MAGNETIC RESONANCE INSPECTION OF SYNTHETIC RUBBER This application claims the benefit of pending provisional application 60/150,659, filed August 25,1999.

BACKGROUND OF THE INVENTION A recent review confirms the routine use of laboratory-based nuclear magnetic resonance (NMR) experiments to study chemical composition, morphology and filler matrix interactions in elastomers. Unfortunately, the cost and complexity of state-of- the-art NMR instrumentation has precluded the use of NMR as an on-line modality for real-time process control. Many NMR experiments commonly used to probe polymersl would be a boon to process engineers if a similar analysis could be realized on-line at full production speeds.

SUMMARY OF THE INVENTION Within the past decade, magnetic resonance (MR) technology has become a proven tool for on-line analysis of industrial processes. Recent measurements suggest that devices capable of inspecting rubber bales (-30 kg) at full production speeds are viable. Analysis of MR relaxation behavior allows quantitation of the various compounds within a typical bale of rubber. It has been demonstrated that concentrations of water in butadiene-based rubbers were detectable at levels of less than 0.01% at a magnetic field strength of 0. 15T and nominal measurement time of 4 seconds. MR imaging methods may be employed to measure the distribution of dissolved species such as water and oil. The demonstrated performance of a potential

on-line MR analyzer for the rubber industry has been achieved through the use of permanent magnets, modular electrical components and intelligent automation software.

In this application, to demonstrate the efficacy of on-line NMR analysis, early experiments on production-sized bales of synthetic rubber are described. Process variables such as moisture and oil content are becoming increasingly important to compounders as the performance level of end products is continually raised. It has been shown that even the level of humidity of storage warehouses can affect the cure rate and scorch time of compounded rubbers. 2 The ability of manufacturers to supply (and store) material that meets strict quality requirements hinges upon their continuous knowledge or process parameters during the manufacturing process.

Other objects and advantages of the invention will be obvious and apparent from the specification.

REFERENCES 1. P. Blumler and B. Blumich, RUBBER CHEM. TECHNOL. 70,468 (1997).

2. J. Butler and P. K. Freakley, RUBBER CHEM. TECHNOL. 65,374 (1992).

3. T. W. Schenz, B. Dauber, C. Nichols, C. Gardner, V. A. Scott, S. P. Roberts and M. J.

Hennessy,"Online Magnetic Resonance Imaging for Detection of Spoilage in Finished Packages,"in"Advanced in Magnetic Resonance in Food Science,"P. S.

Belton, B. P. Hills and G. A. Webb Eds., Royal Society Press, Cambridge, 1999.

4. E. L. Hahn, Phys. Rev. 80,580 (1950).

5. R. R. Ernst, G. Bodenhausen, and A. Wokaun,"Principles of Nuclear Magnetic Resonance in One and Two Dimensions,"Clarendon Press, Oxford, 1987.

6. H. Y. Carr and E. M. Purcell, Phys. Rev. 94,630 (1954).

7. S. Meiboom and D. Gill, Rev. Sci. Instrum. 29,688 (1958).

The invention accordingly comprises the several steps in the relation of one or more of such steps with respect to each of the others, which will be exemplified in the method hereinafter disclosed, and the scope of the invention will be indicated in the claims.

BRIEF DESCRIPTION OF THE DRAWINGS For a fuller understanding of the invention, reference is had to the following description taken in connection with the accompanying drawings, in which: Figure 1 is a photograph of a sample of IBR sliced from a production bale; Figure 2 is a photograph of an NMR image of the sample of Figure 1; Figure 3 is a photograph of a"wet"SBR sample; Figure 4 is a photographic NMR image of SBR with greater than 10% aromatic oil content; Figure 5 is a graph illustrating a one-dimensional NMR profile of SBR; Figure 6 is Table I of spin-spin relaxation values of synthetic rubbers; Figure 7 is Table II of spin-spin relaxation values of SBR following storage; Figure 8 is MR images of"wet"rubber sample; Figure 9 is MR images of"dry"rubber sample; Figure 10 is MR images of"high oil"rubber sample; Figure 11 is Table III presenting T2 measurement data on rubber samples; Figure 12 is T2 data acquired on a rubber bale with a single spin-echo sequence; Figure 13 illustrates spin-echo components of"wet"SBR; Figure 14 is Table IV of T2 values of major components in"wet"rubber samples;

