REVIEW OF SCIENTIFIC INSTRUMENTS 85, 045109 (2014)
Enhanced reaction rates in NDP analysis with neutron scattering R. Gregory Downinga) National Institute of Standards and Technology, Chemical Sciences Division, Gaithersburg, Maryland 20899, USA
(Received 27 February 2014; accepted 23 March 2014; published online 14 April 2014) Neutron depth profiling (NDP) makes accessible quantitative information on a few isotopic concentration profiles ranging from the surface into the sample a few micrometers. Because the candidate analytes for NDP are few, there is little interference encountered. Furthermore, neutrons have no charge so mixed chemical states in the sample are of no direct concern. There are a few nuclides that exhibit large probabilities for neutron scattering. The effect of neutron scattering on NDP measurements has not previously been evaluated as a basis for either enhancing the reaction rates or as a source of measurement error. Hydrogen is a common element exhibiting large neutron scattering probability found in or around sample volumes being analyzed by NDP. A systematic study was conducted to determine the degree of signal change when neutron scattering occurs during analysis. The relative signal perturbation was evaluated for materials of varied neutron scattering probability, concentration, total mass, and geometry. Signal enhancements up to 50% are observed when the hydrogen density is high and in close proximity to the region of analysis with neutron beams of sub thermal energies. Greater signal enhancements for the same neutron number density are reported for thermal neutron beams. Even adhesive tape used to position the sample produces a measureable signal enhancement. Because of the shallow volume, negligible distortion of the NDP measured profile shape is encountered from neutron scattering. [http://dx.doi.org/10.1063/1.4870414] INTRODUCTION
Neutron beams make possible the extraction of quantitative information about composition and the elemental distribution in materials, often nondestructively. The neutroninduced radiation is used to uniquely characterize the sample or corroborates measurements obtained through other analytical techniques. A major disadvantage with neutron beam techniques is that available neutron fluence rates are notably less than the particle fluence rates accessible by even modest laboratory probes based upon electrons, protons, and photons beams. As a result, data acquisition by neutron techniques is often protracted in comparison. Signal yield from neutron interactions and detection efficiency are improved by a variety of methods to make better use of the available neutrons. Detector positioning, neutron optics, improved electronics, and background mitigation are all critical experiment design parameters. Techniques like neutron depth profiling (NDP)1, 2 and prompt gamma activation analysis (PGAA)3 benefit from these tactics. In a typical analysis scenario the samples are inserted into a highly directed stream of neutrons, the beam passes through the sample substantially unperturbed because of small sample mass and limited neutron interaction cross section. If, however, the sample itself or the in-beam environment adjacent to the sample contains a strong neutron scatterer (e.g., H, C, . . . ), then a substantial fraction of the neutrons is scattered into 4π geometry space. Two changes occur to a fraction of those scattered neutrons: (1) the path length of the neutron is lengthened or
a) [email protected]
0034-6748/2014/85(4)/045109/5/$30.00
shortened4 through the sample and (2) the energy of the neutron may be altered. Previously, Mackey et al.,5 Lindstrom et al.,6 Thompson et al.,7 and references within these papers described how neutron scattering in hydrogenous materials alters the effective reaction rate and the background radiation. They have demonstrated that variables such as sample composition, geometric shape of the sample, and shifts in the neutron energy by way of neutron scattering will increase or decrease the bulk reaction rate compared to samples without significant scattering properties. This realization has necessitated tight control on how samples are prepared and then oriented with respect to the neutron beam for quantitative activation analysis. If not managed, the effects bias the concentration determination and render unreliable the attempt for analytical results. This is especially true for sample analysis in cold neutron beams with the increased scattering cross section and the consequential upshift in neutron energy for neutrons that scatter. Conversely, if hydrogenous or other materials with a significant neutron scattering cross section are systematically controlled, the increase in reaction rates from neutron scattering could provide benefit and improve detection limits or reduce analysis time. Neutron scattering is more easily exploited by the NDP technique where the depth of analysis is not a bulk sample response as with PGAA, but rather is limited to depths of less than 100 μm relative to the sample surface. More typical are the cases where the depth of NDP analysis is less than 20 μm. This also implies that the angle of the sample in the beam is of little concern, which is not the case for PGAA. In this work, sample environments were varied stepwise to study the change in reaction rate for NDP measurements and evaluate potential measurement errors.
