DE69130574T2 - REAL-TIME ANALYSIS WITH MAGNETIC RESONANCE AND THEIR INDUSTRIAL APPLICATION - Google Patents

Description

SCOPE OF INVENTION

The present invention relates to an instrument for measuring the type and amount of lattice-bound and free magnetically active nuclei in successive samples of a process material flow by pulsed nuclear magnetic resonance (NMR) techniques and in particular to the application of such measurement to industrial process control of moisture content, polymer content, crystallinity fraction as well as to other component content analyses and other parameters.

BACKGROUND OF THE INVENTION

Magnetic resonance imaging (MRI) and NMR techniques have grown rapidly over the last 40 years, particularly in the medical instrumentation field, where in vivo examinations can visualize various parts of the human body, and in clinical research laboratories. In addition, these techniques have found application and interest in the industrial instrumentation and control fields. The present invention enables an effective use (technically and economically) of pulsed NMR techniques in industrial fields to replace or complement existing energy-based optical and radiation instruments.

Pulsed NMR spectroscopy works with a burst or pulse that has the task of exciting all nuclei of a certain nuclear species of a sample being measured (the protons or the like of such a sample having first been subjected to a precession in an essentially static magnetic field); in other words, the precession is modified by the pulse. After the pulse is applied, a free induction decay (FID) of the magnetization in related to the excited nuclei. A traditional Fourier transform analysis produces a frequency domain spectrum that can be used advantageously in studying the nuclei of interest. The duration of the pulses, the time between pulses, the pulse phase angle and the composition of the sample are parameters that affect the sensitivity of this technique. These frequency domain techniques are not easily applied in industrial applications, especially in online applications.

US 4701705 describes the application of nuclear magnetic resonance (NMR) for moisture measurements on material flowing in a pipe.

It is also known to perform NMR measurements using a laboratory system operating at a frequency of 27 MHz, where samples are placed in glass vials and placed in a detector coil in the gap of an electromagnet (see e.g. COMPOSITES EVALUATION, PROC. OF SECOND INTERN. CONF. ON TESTING, EVALUATION AND QUALITY CONTROL OF COMPOSITES - TEQC 87, 22 September 1987, Guildford, GB, pages 23-27; G.A. Matzkanin et al. "NDE OF MOISTURE ABSORPTION IN COMPOSITES USING NUCLEAR MAGNETIC RESONANCE").

It is also known to use a Bruker PC20 NMR measuring device operating at 20 MHz for studying the dynamics of the crystallization of a polymer, the phase separation process in polymer mixtures and in binary liquid mixtures, the gelatin process, the polymerization process and biological and chemical reactions (see JOURNAL OF APPLIED PHYSICS, Vol. 60, No. 4, August 1, 1986, New York, USA, pages 1306-1309; H. Tanaka et al. "REAL-TIME PULSED NUCLEAR MAGNETIC RESONANCE MEASUREMENT SYSTEM FOR THE STUDY OF NONEQUILIBRIUM PHENOMENA IN POLYMERS").

It is an object of the present invention to provide a to provide an improved measurement system that leads to accurate and rapid determination of the types and amounts of nuclear species of interest.

It is a further object of the invention to use a time domain analysis for providing such a system.

A further object of the present invention is its application to the industrial online problems of measuring control processes.

The main variable of interest is humidity, with discrimination between free and bound water in an organic or inorganic substance based on hydrogen nucleus precession analysis. However, other parameters based on hydrogen or other sensitive species such as sodium can also be measured. It is an object of the present invention to enable a variety of different such measurement tasks.

Another task is to perform the dynamics of industrial online applications including variations in density, temperature, packing and size factors, friction and static electricity, vibration as well as frequent, repetitive, cyclic and non-cyclic measurements.

It is a further object of the invention to use magnetic resonance techniques (NMR) in polymer analysis including dense melt index and/or isostatic index measurements.

Another task is to increase the accuracy and reliability of the data obtained.

It is an object of the invention to achieve the necessary practical level of economy in accordance with the above objects.

It is a further object of the invention to provide all the features of an accurate and rapid determination of the types and amounts of the nuclear species of interest, the use of time domain analysis in such a system, its application to the industrial on-line problems of monitoring and controlling processes, e.g. the measurement of free and bound water in organic and inorganic substances (based on precision analysis with modified hydrogen nuclei) or other parametric measurement (based on hydrogen or other sensitive species, including e.g. sodium-23 or carbon-13) taking into account the dynamics of industrial on-line applications including variations in density, temperature, packing and size factors, friction and static electricity, vibration and frequent, repetitive cyclic and non-cyclic measurements.