Figure 15 is MR images of a large,"wet"IBR samples; Figure 16 are front and back photographs of the sample of Figure 15; Figure 17 shows the"wet"sample in the test set up; Figure 18 shows the large IBR sample in the test set up; Figure 19 show one-dimensional MR profiles of"wet"rubber; Figure 20 is Table V of one-dimensional MR profile data for"wet"rubber samples; Figure 21 is Table VI of moisture content of rubber samples; Figure 22 is a graph of sample proton density based on NMR single amplitude at time = 0; Figure 23 is Table VII of rubber sample properties; Figure 24 is a graph of normalized BR signal magnitudes with errors; Figure 25 is a correlation plot of MR spin-echo ratios and percent volatiles; Figure 26 illustrates a spin-echo NMR sequence; and Figure 27 illustrates a spin-echo NMR imaging sequence.

DESCRIPTION OF PREFERRED EMBODIMENTS EXPERIMENTAL Production samples of synthetic rubbers were obtained from several major U. S. manufacturers. Full sized bales (>30 kg) of BR were analyzed for moisture content.

Specific, smaller portions of styrene-butadiene rubber (SBR) and isoprene-butadiene rubber (IBR) were analyzed for moisture or oil content.

The key consideration in this work was the ability to probe a complete, commodity-sized bale of rubber (>30 kg) as the sample. This large sample precludes the use of high-field, superconducting NMR instruments commonly found in polymer research laboratories.

However, NMR imaging systems, like those currently used in hospitals (usually referred to as Magnetic Resonance Imaging or MRI), are large enough to accept samples of this size without rigorous modifications. Intermagnetics General Corporation, an assignee of this application, a leading supplier of magnets and other components for clinical imaging, produces an NMR imaging system which has already found use on-line as a quality assurance instrument in the food industry. 3 The permanent magnet operates at 0.15 Tesla (6.2-MHz 1 H frequency) and does not require power or cryogens to maintain stability. Samples of rubber were positioned in the magnet on a motorized bed, whereas industrial systems in the food industry utilize fully automated non-metallic conveyors. 3 A selection of radio frequency (RF) signal transmission and detection antennae (RF coils) were used depending upon the size of the sample at hand. For full bales, a"body coil"designed to accommodate the torso of an average sized human was used. Smaller samples were studied using three cylindrical coils measuring 292 mm long by 102,152 and 259 mm in diameter. The coil giving the best filling factor was chosen based upon sample size and geometry.

Conventional Hahn spin echo experiments 4 with echo times ranging from 0.1 ms to 5 s were used to determine spin-spin relaxation times of components dissolved in the smaller rubber samples. Square 90° and 180° RF pulses of widths < 100 lis were used for the T2 measurements carried out in the cylindrical RF coils. To maximize efficiency, the cylindrical coils were used for both receiving and transmitting. RF signals.

NMR images were acquired using a conventional spin echo sequence employing a single read-out gradient applied along the frequency-encoding axis during

signal digitization. Images were obtained using both the smaller coils and the body coil. The body coil was implemented as a receive only coil in conjunction with a larger transmit-only coil which is built into the magnet. The echo and acquisition delay times were varied and are reported below for each image. Single axis NMR profiles were obtained using a spin echo imaging sequence with the phase encode gradient disabled.

In all images and profiles, a 1.5 ms slice selective RF pulse (sync shaped) was used.

Signal averaging was accomplished in all experiments with phase cycling to remove DC offset and other artifacts.