85, 045109-1
045109-2
R. Gregory Downing
Rev. Sci. Instrum. 85, 045109 (2014)
EXPERIMENTAL
Measurements were conducted in the NDP instrument1 on the Neutron Guide 1 (NG1) end station at the NIST Center for Neutron Research (NCNR). Since 2012, the neutron spectrum at the sample plane peaks with a wavelength of ≈0.4 nm (5.2 meV) and an upper energy cutoff regulated by 2 liquid-nitrogen cooled Be-Bi filters.8 The thermal-energy equivalent beam fluence rate was determined to be greater than 1 × 109 n/cm2 /s.9 Prior to 2012 the neutron optics at this end station were optimized for a small angle neutron scattering instrument that restricted the available neutron fluence rate to 7.6 × 108 n/cm2 /s (thermal-energy equivalent). NDP measurements presented here were conducted during both time periods but are normalized to a common sample; therefore, variations in the neutron fluence rate are not a complicating factor. The full-width-half-maximum diameter of the neutron beam at the sample plane is ≈12 mm. Two samples, previously characterized for 10 B content and depth distribution, served as the control materials throughout this study. The samples consisted of a silicon wafer substrate of 475 μm or 300 μm thickness. One face of each sample is coated with ≈100 nm of borosilicate glass (BSG). Consequently, a NDP charged particle energy spectrum distinctly resolves all four energy peaks predicted from the 10 B(n,α)7 Li reaction (Figure 1). The 10 B content of the BSG layer shown in Figure 1, obtained from the 300 μm sample, is 5.22 ± 0.03 × 1015 atoms/cm2 . The uncertainty is a 95% confidence tolerance interval including random measurement error and systematic uncertainties. The sample area analyzed for each measurement was defined by a Teflon sheet containing an aperture ≈10 mm in diameter. Control samples completely filled the aperture and were entirely illuminated by the neutron beam during irradiation. The Teflon sheet was 0.5 mm thick; thus, reaction particles not originating within the aperture were unable to penetrate the sheet but rather are absorbed by the Teflon. The sheet was centered over a 0.8 mm thick, 150 mm diameter aluminum disk. A large diameter hole was cut at the center 2000
10B(n,α)7Li
Particle Count
1500
1000
500
0 500
1000
1500
2000
2500
3000
Channel Number FIG. 1. The 4 particle energy peaks measured for a control sample using a multichannel analyzer during NDP analysis of the 10 B(n,α)7 Li reaction. Only the peak area centered about channel 2100 was used in this work.
FIG. 2. Two perspective views (not to scale) of an aluminum disk mount used to reproducibly aperture and position the sample in the NDP chamber. The upper figure is from the perspective of approaching neutrons. The bottom view is an edge-on perspective of the sample mount showing the configuration of various components. False color legend: Teflon mask (aqua), sample (red), mounting ring (grey), and neutron scattering material (blue).
of the Al disk. The entire assembly is depicted in Figure 2. Also shown is the shallow notch cut into the outer rim of Al disk to ensure reproducible mounting and orientation of the holder for each measurement. Both the sample and the NDP chamber environment remained at room temperature. A transmission-type silicon surface-barrier detector was positioned at a distance of ≈120 mm from the sample. The detector face is circular with an active area of 50 mm2 concentric and parallel to the sample surface defined by the Teflon aperture. The thinness of the transmission-type detector (≈40 μm) significantly reduces background noise in the spectrum arising from ambient prompt gamma and x-ray radiation present in the NDP chamber during analysis. Radiation and electronic background noise is seen increasing in intensity at the lower energy region of the spectrum, but it does not contribute counts in the two right-most (higher energy) alpha peak areas (Figure 1). A neutron monitor operates concurrently with each sample spectrum acquisition to account for any small variations in neutron flux between runs. Each NDP spectrum was normalized to the respective integrated neutron flux monitor count. The consistent use of long sample counts reduced the statistical error of neutron flux correction to about 0.2% ± 0.1% uncertainty. The first set of charged-particle spectra were collected by attaching