It is a further object of the invention to apply such magnetic resonance techniques in polymer analysis, including density, all with improved accuracy and reliability of the data obtained and while achieving the necessary level of practical economy.

It is a further object of the present invention to further extend these capabilities in relation to industrial online processing and the like in application to mixed species (or mixed phases) of NMR active materials and in particular to food and plastic materials (and applicable to many other NMR active materials) with a third component such as oils/fats or solvents in addition to two main components (moisture/solids, crystalline/amorphous).

The present invention provides a magnetic resonance system for industrial process monitoring, comprising:

(a) a means for accessing successive samples from an industrial process, placing the samples in a sample measuring area and removing successive samples from the area,

(b) means for applying a base magnetic field to the region to cause precession of sample nuclei therein and for applying a local resonant excitation pulse to the region to modify the precession,

(c) means for defining a receiving antenna coil and a signal conversion means, cooperating with the sample to cause a precession of nuclei to be measured and a relaxation which is detected at the coil as free induction decay, and

(d) a means for creating a digitized version of the free induction decay and analyzing it into bound (Gaussian) and unbound (exponential) components of the decay curve and for extracting a ratio between the detection (intercept) of one of the components and the sum of the detections (intercepts) of the components,

wherein the means (a) to (d) are arranged:

(1) to stabilize base field application and excitation and conversion conditions in response to changes in one or more parameters of sample conditions in the measurement range;

(2) to effect such measurement and ratio extraction and associated stabilization on a very rapid repeat basis of successive samples from the process; and

(3) for converting at least one thousand data points in each decay curve and effecting a reduction in at least one hundred points thereof so that sample component proportions can be reliably and repeatedly measurable from sample to sample in rapid succession for real-time on-line monitoring, the system further comprising: means for extending each of a plurality of decay curves with an echo refocus/decay technique to acquire data relating to at least one further subcomponent of the sample for increasing the accuracy of the ratio determination of (d) and/or for identifying the proportion of the at least one further subcomponent.

The NMR system provides reliable extraction of free induction decay in a manner that is practical in an on-line industrial context and from an economic standpoint. The system is characterized by achieving a baseline magnetic field homogeneity to a reasonable degree and correcting for inhomogeneity effects, by temperature stabilization to a reasonable degree and correcting for thermal drift effects, and by performing multiple runs (50-500) for each measurement with digital data reduction and applying statistical techniques or other data manipulation for an industrially effective measurement. These data can be represented for discussion/analysis as a free induction decay (FID) curve with attention to time sequence components of a first, very fast, essentially Gaussian part, followed by a slower, essentially exponential region, representative of proton relaxation following initial excitation by a pulse of transmitted and resonantly coupled radio frequency energy, which induces a modification of the precession of protons in the measured sample in a highly static magnetic field.

The Gaussian FID part is based on measurement data points of magnetization decay of immobile or highly confined protons present in the sample and picked up at the receiver of the NMR system. This contribution is usually based on the response of bound proton species such as the hydrogen or hydrate content of chemical compounds (as well as for many other NMR-active species such as sodium-23 or carbon-13 which are chemically bound). The exponential FID contribution is usually based on little or no confined NMR-active species such as moisture which is physically present in a sample but essentially not chemically bound to it. The Gaussian and exponential FID contributions and the overall FID can be extrapolated to a decay origin which is usually set at the time center of the excitation pulse. Zero-time acquisitions of these curves (ie FID and one or more of these component curve sections) yield ratio data with the FTD section(s) of one or more of the curve sections, for example to determine the proportion of free versus bound water in a wet material (eg for process control of industrial chemicals, minerals and metals, agricultural commodities, processed foods by determining the moisture content for upstream or downstream correction or for acceptance/rejection purposes). Instead of determining the moisture in eg a food product, the task may be to determine the ratio of relatively crystalline to non-crystalline components of a material, eg hard and soft components of a plastic material, and this is done in a manner analogous to moisture measurement. Tightness can also be determined using the invention since the FID can be predictably measured in Depending on the tightness varies.

Within such a measurement, a source of error may arise if the material has a fat or oil or organic solvent component that has an exponential response similar to that of moisture content or the like. Such fats or oils or solvents may be present in food. In some cases, it may be desirable to actually detect such fat, oil or solvent content - and not simply separate it from the detection of moisture or the like. Likewise, residual solvents in plastics or other industrial materials may appear as an error factor to be eliminated and/or as a target parameter to be measured. The NMR response of such a solvent or fat/oil component can be isolated using a Hahn spin echo technique as an extension of each of several measurement cycles. When an FID has progressed from an original excitation and a Gaussian and exponential response section, then a new secondary refocusing pulse (usually a 180º or pi rotation compared to the original 90º or pi/2 excitation pulse) is applied to generate an echo, and then a third pulse (also pi) is applied to obtain a second echo that perfectly reflects the solvent or oil/fat content. The second echo can be treated independently of stored data from the first FID and then subtracted from the first FID to obtain a valid first FID intercept (excluding oil/fat effects) which is reducible via ratio analysis to the intercept of the Gaussian part of the FID curve to derive valid data on solids/liquid content (or crystallinity or moisture).