RESULTS AND DISCUSSION A. RELAXATION MEASUREMENTS Following excitation by a preparative RF pulse, polarization within a group of spins can be lost by interactions with the surroundings via numerous pathways. These relaxation phenomena are well understood in the art. The detection of dissolved components within a solid matrix may be performed by analysis of the spin-spin relaxation, or T2 behavior of the sample. T2 relaxation pathways are generally impeded by molecular interactions and lead to larger T2 values in samples with increased molecular mobility.

In a sample containing multiple, independent components, each with correlation times significantly different from the others, the relative concentrations of each component may be quantified by a mathematical decomposition of NMR signal lifetimes and amplitudes. Hahn echo experiments may be used to monitor the spin-spin relaxation behavior of a sample in the presence of magnetic field inhomogeneities.

This is especially useful when dealing with large sample volumes required for production sized bales of synthetic rubber. Data collected from a series of Hahn echo experiments, where each echo is acquired in separate experiments or serially via a CPMG experiment, 6, 7 can be fit to a collection of exponential decay functions of the type S = Ae (-t/K) where S is the NMR signal amplitude at time t; A is the zero-time amplitude of the NMR signal, and K is the decay time constant, or T2.

The signal from each component of the sample will follow exponential decay with a decay time specific to the component's mobility and amplitude proportional to the components concentration. T2 values reported in this work were calculated by fitting a series of single Hahn echoes to multiple exponential decay functions of the type S=Ale (-t/Kl) + A2e (-t/K2) +A3e (-t/K3) +ete. where the number of terms is equal to the number of components in the sample.

Table I (Figure 6) reports the T2 values between 1 and 1000 ms for the rubber samples studied in this work. The relative intensities reported in column four are calculated from the zero-time amplitudes for each component.

Manufacturers have exercised the practice of qualitatively determining moisture content in bales on the production line by measuring the relative number and size of white spots visible to the eye in and on the bale. In most cases, the bales must be allowed to cool and cold flow for a few hours or days before the moisture spots are visible. Correlating the visible spots with NMR data was a preliminary goal in this work.

The IBR sample shown in Figure 1 contained numerous white spots measuring 10 to 50 mm in diameter. Figure 1 is a photograph of a 10 kg sample of IBR sliced from a production bale. Many"spots"of moisture are visible on the surface. The sample measured 400 mm wide, 275 mm in height and 100 mm thick. The T2 data for this sample reported in Table I reports that two components were detected in the sample. The 700 ms component was assumed to be due to the moisture spots, but could not be verified until the imaging experiments reported below were completed.

Three samples of SBR were tested: wet, dry and containing >10% aromatic oil.

The SBR samples containing moisture and oil were analyzed for spin-spin relaxation behavior and compared against a dry control. The wet sample arrived with light spots visible on the surface and the oil-bearing sample was black and opaque. Three distinct T2 values were observed in the wet sample versus two for the dry sample. The two faster components in the wet sample were similar in value and concentration to the components in the dry sample, however the third component possessed a considerably longer T2 (400 ms) than the others. This suggested that the longer T2 component may be due to moisture.

Three full-sized bales of BR were supplied with the manufacturer's estimate of moisture content based solely on appearance. Correlations between the qualitative assignment of moisture content in the bales of BR by the manufacture and NMR data were moderately conclusive. The bale considered"evenly wet"based on an overall lighter color of the bale tested higher in concentration of the longer, 100 ms component.

The"dry"and"spotted"bales were of similar color, and not significantly different on the basis of multiple Hahn echo experiments.

Evidence for the assignment of the 400 ms component in the wet SBR sample to moisture was provided one month later following storage of the samples in a low humidity (&lt;15% relative) environment. Table II (Figure 7) reports the T2s of the three SBR samples at low humidity one month following the initial analysis. Following dry storage, both the wet and dry samples were expected to reach an equivalent state of hydration. All SBR samples gained concentration in the shortest T2 component suggesting loss of intensity in the longer component (s). Furthermore, the 400 ms component previously detectable in the wet sample, as well as the visible spots, had vanished following the month of dry storage. In lieu of samples with known amounts of moisture and oil, the assignments of the remaining components may be hypothesized as follows: residual moisture and oil in SBR may be attributed to the components greater than 10 ms whereas the components shorter than 1 ms are likely due to the polymer backbone.