hydrogenous neutron scatterers of known thicknesses and masses in the configuration shown in Figure 2. Measurements were taken with scatterers in an arbitrary thickness order. Individual disk-shaped scatterers were 40 mm–50 mm diameter and ≈0.49 mm thick. The larger diameter pieces were used in earlier measurements and the 40 mm diameter pieces were used in the most recent measurements. They stacked immediately behind (downstream) and centered flat against the control sample. High-purity Nylon-6, a semi-crystalline polyamide solid, was selected for the scatterer based on properties of high hydrogen density, low neutron absorption, and mechanical properties. The Nylon-6 has a molecular composition of C6 H11 ON with
R. Gregory Downing
an average bulk density of 1.12 g/cm3 . From the formula weight of 113.16 g/mole, the hydrogen density is calculated as ≈0.11 g/cm3 . Bound hydrogen has a nominal scattering cross section of ≈80 barns for the neutron energy spectrum delivered to the NDP instrument. A second set of NDP measurements was collected using Teflon disks in the same geometry as with the previous Nylon experiment. Teflon has various formulations but do not contain hydrogen. However, fluorine has a nominal 4 barn scattering cross section. The Teflon sheets were ≈0.49 mm thick. Again, individual disks 25 mm in diameter were stacked in known but random thicknesses centered behind the control sample. Spectra were collected to detect correlations in reaction rate to the applied mass. In addition, masses of other materials including paraffin, low density polyethylene (LDPE), adhesive tape, silicon, and a lithium rich polymer were evaluated as neutron scatterers in geometrically equivalent NDP experiments. A third dataset was collected by positioning a single 50 mm diameter, 8 mm thick disk of Nylon-6 downstream of the sample between spectrum acquisition. The spectra were taken in a sequence of increasing distances to evaluate the geometry of scatterer upon the sample reaction rate. Similar to the previous datasets, the scatterer was kept aligned perpendicular to the neutron beam and concentric with sample center. In a final measurement set, the series of Nylon-6 disc thicknesses, described above, were centered and attached to the same side of the sample as the approaching neutron beam. That is, the scatterer intercepted the neutron beam and the neutrons must pass through the scatterer prior to reaching the sample. In this scenario the sample holder (Figure 2) was simply reversed 180◦ and mounted in the NDP chamber so the BSG side of the control sample faced an antipode detector. The detector is equal in distance from the sample as the upstream detector in the previous three sets of measurements. The spectra obtained from all four measurement sets are described below. RESULTS
As downstream masses of hydrogenous material adjacent to the thin control sample were increased, the reaction rate of 10 B(n,α)7 Li increased rapidly first, then slowed and asymptotically approached a maximum enhancement value. The change in reaction rate is plotted in Figure 3 as a function of scatterer thickness for various materials. These data are corrected where necessary for spectral background noise, variations in count times, neutron fluence variations, and then normalized to the observed reaction rate of the hydrogen-free 10 B control sample. Fitted lines to the datasets illustrate the trend in reaction rate with increasing scattering mass. Counting uncertainties for the data points are negligibly small; however, slight variations in the spacing of the disk stack, up to 20 disks, account for some variations observed among repeat analyses. The considerable hydrogen density (≈7.2 × 1022 atoms/cm3 ) in Nylon-6 is comparable to the hydrogen density of paraffin wax, polyethylene, and natural rubber. As
Rev. Sci. Instrum. 85, 045109 (2014) 1.6
Nylon Teflon Various (Labeled) Adhesive Tape
1.5
Reaction Enhancement
045109-3
Paraffin
LDPE
Nylon-6
1.4
1.3
1.2
1.1
Silicon 1.0 0
1000
2000
3000
4000
5000
6000
Thickness of Scattering Material (μm) FIG. 3. Graph of reaction enhancement as a function of increasing neutron scatterer thickness placed behind sample and normalized to the hydrogenfree control sample. Statistical counting error for each point is less than 1%. Lines are fitted to the data using the equation in Table I.