The invention also encompasses the direct acquisition of data with appropriate calibration from the tertiary FID to show solvent or fat/oil content. Such techniques can be extended to some extent to distinguish among subcomponents of such solvents or fats/oils.

The measuring system of the invention comprises economically miniaturized and industrially hardened parts compared to the widely used laboratory systems. A substantially fixed magnetic field is generated by closely spaced pole pieces with a field of 4000 to 8000 gauss (nominally about 4700 gauss). Helmholtz coils are provided which are adjustable to achieve rapid adjustments for the precise, correct field and are superimposed with coarser and slower adjustments of the thermal environment. This is to ensure that the product of a material-related constant (gamma) multiplied by the magnetic field strength, which is the resonant frequency, matches the excitation frequency. Further fine adjustment is made during signal processing as described below.

The system of the present invention conveniently handles large data streams through features described below that are related to the thermal controls to achieve a measurement system that meets the above objectives. The materials of construction are also included in the reliability considerations as described below. Measurement of a sample is often completed in less than a few minutes (as opposed to hours of measurement in many prior art systems).

The measurements carried out according to the invention on the basis of section ratios and/or integrated areas under curves and/or peak analysis are independent of sample weight or volume within a measuring range, whereas precise weight measurement is a necessary feature and a limitation of many state-of-the-art systems.

Previous efforts at industrial on-line measurements of the same parameters discussed here included gravimetric, radioactive, acoustic, optical and electrostatic/capacitive non-NMR systems, but none were fully satisfactory for the purpose at hand, and the use of NMR to support continuous industrial processes pushed the focus to off-line laboratory instruments [at considerable cost and yet with inadequate sampling capabilities], or some earlier efforts in the 1980s with industrially hardened pulsed NMR instruments using only one or two data points for FID analysis.

Laboratory procedures for frequency domain NMR analysis have been described for crystallinity content determination, e.g., in Spiess, "Molecular Dynamics of Solid Polymers As Revealed By Deuteron NMR," 261 Colloid & Polymer Science 193-209, (1983), and Kauffman et al., "Determination of Transition Temperatures and Crystalline Content of Linear High Molecular-Weight Polyethylene by Proton NMR Spectroscopy," 27 Jl. of Polymer Science 2203-2209 (1989). Time domain analysis using pulsed and multi-pulsed NMR/free induction decay in coal to detect free radicals therein is shown in the laboratory systems of Gerstein et al (Iowa State University Ames Laboratory), published in "Utility of Pulse Nuclear Magnetic Resonance In Studying Protons In Coals," 81 Jl. of Phys. Chem. 566-571 (1977), and "1H Nuclear Magnetic Resonance Studies of Domain Structures In Polymers," 52(9) J. Appl. Phys. 5517-5528 (1981). The IBM Federal Systems Division Model and Brucker GmbH Model P201 instruments or earlier instruments and the description in U.S. Patent 4,430,719, issued to Pearson on February 7, 1984, are the earlier attempts at industrial application of NMR techniques mentioned above. Pearson's work was incorporated into industrial plant control work by Kaiser Aluminum & Chemical Corp. in 1985. The instrument was designed for a Auburn International, Inc. offered the Pearson/Kaiser product for sale in 1987-1988, but it did not meet the requirements for industrial on-line monitoring.

Further tasks, features and advantages are apparent from the following detailed description of preferred designs in conjunction with the accompanying drawings.

Figures 1 and 2 are side and cross-sectional views of a preferred embodiment of the invention including electrical components in block diagram form;

Fig. 3 shows the voltage-time waveforms of free induction decay (FID) of the embodiment of Fig. 1-2 during operation;

Fig. 4 is a flow chart of measurement steps using the device of Figs. 1-2, including its signal processing segments (the activity of which is illustrated by the waveform of Fig. 3);

Fig. 5 is a voltage-time waveform for free induction decay (FID) of the embodiment of Fig. 1-2, modified to utilize the Hahn spin echo;

Fig. 6 is a flow chart of measurement steps using the embodiment of Figs. 1-2 modified for Hahn spin echo use, including its signal processing elements (the activity of which is described by the waveform of Fig. 5) and also indicating density and moisture content;