B. NMR IMAGING The correlation of visible moisture spots with NMR data is most easily accomplished by spin-echo imaging of the rubber samples. Figure 2 shows a representative image obtained from the 10 kg sample of IBR. Figure 2 is an NMR image of the 10 kg sample of IBR pictured in Figure 1. The light areas (spots) in the image correlate with the white spots visible on the surface of the sample. Imaging parameters: TE=19 ms, TR=500 ms, 128x128 pixels, 3.28 mm/pixel scale, 256 averages, slice thickness = 250 mm. The parameters of the imaging sequence were chosen to be sensitive to components with T2 values greater than 10 ms, however components with longer T2s appear with greater intensity. A series of images acquired with echo times ranging from 19 tol60 ms (not shown here) clearly indicate that the large white spots visible on the sample correlate with the 400 ms component (moisture) detected in the previous T2 experiments.

An image obtained from the wet SBR samples is shown in Figure 3. Figure 3 is an NMR image of a"wet"SBR sample. Moisture is not evenly dispersed in the sample as evidenced by the irregular distribution of bright regions (spots). Imaging parameters: TE=31 ms, TR=500 ms, 128x128 pixels, 1.56 mm/pixel scale, 64 averages, slice thickness = 10mm. The non-uniform distribution of signal intensity suggests a similar distribution of moisture within the sample. Images obtained from the dry samples under the same experimental conditions (not shown) show barely enough signal to outline the shape of the sample. Comparison of the image in Figure 3 with an image obtained from the oil-incorporated sample in Figure 4 reveals a marked difference in signal uniformity. Figure 4 is an NMR image of SBR with > 10%

aromatic oil content. The uniform distribution of signal in the image suggests an even dispersion of oil in the sample. Imaging parameters: TE=31 ms, TR=500 ms, 128x128 pixels, 1.56 mm/pixel scale, 64 averages, slick thickness=lOmm The homogeneous blending of oil into the SBR is apparent in Figure 4 confirming NMR imaging's utility in determining the dispersion of liquid components within synthetic rubbers.

Images obtained on the bales of BR (not shown) yielded similar results: the spots visible on the surface were seen as bright regions in the images, however the bright regions were only detected on the surface, and not within the thickness of the bale.

The Figures 1-4 based on digital images that were filed in the provisional application that is relied on for priority in this case and is incorporated in its entirety herein, including its appendices.

C. ONE-DIMENSIONAL NMR PROFILES The two dimensional images of rubber in Figs. 2-4 clearly show the sensitivity of NMR to moisture in bale-sized samples of synthetic rubber. Unfortunately images are impractical in an industrial environment because of the amount of time required to collect enough 2D spatially resolved signal-on the order of minutes. A successful process analyzer must collect its data at process speeds and in a rubber manufacturing plant this constraints is on the order of a few seconds. Bulk measurements of moisture content may be obtained by Hahn echo experiments in as little as a few milliseconds,

but detection of localized areas of moisture, which lead to formation of visible spots, must be done with a spatially sensitive technique.

Single-axis, one dimensional NMR profiles strike a useful compromise between imaging and bulk measurements by allowing the spatial reporting of signal along one axis of the sample in an experiment time of just a few seconds at 6.2-MHz.

Profiles obtained from the wet and dry SBR samples are shown in Figure 5. Illustrated are one-dimensional NMR profiles of SBR. The profiles are, in order of increasing average intensity, dry, wet and oil-impregnated. The wet sample profile shows a non- uniform distribution of moisture. The dry sample profile has been increased in scale for visibility. The intensity profile for the dry sample shows significantly more uniformity than the profile for the wet sample.