expected, the reaction rate correlates with mass, hydrogen density, and neutron scattering cross section. The relative reaction rate for nylon approaches a maximum ratio near 1.5 for the experimental conditions at the NIST NDP facility. Measurements were also conducted using Teflon and intrinsic single crystalline silicon, both shown in Figure 3. Again, these later two materials do not contain hydrogen; however, the scattering cross section of fluorine is about 4 barns. The 2 mm thick silicon wafer exhibited no measureable reaction enhancement. Not shown are two other materials evaluated for reaction enhancement: (1) a series of polycrystalline AlN plates that produced a reaction enhancement slightly greater than that seen for Teflon, and (2) Lithoflex, a Li-enriched polymer typically used for gamma-free neutron shielding [Ref. 2 - page 31] which gave no measureable enhancement. Strips of common transparent adhesive tape are occasionally applied to secure small samples to the mounting hardware for NDP analysis. A simple test was conducted to evaluate the potential for reaction enhancement due to the presence of hydrogenous tape during a measurement. Strips of tape 18 mm wide and cut to 50 mm lengths were progressively applied to the reverse side of the control sample between spectrum acquisitions. The thickness of a single strip, including the adhesive on one side, is ≈0.59 μm. Of course this may vary by composition, brand, batch, or even the location along the length of a roll. With increasing thickness (number of strips), the rapid rise in reaction rate was observed as with other hydrogenous materials. The enhancement curve seen in Figure 3 trends just below that observed for nylon of similar thickness. In this experiment 0, 3, 6, 10, 20, 25, and 30 strips of tape were applied. Variations in the number and size of sub-millimeter diameter air pockets trapped in the adhesive layer between successive tape layers could foreseeably introduce variation in the results. It is important to note that even a single layer of tape was sufficient to increase the reaction rate by over 3% and then decreases to an average of about 2% enhancement per layer as 10 layers of tape were applied.
045109-4
R. Gregory Downing
Rev. Sci. Instrum. 85, 045109 (2014) 1.2
1.6
1.0
Reaction Reduction
Reaction Enhancement
1.5
1.4
1.3
1.2
0.8
0.6
0.4
0.2
0.0
1.1
0
2000
4000
6000
8000
10000
Distance Separating Sample from Scatterer (μm)
1.0 0.1
1
10
100
Distance between Sample Surface and Nylon Surface (mm) FIG. 4. Graph of the measured enhancement to the reaction rate as an 8 mm thick disk of Nylon was moved away from the back of the control sample. The counting error at each point is less than 1%. The dashed line uses the fit equation and coefficients found in Table I, labeled “Scatterer Separation”.
Measurement results are graphically presented in Figure 4 for the condition of physically withdrawing a 5 mm thick, single mass of nylon from sample surface opposite of the impinging neutron beam. At discrete distance increments, spectra were collected to determine count rate. The reduction in rate enhancement follows the formula given in Table I and approaches no enhancement by the distance of 40 mm and beyond. It is important to note that the scattering masses approach the BSG layer on the control sample surface no closer than 0.3 mm because of thinnest silicon wafer thickness. When extrapolated to zero distance, the fitted line implies that if the boron was located in the hydrogenous material, the reaction enhancement would increase over that observed in this work. The final measurement set in this series of scattering experiments was obtained by stepwise increasing the hydrogenous mass placed between the control sample and the approaching neutron beam. The neutron beam must pass through the series of nylon masses prior to entering the adjacent control sample. As before, the change in reaction rate is normalized to the bare control sample with the counting uncertainty smaller than the displayed data points. Data obtained are shown in Figure 5. Variation in the stacking closeness of fit for the nylon disks is expected to produce minor variations TABLE I. The equation and the coefficients obtained by fitting the thickness values of nylon, Teflon, or the separation distances found in Figures 3 and 4 to the observed reaction rates from a control sample. The independent variable “x” is the material thickness or separation distance of the scatterer from the control sample surface. The dependent variable “y” is the relative reaction rate. This equation is similar to that of equation 5-48 by Duderstadt and Hamilton for neutron albedo.10 y = (a + cx)/(1 + bx) Material Nylon-6 Teflon Scatterer separation
a
b
c
1 1 1.5832
0.0015397 0.30610 0.51589
0.0023860 0.36951 0.49965
FIG. 5. Graph showing the reduction in neutron reaction rate with increasing mass of hydrogenous scatterer placed in the neutron beam path immediately prior to the sample. Two materials were evaluated: nylon and polypropylene.