Fig. 7 is a voltage (or other variable intensity parameter) time trace for FID curves of samples of different densities using the above-mentioned modified embodiment of the present invention, yielding relative and absolute density measurements of samples;

Fig. 8 shows a calibration curve for an absolute density measurement with the above-mentioned embodiments of the present invention applied to polyethylene; and

Fig. 9A, 9B, 9C Decay curves of intensity (I) versus time (t) illustrating operation of the modified embodiment according to further aspects thereof for determining decay time constant shifts and scaling bias factors and for measuring a solvent or oil/fat or similar other component of a sample by simple peak values (averaged) rather than a (reused) full FID analysis.

DETAILED DESCRIPTION OF PREFERRED DESIGNS

Fig. 1 to 2 show a side view and a cross-sectional view with block diagram inserts of a first embodiment of the invention. Material flows through an industrial process line IPL as indicated by the arrow A. Part of the material is a probe P and is directed through an inlet line LI into a sample region S1. Said region is defined by a tubing 98, typically 1 foot (0.3 m) long, made of a substantially non-magnetic, non-conductive material which does not in itself generate significantly interfering FID signals (glass, certain ceramics, certain plastics or hybrid materials). The sample region is defined between the inlet and outlet valves V1 and V2. Gas jets J are also provided. These are repeatedly pulsed on and off to agitate flowing sample materials during the introduction and expulsion of sample. Region S2 is the critical section of the sample. It is surrounded by a sample coil 100 which is tuned to resonance and is controlled by a tuning circuit 102 and an associated transmitter/receiver control 104. Grounded loops 101 form active shields according to Lenz's law, provided above and below the coil 100 to help shape the field of the coil 100 - that is, to contain the field created by an excitation pulse. The controller 104 includes an on-board microprocessor and the necessary power supplies, memory, program and I/O decoding for interfacing with the hardware shown and with an external microcomputer 106 with keyboard 108, monitor (or other display) 110, recorder 112 and/or process controller 114 (for controlling the process at IPL). The operator initiates and controls the operation from the display keyboard 108 and the resulting data and signals are subsequently displayed on the display 100 and processed in 110, 112 and/or 114. The computer 106 also controls the operating conditions of the instruments.

The region S2 of the tubing 98 and coil 100 is in a static but adjustable crossover magnetic field defined by a magnet assembly 116 comprising a yoke 118, pole pieces 120, surrounding Helmholtz lobes 124 and a coil current generator 122. The critical sample region S2 of type 98 [sic...*1] and magnet are housed in a metallic (but not ferromagnetic) box 126 having highly thermally conductive shield plates 128 and internal compartments 130 and a mutually related total mass to minimize harmonics and other interference with a microwave signal radiated from coil 100 to a sample and/or returned from the sample for pickup by coil 100 and its tuned circuitry 102 and transmit/receive control 104.

The magnet assembly 116 with the yoke 118 and other parts therein as shown in Fig. 1-2 is in turn located in an environmental control chamber 132 with gas filling and purging elements (not shown), an internal gas heater 134, a motor M for driving the fan 136, and a temperature sensor 138 which can be applied to the yoke or other sensing area whose temperature reflects the temperature at pole pieces 120 and the sample area therebetween. A thermal control 140 processes temperature signals from 138 to adjust heating/circulation at 134/136 as coarse control and to adjust the current through the Helmholtz coils 124 on magnetic pole pieces 120 as sensitive and fast fine control, as well as to execute general control commands from the computer 106. Further thermal stabilization is provided by a control heat exchanger 142 with a pump 144 and coils 146 attached to the yoke 118, and coils 148 attached to the plates 128 of the box 126. Further or alternative thermal stabilization may be achieved by a controlled heat exchanger [sic] or a hot air flow ring [sic...*2], the ring ANN being defined between concentric cylinders W1 and W2 running the full length of the sample tubing SIA, with a hot air flow being passed through the ring length.

The strength, consistency and constancy of the magnetic field between poles 120 in region S2 of the sample are thus controlled by a uniform base magnetic field throughout region S2. The Helmholtz coils 124 are energized by the coil current controller 140 to precisely adjust the final magnitude of the field in which the sample is located. This field is the vector addition of the fields due to the magnetic poles 120 and the Helmholtz coils 124. The controller 140 adjusts the current through the Helmholtz coils 124 using current generators. The coils 124 are wound around the magnet 110 so that the magnetic field generated by the current in the coils 114 can be added to or subtracted from the field generated by the magnetic pole pieces 120. The magnitude of the current through the coils 124 determines the strength of the field that is added to or subtracted from the field due only to the magnetic pole pieces 120 (and the associated yoke structure).