Quantitation of moisture in SBR was accomplished by the simultaneous profiling of the rubber sample and a standard vial, containing a known amount of water. The T2 of the water standard was adjusted with copper sulfate to approximate the 400 ms T2 of the moisture in the rubber. Integration of the areas under the rubber and standard portions of the profile allowed direct calculation of the moisture content in the SBR. Following this method, the weight percent of moisture in a sample of SBR can be determined to 0.01% SUMMARY Preliminary NMR investigations have shown that moisture defects or high oil content in SBR, BR and IBR can be detected by an industrially ruggedized imaging instrument. The most rapid, bulk measurements may be obtained by conventional spin- echo analysis. Imaging experiments may be used to locate areas of high moisture or

water, but are inefficient for process control at full production speeds. One dimensional MR profile experiments are effective for quickly determining the distribution of moisture or oil in a full bale.

Whereas the above description speaks of rubber manufacturing, it should be understood that the inspection techniques of the present invention are applicable to evaluation of other non-magnetic materials during on-line operations to detect voids and abnormal components, desirable, undesirable, or of no consequence.

The following Appendices A-D accompanied the provisional application that was filed in the United States Patent and Trademark Office on August 25,1999, Application No. 60/150,659.

Appendix A-Results from Preliminary Experiments-Rubber Analysis by MR The first group of samples was comprised of"wet","dry"and"high oil"content rubber, and one full sized"dry"bale. The small dry sample consisted of two small blocks cut from the same original bale. The three smaller samples were approximately the same size, about 1/5 the size of a whole bale. Notable observations made during the unpacking of the samples included the presence of whitish spots on the wet sample and the deep black color of the high oil sample. Other than the black sample, all samples were a light yellow, semi-translucent color, not unlike dried rubber cement. Once unpacked from the shipping container, all samples were stored inside the imaging suite, within their own plastic bags, at 70 °F and 20% humidity.

Imaging.

The first experiments were to obtain good quality images from the wet sample to determine if the irregular water inclusions could be detected by MR imaging. Figure 1 reports example images obtained using 1282 points, 64 averages, 500 ms TR, 31 ms TE,

10 mm slice @ 15 mm spacing, and 200 mm FOV. The SPC-10 RF coil was used in TX/RX mode. No noticeable coil loading was observed on any sample.

Figure 8 presents MR images of"wet"rubber sample. Exponential filtering value of 2 was used to suppress noise. The images in Figure 8 indicate a non-uniform distribution of a liquid component in the bale, presumably water. For comparison, an identical experiment was performed on the dry samples using the same acquisition parameters and RF coil. Figure 9, reports the images obtained using the same reconstruction parameters and pixel intensity scaling values used in Figure 8.

Figure 9 shows MR images of"dry"rubber sample. Identical acquisition and reconstruction parameters were used in both Figures 1 and 2. The ghostly images of the dry sample in Figure 9 can just barely be seen above the noise. Since the same intensity scales were used in both Figures 8 and 9, the dry samples must not contain the liquid component present in the wet sample. This strongly supports the selection of water as the component present in the wet sample.

Figure 10 shows MR images of"high oil"rubber sample. Intensity is uniformly distributed through sample. Figure 10 reports images acquired on the high oil sample according to the same experiment used in the previous figures. The relatively high S/N is likely due to the oil content of the sample and the fact that it is uniform and not localized, as in Figure 8 indicates that the oil is evenly distributed through the sample.

The dark fissures seen along the top edge is due to a cut in the sample.

The images obtained on the rubber samples suggest that water and oil can be detected spatially within a rubber bale.

Relaxation measurements.

Using a single spin-echo sequence and nine different echo times ranging from 0.25 to 25 ms, data was obtained that suggests at least two separate components in the dry and high oil samples, and three component decay in the wet sample. Table III (Fig. 11) reports the component T2 values and their relative magnitudes in the three samples.

The data, acquired with a single spin-echo sequence, was processed by taking the highest magnitude point in the echo, and plotting it versus the echo time as shown in Figure 12.