in the fit to the curve and for reproducibility of measurements at a given thickness value. DISCUSSION
A series of measurements were conducted to investigate the influence of neutron scattering upon the effective reaction rates from the sample itself and the environment surrounding sample during NDP measurements. The effect has potential value or may cause significant inaccuracies in quantitative NDP analysis if uncorrected. Neutron research facilities designs vary and supply characteristic neutron energy spectra which affect the neutron capture and scatter probabilities in a sample. The average neutron energy of the NDP instrument at NIST is ≈5.2 meV. In contrast, typical thermal neutron beam provides an average energy of slightly greater than 25.8 meV. When the lower energy neutron backscatters from a room temperature sample, it gains energy from the interaction. The upshift of neutron energy correspondingly lowers the neutron capture probability, following a predictable inverse relationship. The temperature differential of a thermal neutron beam and a thermal sample is small; therefore, the neutron energy change is negligible and facilities that use of thermal neutron beams benefits more so from backscattered neutrons. Dr. Sacit Cetiner,11 using NDP at the Breazeale Nuclear Reactor at the Radiation Science and Engineering Center, The Pennsylvania State University, reports a greater reaction rate in a thermal neutron beam NDP measurement than reported here with cold neutron beams. In unpublished work, he used a portion of the BSG control sample and the nylon used in this study for a series of back scattering measurements. Nonetheless, a 5 meV energy neutron beam, equal in fluence rate to a thermal beam, will produce the higher reaction rate because the capture cross section is initially twice greater before incurring gains from the scattered neutrons. Several caveats are important to consider before applying the reaction enhancement technique in analytical NDP work. Quantitative determinations are frequently established by comparing a standard of known analyte abundance to a sample of unknown abundance in a comparable geometry. The usual standard of comparison at the NIST NDP facility is
045109-5
R. Gregory Downing
the control standard used in this study. The potential for error is nontrivial when samples containing differing amounts of neutron scatterers, i.e., hydrogenous polymers, are compared to a standard not matched with similar scattering properties. Or, if an adjacent scattering mass is in the neutron beam and in close proximity to the region of sample being measured, additional error may be introduced. Another confounding factor involves the use of adhesive tape to secure small samples in the NDP chamber. A single tape strip produces 3% reaction enhancement when placed across the back of the thin sample. Minimizing the use of tape to the edge of the sample or use of alternative methods of mounting small samples is recommended for the highest quality measurements. A Teflon sheet is low background material that compressively holds the sample in place when tape is applied outside the neutron beam. However, as shown in Figure 3, the use of Teflon only reduces scattering enhancement. Where practical, aluminum foil or silicon wafer would serve as a better support. Several questions arose during this work that could serve as basis for future studies. First is whether the analyte concentration profile is disproportionally enhanced with respect to depth when a measurement is performed within or in close proximity to a hydrogenous material? Reaction rate losses and gains over the typical escape depth of a charged particle appear as only a few percent considering the values obtained in Figures 3–5. More probably, the blurring from charged particle energy straggling over the profile range likely exceeds non-uniform reaction rates taking place in the sample. A second issue is when the sample area is small compared to the beam size and the scatterer environment is larger than the sample. As the lateral adjacent scatterer remains in the beam and in close proximity to the sample, the normalized neutron fluence rate at the sample would seemingly increase proportional to the decreasing sample area for a fixed size scattering mass. The same enhancing conditions are met by the use of a masking aperture on a large sample containing a neutron scatterer. Another consideration is when the neutron fluence is nonuniform across the beam area striking the sample within a scattering environment. Scattering tends to spatially redistribute neutrons uniformly into a spherical geometry. For instance, the scattering of Gaussian intensity beam profile would redistribute more evenly which may not change the effective flux, but would reduce the need for precise repositioning of sample and standard for comparability, or reduce the influence of analyte non-uniformities across the sample on the NDP profile. Finally, the question arises; can sample mounts using a hydrogenous structure further increase the neutron reaction rate? A few attempts to customize a neutron scattering environment for NDP samples were tested. The most involved piece was a roughly 100 mm × 100 mm × 100 mm polyethylene mass. The control sample was mounted at the center of the mass with a 20 mm hole drilled for the neutrons to directly illuminate the sample. A second 20 mm hole entered the mass at 45◦ for a detector to view the sample area illuminated by the neutron beam. No reaction enhancement was measured
Rev. Sci. Instrum. 85, 045109 (2014)
beyond that described in Figure 3 with a simple hydrogenous backing. In summary, analyte reaction rates in samples are enhanced and controllable for NDP measurements through careful and reproducible positioning of neutron scatterers. The reaction gain improves sample throughput or detection limits. Cold neutron NDP at NIST benefits nearly 50% in reaction rate when a 5 mm or greater thickness of hydrogenous material was placed to the back of a thin borated sample. Interference from background noise did not increase because of the geometrical drop in the scattered neutron fluence at typical detector operating distances. Although, radiation safety factors must be considered as neutrons are scattered into a spherical geometry. The reaction enhancement conveniently reaches maxima with a relatively small scattering mass. As most NDP instruments in the world operate using low fluence rate thermal neutron beams and limited operating schedules, exploiting the reaction enhancement is of broad benefit. ACKNOWLEDGMENTS
The author wishes to acknowledge helpful discussions with Dr. Sacit Cetiner, Oak Ridge National Laboratory, Dr. Robert Williams, National Institute of Standards and Technology (NIST), Dr. Jeremy Cook, NIST, and the determination of neutron fluence by Dr. Richard Lindstrom, NIST. Certain commercial products are identified in this paper in order to specify the experimental procedures in adequate detail. This identification does not imply recommendation or endorsement by the authors or by the National Institute of Standards and Technology, nor does it imply that the products identified are necessarily the best available for the purpose. Contributions of the National Institute of Standards and Technology are not subject to copyright. 1 R.