The actual determination of the current through the Helmholtz coils is done using the magnetic energy and resonance techniques described below in preliminary runs and by adjusting the Helmholtz current until the maximum sensitive resonance is reached.

The main electrical control elements are located in the tuner 102, including coils 100 and 101, variable capacitors 102-1 and 102-2, resistor 102-3 and diodes 102-4, and are designed to tune to a Q value of twenty to fifty designed to achieve resonance of coil 100 and controller 104 including transmit/receive switch 104-1, transmitter 104-2 and receiver 104-3, crystal oscillator 104-4, pulse gate generator (PPG) 104-5 and phase shifter 104-6. The crystal presents a nominal 20 MHz carrier which is phase modulated or demodulated by the MOD, DEMOD elements of transmitter 104-2 and receiver 104-3. The receiver includes variable gain amplifier elements 104-31 and 104-32 for operation. The received analog signals are fed into a high speed flash converter 105-1 and an internal (to the instrument) CPU element 105-2 which sends data to an external computer 106 provided with keyboard 108, monitor 109, modem 110, recorder elements 112 and process control elements 114 eg for controlling valves V1, V2 via valve control element 115 and/or to coil current controllers 122, all with D/A converters (not shown).

The excitation of coil 100 and the excitation precession of the proton content of the sample and the subsequent relaxation/decay produce a received FM signal which, after demodulation, controlled amplification, A/D conversion and point plotting, has the free induction decay (FID) waveform C shown in Fig. 3.

Fig. 3, the voltage-time curve, shows the elements of a "cycle" (with subcycles (+) and (-)) of sample excitation and free induction decay. In each (+)/(-) subcycle, an excitation energy pulse is applied. The center of the excitation pulse is assumed to be t0. The electronic components of the transmitter/receiver 104 do not receive effectively until the saturation effects are overcome at t1. Then a useful C(+) or (C-) curve is formed. The signal processing devices can add or subtract subsequent C+ and C- shapes for useful adjustment as described below.

The FID curve data is stored in the external computer 106 where a program determines the best curve that fits each stored FID curve. The FID curve C has two component parts shown in Figure 3 as A and B. The A curve, which dominates the first part of the FID curve, is a Gaussian curve, while the B curve, which dominates this latter part of the FID curve, is an exponential decay curve. The Gaussian and exponential portions are respectively governed by the bound and unbound proton content of the samples (e.g., (1) water or hydration molecules and other water (moisture) content of the sample bulk, (2) crystalline and amorphous contents where both occur, including mixtures of strongly and slightly polymerized materials, and (3) components of a mixed elastomer/polymer). Determining the type of curve that makes up the FID curve C is important because once known, the curves can be extended back to a time origin (shown as t0-1, i.e. excitation of a cycle 1) which is the theoretical center of the transmitted burst signal. This is important because there are saturation effects of the electronic components of the instrument that occur from the end of the burst signal to t1. During this time, no accurate measurements can be made, and yet the region of interest under the curve, which is a measure of the number of nuclei in the sample, runs from t0 to t4, and after that the curve becomes too small to be of any concern and the electronics require recovery time to prepare for the next cycle (starting with a Impulse with center at t0-2).

Each (sub)cycle continues to t5 to allow for recovery, i.e. essentially full relaxation of the sample protons - and only then begins a new transmit signal burst (t0-2). An excitation pulse interval usually lasts five to ten microseconds, the time t0 - t1 is five to fifteen microseconds (the shorter the better), t1-t2, where effects due to bound nuclei (Gaussian) are predominant, has a duration of five to fifteen microseconds (a critical measurement is taken in a narrower range tm); t2-t3 is an ambiguous transition region of fifteen to twenty-five microseconds duration, t3-t4 is a region where the unbound (exponential) component is dominant, and it has a duration of 300 to 500 microseconds during this part of the decay. The densest exponential curve is fitted to the C curve in the region t3 - t4 where the C and B curves are essentially equal, and this B curve is extrapolated back to t0, i.e. creating a curve B and its phantom B component and its B0 intercept at t0. This exponential curve is subtracted from the C curve in the region (t2-t3) where the A (Gaussian) curve is dominant. The resulting curve is then fitted to the least squares Gaussian curve. This best Gaussian curve is then extrapolated back to t0 to create A including its phantom A component and A0 intercept at t0.

The resulting data used in the computer 106 (Figs. 1 to 2) are the A-curve and the B-curve and ultimately their intercepts at t0 and their B0/A0+B0 ratio. Each of these curves (and their intercepts) has been experimentally and theoretically related to the same nuclei of interest, but with the group of nuclei giving the A-curve (Gaussian) arranged in a lattice structure bound. The nuclei that give rise to the B curve (exponential) are unbound or relatively unbound.