From this data, double and triple exponential decay functions of the type f (x) = a * e (~ were fit using a least linear squares line-fitting software package (Figure 13). The errors reported in Table III were taken from the fit results and are based on the quality of the fit. For all fits the regression value R2 was greater than 0.999 indicating exceptionally good modeling of the data by the functions. It should be noted that single exponential fits gave poor results (R2 < 0.98) on all samples, and a double exponential fit on the wet sample was also poor. The triple exponential fit on the wet sample was nearly perfect: R = 1.0000. The moisture in the wet sample may be responsible for the long decay component, 70 ms, not seen in the other samples. This can be shown graphically in Figure 12 where the wet sample decay curve seems to level off and cross the other two curves. Longer echo times show this decaying to zero after 200 ms, but echo times greater than 60 ms were not possible with the"rapid"spin-echo pulse sequence.

Appendix B-Internal Progress Report Numerous samples of commodity grade rubber and production intermediates were obtained from various vendors. The samples were analyzed to determine the MR

behavior of rubber moisture content, oil content and impurities. Determining moisture in blocks of styrenebutadiene (SBR) and isoprene-butadiene (IBR) to levels of 0.01% was possible. Moisture content was estimated for the three"wet"samples.

Measurement of rubber oil content should also be feasible to similar levels of accuracy.

Images of bales containing moisture show localized regions of intense signal suggesting a non-uniform dispersion of moisture.

Quantification of moisture in rubber bales was accomplished through the use of one- dimensional MR profile experiments. A water standard was attached to the sample for calibration of signal integrals. The first step was to measure T2 of the water component in each rubber sample so that a water standard with matching T2 could be prepared.

Nine spin-echo times ranging from 10 to 5000 ms were used to calculate rubber relaxation times. Table IV reports the major T2 component values of the"wet" samples. The spin echo data acquired on the wet samples could only be suitably fit by a double-exponential decay function. The T2 values calculated from these fits are similar; each sample contains a fast component on the order of 10s of milliseconds and a slower component about 20x greater.

The component with the longest T2 was assumed to be water. This assumption is in agreement with the T2-weighted images in Figure 15 which show correlation between the visible"white spots" (Figure 16) and regions of high signal intensity. The images in Figure 15 were acquired under different TE and TR times. Figure 15 shows MR images of the large, wet IBR sample. TE/TR values in milliseconds are listed in each image. Intense images of water with good separation from rubber background are found at long TEs and TRs. Note how signal intensity of the"spots"is greatest when

TR is greatest. Also, increased TE times leads to better definition between the spots and the rubber"background".

Figure 16 shows front and back photographs of the large, wet IBR sample. The light areas on the block correlate with the bright areas imaged in Figure 15.

Based on the images in Figures 15 and 16, the longer T2 component in each wet sample was assumed to be due to moisture. A sample of water with a single T2 value similar to the moisture component in the rubber samples was prepared. A copper sulfate concentration of 0.6 mMol has a T2 of about 850 ms. A sample of this solution was placed in a small vial to be used as the internal standard in the following profile experiments.

One dimensional profiles were obtained by a standard MRI pulse sequence with phase encoding and slice selection gradients disabled. The sample under test and the internal standard were placed inside the IMIG-MRI body coil as shown in Figures 17 and 18.

This coil had sufficient room to accommodate all the rubber samples we received.

Depending upon the geometry of the sample under test, either a horizontal or vertical read encoding gradient was used on a transverse-slice experiment. Acquisition parameters for all samples were: 2.00 KHz bandwidth, 256 data points, 1 view, 8 averages, 500 mm FOV, 3s TR and 400 ms TE.

Figure 17 shows layout of SBR wet rubber sample and standard vial in the IMIG-MRI body coil. The small vial is to the right of the rubber sample. Figure 18 shows the large IBR sample tightly fit the IMIG-MRI body coil. The standard sample is laying on top of the bottom foam block.