G. Downing, G. P. Lamaze, J. K. Langland, and S. T. Hwang, “Neutron depth profiling: Overview and description of NIST facilities,” NIST J. Res. 98, 109 (1993). 2 Neutron Depth Profiling “Chambers,” accessed February 27, 2014, https://sites.google.com/site/nistndp/home. 3 Handbook of Prompt Gamma Activation Analysis with Neutron Beams, edited by G. L. Molnár, (Chemical Research Centre, Budapest) (Kluwer Academic Publishers, Dordrecht, 2004), p. 424. 4 J. R. D. Copley and C. A. Stone, “Neutron scattering and its effect on reaction rates in neutron absorption experiments,” Nucl. Instrum. Methods., Sec. A 281, 593 (1989). 5 E. A. Mackey, D. L. Anderson, R. M. Lindstrom, and G. E. Gordon, “Effects of target shape and neutron scattering on element sensitivities of neutron capture prompt-gamma ray activation analysis,” Anal. Chem. 63, 288 (1991). 6 R. M. Lindstrom, R. R. Greenberg, D. F. R. Mildner, E. A. Mackey, and R. L. Paul, Neutron scattering and neutron reaction rates, in Proceedings of the Second International k0 Users Workshop, Ljubljana, Slovenia, Jozef Stefan Institute, edited by B. Smodis, pp. 50–53 (1997). 7 D. Thompson, S. J. Parry, and R. Benzing, “Evaluation of nuclear effects in the analysis of plastics by neutron activation analysis,” J. Radioanal. Nucl. Chem., Lett. 187(4), 255 (1994). 8 Jeremy Cook, National Institute of Standards and Technology, personal communication (2012). 9 R. M. Lindstrom, “Neutron fluence monitoring by foil activation at the NBSR,” Trans. Am. Nucl. Soc. 83, 268–269 (2000). 10 J. J. Duderstadt and L. J. Hamilton, Nuclear Reactor Analysis (John Wiley & Sons, Inc., New York/London/Sydney/Toronto, 1976). 11 Dr. Sacit Cetiner, Breazeale Nuclear Reactor, Pennsylvania State University, State College, PA, personal communication (2006).