The data can be used as a QC type measurement or as an online control parameter fed back to control a process fed back into the IPL line (Fig. 1) or into associated components (e.g. when drying or baking a food product, when carrying out a continuous chemical or metallurgical reaction process, etc.).

The form of the system's input operating parameters may be wide-ranging and may include previously stored parameters in PROMs and ROMs or inputs entered via a telephone line and modem 110. The generation of the RF signal may be accomplished using many techniques, such as a coil or antenna array. The stationary magnetic field may be generated by electromagnets, permanent magnets, superconducting coil electromagnets, or other standard magnetic field generation techniques.

Fig. 4 is an expanded flow chart showing the measurement steps to achieve an effective industrial measurement. First, a single curve C of free induction decay is made to see if the sample area is clear (Quick FID), in an abbreviated cycle to make a curve C. If the sample area is not clear (N), then the measurement is interrupted so that valve V2 (again) opens and the area is cleared by operation of jets J and gravity. Another Quick FID step creates clearance. Then a sample is introduced by closing valve V2 and opening valve V1; probe P and line LI are adjusted as necessary to ensure that a sample is obtained. Jets J adjust and stabilize the new sample.

The temperature controllers 134-138 and 142-146 described above achieve very coarse and less coarse thermal control, thereby counteracting variations in the sample temperature.

An electronic reference for the signal processing device is established in 30-40 cycles (each with (+) and (-) subcycles in addition to (C+) and (C-) to detect and compensate for offset). Further adjustment is made by coils 24 to adjust HO and this is made possible by ten to twenty field test cycles of FID curve generation. The (C-) FID curve is subtracted from the (C+) FID curve, i.e. the absolute C values are added together to obtain a workable FID derivative - which has a maximum value at resonance. HO is adjusted via the coil current generator 122 and coils 124 until such a maximum value is achieved. These measurements are taken in a reliable range for such a purpose, i.e. the exponential range of t3-t4 [where the above reference measurements are also taken]. A sufficient field adjustment is usually achieved in less than seven cycles.

Then 50 to 500 cycles are performed to obtain a usable measurement. Each of these 50 to 500 cycles involves modulated FM transmit/receive/flash A/D conversion and storage of data. The curves are then averaged for curve fitting as well as to obtain the t0 intercept and the B0/A0+B0 ratio. Similar cycles, but somewhat shortened, can be used for quick FID, field testing and reference correction purposes. Each of the subcycles [(+) and (-)] of each such cycle involves the acquisition of thousands of FID points and the use of hundreds of such points in the data reduction.

Where multiple cycles are used for a single measurement, the amplitudes of (digitized) points on curve C are stored and a least-squares fit of such data points is made. Furthermore, plus and minus points are taken alternately to eliminate zeroing errors as noted above.

The area under a squaring derivative of the FID curve is integrated ("int.") and the area under a squared derivative of the desired exponential curve is integrated and a ratio is created whose square root is L/(KS+L), where L is a liquid (moisture) value derived from the exponential curve, K is a constant, and S is a quantity related to solid (or bound protons) with respect to the Gaussian curve, where KS+L is the total number of protons affected (and capable of being affected) by resonance.

It has also been discovered that with the attainment of higher accuracy and reliability, the composition of the sample tubing can distort readings. If glass is not used (and the use of glass is preferably avoided for industrial use), then hydrocarbon plastics should not be used instead. But fluorocarbons can be effective in several applications because signals from fluorine act off-resonance (with resonant conditions for hydrogen in humidity measurements) and can be distinguished from humidity-related readings with the sensitivity levels needed for such readings, and can be filtered (or distinguished) if necessary. In other cases of higher sensitivity measurements, e.g. for measuring relative proportions of amorphous and crystalline species in mixtures thereof, the sample container should be made of glass or a non-proton ceramic. In all such cases, the aim is to avoid sample containers with species that can couple with transmitted energy and produce an FID decay curve with a misreading of the samples. The present invention includes a system response to the detection of certain such cases that have not been recognized heretofore.

Fig. 5 shows another NMR decay waveform, with like components labeled the same as in the NMR decay waveform of Fig. 3 discussed above. In an operational modification of the method and apparatus of Figs. 1-2 (and in modification of their structure and control setting to effect such modified operation), each waveform is triggered by further pulses centered at mo' and mo" to cause Hahn spin echo responses to occur. The decaying portions of the echoes (hereinafter referred to as D and E) have the same general patterns as the original FID C, except that D and E, particularly the latter, are based primarily on a sample response without reaction of non-fatty solids and non-oil and non-solvent liquid components, while FID C is a measure of significant response to pulse excitation by all solid, moisture, oil/grease or solvent components.

