receiver 5, which excites the hydrogen nuclei. In the following step the measure- ment instrument 5 records the NMR signal (known as a Free Induction Decay or FID) for a period of milliseconds. During this time, some parts of the sample undergo NMR relaxation and return to the original state. The signal amplitude 10 (Fig. 2) at predetermined time (tens of microseconds following the first RF pulse) is proportional to the total amount of hydrogen from moisture of the samples. Therefore, the maximum value of NMR-signal defines the moisture content. In practice this maximum value 10 is often extrapolated from a measurement performed some time after the real maximum 10.

Nuclear Magnetic Resonance-based instrument can easily be configured to yield an electrical signal that is proportional to the content of hydrogen containing liquids in a solid material. The NMR-device is especially well suited for measuring the water content in biomass. When the sample to be measured is very dry, typically meaning water content of less than 20 m-%, the signal-to-noise -ratio is low, which is typically compensated for by increasing the number of successive measurements and averaging them. This easily leads to a long measurement time. The limitation for the time between successive measurements is primarily set by the so-called spin-lattice relaxation time (hereafter called Tl). This is the time required for the deflected average magnetization vector to recover its original value. The recovery is enabled by energy dissipation from the protons to the lattice. If the excitation pulse is applied before the full relaxation, reduced signal amplitude is observed, and the correlation coefficient between the water content and the signal amplitude is altered, and thus calibration will not be valid.

Tl is essentially a function of interaction between the nuclear spin and the lattice,. Generally, the drier the material the shorter the corresponding Tl .This phenomenon can be utilized in optimizing the pulse interval, meaning that S N-ratio for dry samples can be increased significantly for a given total measurement time.

Tl is the time it takes for the nuclear magnetisation to recover approximately 63% [l-(l/e)] of its initial value after being flipped into the magnetic transverse plane. Different tissues have different Tl values. For example, fluids have long Tl ( 1500- 2000 ms), and water based tissues are in the 400-1200 ms range. In accordance with figure 2, T2 characterizes the rate at which the component of the magnetization vector decays in the transverse magnetic plane. It is the time it takes for the transverse magnetisation to reach 37% (1/e) of its initial magnitude after flipping into the magnetic transverse plane. Hence the relation:

M _∑y (t) = M _xy (Q )e ^~t/T2

T2 decay occurs typically 5 to 10 times more rapidly than Tl recovery, and different tissues have different T2s. For example, fluids have the longest T2s (700-1200 ms), and water based tissues are in the 40-200 ms range.

The method consists typically of two steps: 1. Estimating the Tl time for the sample. This can be achieved in accordance with figures 2 and 3 e.g. by measuring the response signal amplitude 10 using successive excitation pulse sequences 2 with a constant number of pulses 3 and a stepwise increasing pulse interval T3, and detecting the minimum interval T3 required for the response signal to remain at the constant (maximum) level 10. As told earlier, the maximum value 10 may be determined by extrapolation from a delayed measurement. Estimation of Tl can be performed in a few seconds.

Another method of estimating Tl is to measure the spin-spin relaxation time T2, and estimate Tl from T2. Typically both decrease when the water content of the sample decreases. Actually the T2 is usually estimated based on the measured value of T2*, which is a combined result of spin- spin-relaxations and decoherence effect caused by inhomogeneity of the primary magnetic field being device specific. A third method of estimating Tl comprises the use of two successive excitation pulse sequences, each of which is preceded by a so-called saturation pulse sequence. The pulse interval in the said excitation pulse sequences is advantageously larger than T2*, but preferably not significantly smaller than Tl . The two successive excitation pulse sequences shall have different pulse intervals tl and t2, e.g. tl =Tl and t2=(2*Tl). The ratio of signal amplitudes A1/A2 obtained with pulse interval tl and t2, respectively, can be calculated from the following equation:

Which can be solved numerically for Tl .

Yet another means to estimate Tl is to use the water signal amplitude per unit mass of sample: the lower the said ratio is (the dryer the sample is), the shorter is the Tl . This estimation method is valid only for a limited range of samples, for example solid biofuels.

Yet another method of estimating Tl comprises the use of two successive excitation pulse sequences, each of which is_optionalIy preceded by a so-called saturation pulse sequence.

Without the saturation pulse sequences, the estimate for Tl can be numerically solved from he equation:

The methods described above are only examples of the possible means to estimate 2, Performing the actual measurement using the minimum pulse interval that yields a constant (maximum) amplitude with sufficient, e.g. 1% accuracy.

Typically, such minimum pulse interval T3 is 5*7 1 . In this way, the number of averaged pulses within 20 s measurement time frame can be increased from approx. 10 (long pulse interval required by wet samples) to approx. 200 (short pulse interval enabled by very dry samples), thus improving the S N by a factor of sqrt(200/l 0) = 4.5. Low NMR signal-to-noise ratio typically obtained from dry samples can be enhanced by shorter measurement interval and thus larger number of individual measurements. The optimum pulse interval is determined using a probe pulse sequence to estimate the spin-lattice relaxation time constant Tl , which advantageously can be used as an input for calculating the low limit for the pulse interval. The disclosed method can improve S/N ratio of very dry samples by a factor of five.

As can be seen from figure 4 the sample with medium moisture content can be measured with shorter pulse interval than the wet sample.