Review of Scientific Instruments is copyrighted by the American Institute of Physics (AIP). Redistribution of journal material is subject to the AIP online journal license and/or AIP copyright. For more information, see http://ojps.aip.org/rsio/rsicr.jsp
Enhanced reaction rates in NDP analysis with neutron scattering R. Gregory Downinga) National Institute of Standards and Technology, Chemical Sciences Division, Gaithersburg, Maryland 20899, USA
(Received 27 February 2014; accepted 23 March 2014; published online 14 April 2014) Neutron depth profiling (NDP) makes accessible quantitative information on a few isotopic concentration profiles ranging from the surface into the sample a few micrometers. Because the candidate analytes for NDP are few, there is little interference encountered. Furthermore, neutrons have no charge so mixed chemical states in the sample are of no direct concern. There are a few nuclides that exhibit large probabilities for neutron scattering. The effect of neutron scattering on NDP measurements has not previously been evaluated as a basis for either enhancing the reaction rates or as a source of measurement error. Hydrogen is a common element exhibiting large neutron scattering probability found in or around sample volumes being analyzed by NDP. A systematic study was conducted to determine the degree of signal change when neutron scattering occurs during analysis. The relative signal perturbation was evaluated for materials of varied neutron scattering probability, concentration, total mass, and geometry. Signal enhancements up to 50% are observed when the hydrogen density is high and in close proximity to the region of analysis with neutron beams of sub thermal energies. Greater signal enhancements for the same neutron number density are reported for thermal neutron beams. Even adhesive tape used to position the sample produces a measureable signal enhancement. Because of the shallow volume, negligible distortion of the NDP measured profile shape is encountered from neutron scattering. [http://dx.doi.org/10.1063/1.4870414] INTRODUCTION
Neutron beams make possible the extraction of quantitative information about composition and the elemental distribution in materials, often nondestructively. The neutroninduced radiation is used to uniquely characterize the sample or corroborates measurements obtained through other analytical techniques. A major disadvantage with neutron beam techniques is that available neutron fluence rates are notably less than the particle fluence rates accessible by even modest laboratory probes based upon electrons, protons, and photons beams. As a result, data acquisition by neutron techniques is often protracted in comparison. Signal yield from neutron interactions and detection efficiency are improved by a variety of methods to make better use of the available neutrons. Detector positioning, neutron optics, improved electronics, and background mitigation are all critical experiment design parameters. Techniques like neutron depth profiling (NDP)1, 2 and prompt gamma activation analysis (PGAA)3 benefit from these tactics. In a typical analysis scenario the samples are inserted into a highly directed stream of neutrons, the beam passes through the sample substantially unperturbed because of small sample mass and limited neutron interaction cross section. If, however, the sample itself or the in-beam environment adjacent to the sample contains a strong neutron scatterer (e.g., H, C, . . . ), then a substantial fraction of the neutrons is scattered into 4π geometry space. Two changes occur to a fraction of those scattered neutrons: (1) the path length of the neutron is lengthened or
a) [email protected]
0034-6748/2014/85(4)/045109/5/$30.00
shortened4 through the sample and (2) the energy of the neutron may be altered. Previously, Mackey et al.,5 Lindstrom et al.,6 Thompson et al.,7 and references within these papers described how neutron scattering in hydrogenous materials alters the effective reaction rate and the background radiation. They have demonstrated that variables such as sample composition, geometric shape of the sample, and shifts in the neutron energy by way of neutron scattering will increase or decrease the bulk reaction rate compared to samples without significant scattering properties. This realization has necessitated tight control on how samples are prepared and then oriented with respect to the neutron beam for quantitative activation analysis. If not managed, the effects bias the concentration determination and render unreliable the attempt for analytical results. This is especially true for sample analysis in cold neutron beams with the increased scattering cross section and the consequential upshift in neutron energy for neutrons that scatter. Conversely, if hydrogenous or other materials with a significant neutron scattering cross section are systematically controlled, the increase in reaction rates from neutron scattering could provide benefit and improve detection limits or reduce analysis time. Neutron scattering is more easily exploited by the NDP technique where the depth of analysis is not a bulk sample response as with PGAA, but rather is limited to depths of less than 100 μm relative to the sample surface. More typical are the cases where the depth of NDP analysis is less than 20 μm. This also implies that the angle of the sample in the beam is of little concern, which is not the case for PGAA. In this work, sample environments were varied stepwise to study the change in reaction rate for NDP measurements and evaluate potential measurement errors.