The times tl', t2', t3', t4' and t1", t2", t3", t4" are defined for D and E respectively as rough analogue values of the points t1, t2, t3, t4 of the curve FID C discussed above. However, there are significantly different echo effects (due to a 180º rotation) compared to C (due to a 90º rotation). FID C has a duration of about one millisecond and each of the echoes has a Duration of usually about 2 milliseconds as controlled by selective timing of P', P" with respect to P; but these intervals can be set considerably higher or lower for different sample materials and different sets of instrument electronics specifications. Almost all of the time of C is occupied in the decay process, compared to the latter half of each of the echoes. The effective decay onset for D and E is t0' and t0" at the actual peaks of these curves (while t0 for C is set at the middle of the excitation pulse). The +C/D/E decay curves are alternated with -C/D/E subcycles for the same purposes as +C/-C subcycles in Fig. 3, as discussed above. The pulses P+/P- inducing C+ /C of FID cause a 90º (pi/2) rotation of protons, while the pulses P+, P', p"+, P"- for +D, - D, +E, -E cause a 180º (pi) rotation, respectively.

The block diagram of operations of Fig. 6 for such a modified system includes the same artifacts as described above in Fig. 4, i.e., the sample area freedom check by a Quick FID, several reference correction cycles and field checks (with simple P+, P- excitation) with subsequent adjustments, and then a measurement essentially according to the application of the Hahn spin echo described in connection with Fig. 5 above.

The signal processing of this measurement involves storing digital data points of the C and E decay curves, subtracting E from C, then processing an adjusted FID C curve essentially as described above in connection with Figs. 3 to 4, using time zero intercepts and/or integrals of Gaussian and exponential daughter curves derived from the adjusted FID C curve. In some cases, the peak of the E Decomposition is used to quantify the solvent or oil/fat component(s).

In addition, the data from FID C before and after adjustment by subtraction of E can be used as a determinant of tightness on an absolute or relative basis. Fig. 7 shows FIDs for polyethylene samples 7-1, 7-2, 7-3 with the highest, a lower and a lowest tightness. The curves differ sufficiently from each other that the differences can be detected by the signal processing means described above.

The corresponding t0" timing of the E peak itself is determined as a preset calibration factor for the instrument as applied to a particular type of material to be analyzed on-line over successive measurement days (or weeks or months). Figs. 9A, 9B, 9C show how this is done. Fig. 9A shows the C and E curves developed by running curves C, D, E of Fig. 5 with a roughly calculated t0" value under non-resonance coupling conditions (intentionally induced at coil 100, Figs. 1-2 compared to a resonance decay curve such as that shown in Fig. 9B) and superimposed. The difference of x-axis intercepts (delta t, determined by shifting the curves until the zero crossings of C and E coincide) yields a correction factor which is applied to the coarse t0" value to yield an adjusted t0"-a value relative to t0 for use in later runs at resonance of FID decay with Hahn spin echo (Fig. 9C). That is, in such later runs the digitizer/flash converter (105-1, Fig. 1) is opened at a time t0a" after time t0. The last peaks (LP) of C and E in Fig. 9 are also analyzed to obtain a Correction factor f, such as the ratio C/E, applied to subsequent E readings to compensate for attenuation of t1 compared to the value that would be obtained if t0" actually coincided with t0, rather than an arbitrary superposition. This is a valid correction because C at LP is entirely due to solvent or oil/grease effects [sic] and the measured E value is smaller than the measured C at LP only due to attenuation.

Furthermore, the intensity I parameter (e.g. voltage) is measured over a range that includes the nominal location t0 and averaged to obtain an average I value that can be used with higher reliability.

Finally, due consideration must be given to the relative amount of protons (or other NMR active species) per gram in each phase (e.g. solid, water, oil/fat) of interest. Appropriate calibration constants can be developed to determine concentrations of such phases or of components or other percentage data, e.g. crystallinity, in a mixture. For example, the fraction P of, say, oil/fat is determined by:

where Ke, Ka, Kb are calibration factors derived from standard materials, and Ao, Bo, Eo are the t0 intercepts after application of the above correction factors. The overall density (d) of a sample material can be determined either by a curve analysis as discussed above in connection with Figs. 7-1, 7-2, 7-3 or for plastics (and other materials) can be determined by the following equation:

where the fraction in parentheses is the crystallinity fraction M and N are empirically derived calibration factors (usually constants for the materials, which are always applied after an initial determination). Densities of the components would normally be additional factors in the equation; but for plastics, the crystalline and amorphous components have closely related fundamental densities, and density factors can be omitted (except where they are built into the M, N factors).