The profiles were integrated in the Specana spectroscopy software package from SMIS.

The method of integration was checked by analyzing the profile obtained from four

vials containing known amounts of doped water. Integration of signal magnitudes was linearly dependent upon the amount of solution contained in the vials.

Representative one-dimensional profiles of the IBR rubber samples are shown in Figure 19. The standard sample was kept separate from the rubber so that baseline could be registered between the two regions of interest. Each sample was integrated by simply summing the magnitude values within each region. The Specana program has a useful utility for performing this function on one-dimensional data. In Figure 19, the MR profiles are of"wet"IBR samples. The sharp peak on right is the vial of standard sample.

Table V (Figure 20) reports the measured intensities of these integrals of the various samples as well as respective T2s, experimental echo times, and calculated proton densities and water contents. Numbers with an asterisk in Table V (Figure 20) are calculated, all others are known. Proton density was calculated according to: D = I exp (TE/T2) and is shown graphically in Figure 22. Where D is proton density, I is integral intensity, TE is the echo time in milliseconds and T2 is the T2 of the component under investigation. The TE of the experiment was chosen to be 20 times greater than the next faster T2 component in the sample. In Figure 22, proton density, or NMR signal amplitude at time = 0 seconds, is shown by a broken line as the reverse extrapolation of the exponential decay function.

Rubber moisture content was calculated by multiplying the ratio of proton densities in each experiment by the amount of standard. Table VI (Figure 21) reports the calculated moisture contents for each"wet"sample.

From the S/N of the profiles in Figure 19, the detection limit of water was estimated to be about one gram in a 2 scan (10 second) experiment. For a full bale, this moisture would be contained in a rubber mass weighing about 11 kg. Therefore, the moisture content would be 0.000088, or about 0.01%.

Following seven weeks of storage in dry (<30% relative humidity) conditions, the "wet"and"dry"SBR samples lost further moisture and reached an equivalent state of hydration. Table VII (Figure 23) reports the T2S of the SBR samples following dry storage. The"dry"and"wet"samples are now indistinguishable. The oil-impregnated sample has lost intensity in the longer, 8 m s component.

Appendix C-Large Polybutadiene Data Set Study Fifteen samples of BR were obtained for moisture/volatile calibration analysis. The samples weighed 250g + 20 g and were analyzed by Hahn echo experiments. Using a small SP-4 RF coil in TX-RX mode, 180° pulse times less than 50 microseconds long were available. Echos were recorded for each sample from 100 ps to 5s. To differentiate between each sample (presumably on the basis of moisture content), Spin- echo ratios were used. The NMR echo amplitude acquired at 5 ms was divided by the echo signal amplitude acquired at 60 ms. This ratio was used to correlate the amount of moisture. Following MR analysis, the samples were returned for moisture/volatile analysis using a heated mill. Data shown in Figure 24 show the relative spin-echo

ratios between each sample standard deviations calculated from three separate MR analyses.

Figure 25 reports the correlation between our spin-echo ratio MR method and the accepted method for determining"volatiles". The MR signal correlates moderately well with the percent of volatiles in the sample.

Appendix D-NMR Pulse Sequences Figure 26 illustrates the spin-echo NMR sequence. The echo time (TE) is equal to two times the delay te2. This sequence was used to acquire data for spin-spin relaxation calculations. The spin-echo ratios calculated in Appendix C utilized this sequence for data acquisition.

Figure 27 illustrates the spin-echo NMR imaging sequence. This sequence was used to acquire all images. NMR Profiles were acquired using this sequence, but with the phase gradient deactivated.

It will thus be seen that the objects set forth above, among those made apparent from the preceding description, are efficiently obtained and, since certain changes may be made in carrying out the above method without departing from the spirit and scope of the invention, it is intended that all matter contained in the above description shall be interpreted as illustrative and not in a limiting sense.

It is also to be understood that the following claims are intended to cover all of the generic and specific features of the invention herein described and all statements of the scope of the invention which might be said to fall therebetween.