For other materials the equation would be:

where da, db and de are component densities. For some materials, statistical techniques can be applied to superimpose more complex and inherently more accurate non-linear functions on sets of data points (plots of density and/or fraction [or other parameter] determinations according to the method and apparatus of the present invention versus laboratory determinations of the same parameters for the same samples) and thus derive a calibration function with a second or third order fit. Subsequent determinations of the invention can be adjusted by such a calibration function.

Fig. 8 shows a calibration curve of such Polyethylene impermeability measurements by the present invention (y-axis) versus standard (e.g., oil bath flotation) impermeability measurements (x-axis) illustrating the effectiveness of the present invention. Other standard methods include acoustic permeability.

It will now be apparent to those skilled in the art that further embodiments, improvements, details and applications are possible in accordance with the above disclosure and within the scope of the present patent, which is limited only by the following claims prepared in accordance with patent law, including the doctrine of equivalents.

Claims (13)

1. Magnetic resonance system for industrial process monitoring, comprising: (a) means (P) for accessing successive samples from an industrial process, for placing the samples in a sample measuring area (51) and for removing successive samples from the area, (b) means (116) for applying a base magnetic field to the region to cause precession of sample nuclei therein and for applying a local resonance excitation pulse to the region to modify the precession, (c) means (100) for defining a receiving antenna coil and a signal conversion means, cooperating with the sample to cause a precession of nuclei to be measured and a relaxation which is detected at the coil as free induction decay, and (d) means (104, 106) for creating a digitised version of the free induction decay and analyzing it into bound (Gaussian) and unbound (exponential) components of the decay curve and extracting a ratio between the detection (intercept) of one of the components and the sum of the detections (intercepts) of the components, wherein the means (a) to (d) are arranged: (1) to stabilize base field application and excitation and conversion conditions in response to changes in one or more parameters of sample conditions in the measurement range; (2) to effect such measurement and ratio extraction and associated stabilization on a very rapid repeat basis of successive samples from the process; and (3) for converting at least one thousand data points in each decay curve and effecting a reduction in at least one hundred points thereof so that sample component proportions are reliably and repeatedly measurable from sample to sample in rapid succession for real-time on-line monitoring, the system further comprising: means for augmenting each of a plurality of decay curves with an echo refocus/decay technique to acquire data relating to at least one further subcomponent of the sample for increasing the accuracy of the ratio determination of (d) and/or for identifying the proportion of the at least one further subcomponent. 2. The system of claim 1, wherein the means for applying the base magnetic field comprises a Helmholtz coil (124) with coarse thermal stabilization of an outer sample processing region and intermediate stabilization of an inner sample receiving region (S2) located adjacent to pole faces of the coil, and wherein the means for stabilization comprises coarse thermal ambient coils for the sample region and fine adjustment via a change in the electromagnetic field to maintain resonance in the sample region. 3. A system according to claim 2, comprising means for effecting a further measurement to counteract a thermally induced deviation by adjusting the extracted decay signal. 4. The system of claim 3, wherein the means for adjusting the extracted decay signal comprises means for influencing the analysis of the free induction decay curve by a true integration of at least one functionally discrete component thereof. 5. A system according to claim 4, wherein the means for influencing the analysis comprises an extension of bound and unbound components of the decay curve to a reconstructed time-zero section that coincides with or is related to the time of the excitation leading to the decay. 6. System according to one of the preceding claims, comprising a sample container (98) which is substantially free of sample degradation-inducing ingredients in proportion to the sensitivity of the measurement to be performed. 7. The system of claim 6, wherein the sample container is a fluorocarbon tube, a ceramic tube, or a glass tube. 8. System according to one of the preceding claims, arranged for measuring moisture content or for detecting proportions of amorphous and crystalline components of mixtures thereof. 9. A system according to any one of claims 1 to 7, wherein the echo refocus/decay technique uses a Hahn spin echo. 10. The system of claim 9, wherein two Pi refocus pulses are applied to generate the Hahn spin echoes. 11. A system according to claim 9 or 10, comprising means for determining a time window for gating the spin echo detection at a correct zero-axis section that can be related to an original section of the free induction decay curve. 12. A system according to claim 11, comprising means for effecting such time window determination and for compensating for the attenuation of the spin echo response with respect to the initial free induction decay response by an off-resonance comparison. 13. System according to one of claims 1 to 12, arranged so that it outputs proportionality, crystallinity, humidity or density values.