85, 045109-1
045109-2
R. Gregory Downing
Rev. Sci. Instrum. 85, 045109 (2014)
EXPERIMENTAL
Measurements were conducted in the NDP instrument1 on the Neutron Guide 1 (NG1) end station at the NIST Center for Neutron Research (NCNR). Since 2012, the neutron spectrum at the sample plane peaks with a wavelength of ≈0.4 nm (5.2 meV) and an upper energy cutoff regulated by 2 liquid-nitrogen cooled Be-Bi filters.8 The thermal-energy equivalent beam fluence rate was determined to be greater than 1 × 109 n/cm2 /s.9 Prior to 2012 the neutron optics at this end station were optimized for a small angle neutron scattering instrument that restricted the available neutron fluence rate to 7.6 × 108 n/cm2 /s (thermal-energy equivalent). NDP measurements presented here were conducted during both time periods but are normalized to a common sample; therefore, variations in the neutron fluence rate are not a complicating factor. The full-width-half-maximum diameter of the neutron beam at the sample plane is ≈12 mm. Two samples, previously characterized for 10 B content and depth distribution, served as the control materials throughout this study. The samples consisted of a silicon wafer substrate of 475 μm or 300 μm thickness. One face of each sample is coated with ≈100 nm of borosilicate glass (BSG). Consequently, a NDP charged particle energy spectrum distinctly resolves all four energy peaks predicted from the 10 B(n,α)7 Li reaction (Figure 1). The 10 B content of the BSG layer shown in Figure 1, obtained from the 300 μm sample, is 5.22 ± 0.03 × 1015 atoms/cm2 . The uncertainty is a 95% confidence tolerance interval including random measurement error and systematic uncertainties. The sample area analyzed for each measurement was defined by a Teflon sheet containing an aperture ≈10 mm in diameter. Control samples completely filled the aperture and were entirely illuminated by the neutron beam during irradiation. The Teflon sheet was 0.5 mm thick; thus, reaction particles not originating within the aperture were unable to penetrate the sheet but rather are absorbed by the Teflon. The sheet was centered over a 0.8 mm thick, 150 mm diameter aluminum disk. A large diameter hole was cut at the center 2000
10B(n,α)7Li
Particle Count
1500
1000
500
0 500
1000
1500
2000
2500
3000
Channel Number FIG. 1. The 4 particle energy peaks measured for a control sample using a multichannel analyzer during NDP analysis of the 10 B(n,α)7 Li reaction. Only the peak area centered about channel 2100 was used in this work.
FIG. 2. Two perspective views (not to scale) of an aluminum disk mount used to reproducibly aperture and position the sample in the NDP chamber. The upper figure is from the perspective of approaching neutrons. The bottom view is an edge-on perspective of the sample mount showing the configuration of various components. False color legend: Teflon mask (aqua), sample (red), mounting ring (grey), and neutron scattering material (blue).
of the Al disk. The entire assembly is depicted in Figure 2. Also shown is the shallow notch cut into the outer rim of Al disk to ensure reproducible mounting and orientation of the holder for each measurement. Both the sample and the NDP chamber environment remained at room temperature. A transmission-type silicon surface-barrier detector was positioned at a distance of ≈120 mm from the sample. The detector face is circular with an active area of 50 mm2 concentric and parallel to the sample surface defined by the Teflon aperture. The thinness of the transmission-type detector (≈40 μm) significantly reduces background noise in the spectrum arising from ambient prompt gamma and x-ray radiation present in the NDP chamber during analysis. Radiation and electronic background noise is seen increasing in intensity at the lower energy region of the spectrum, but it does not contribute counts in the two right-most (higher energy) alpha peak areas (Figure 1). A neutron monitor operates concurrently with each sample spectrum acquisition to account for any small variations in neutron flux between runs. Each NDP spectrum was normalized to the respective integrated neutron flux monitor count. The consistent use of long sample counts reduced the statistical error of neutron flux correction to about 0.2% ± 0.1% uncertainty. The first set of charged-particle spectra were collected by attaching hydrogenous neutron scatterers of known thicknesses and masses in the configuration shown in Figure 2. Measurements were taken with scatterers in an arbitrary thickness order. Individual disk-shaped scatterers were 40 mm–50 mm diameter and ≈0.49 mm thick. The larger diameter pieces were used in earlier measurements and the 40 mm diameter pieces were used in the most recent measurements. They stacked immediately behind (downstream) and centered flat against the control sample. High-purity Nylon-6, a semi-crystalline polyamide solid, was selected for the scatterer based on properties of high hydrogen density, low neutron absorption, and mechanical properties. The Nylon-6 has a molecular composition of C6 H11 ON with
R. Gregory Downing
an average bulk density of 1.12 g/cm3 . From the formula weight of 113.16 g/mole, the hydrogen density is calculated as ≈0.11 g/cm3 . Bound hydrogen has a nominal scattering cross section of ≈80 barns for the neutron energy spectrum delivered to the NDP instrument. A second set of NDP measurements was collected using Teflon disks in the same geometry as with the previous Nylon experiment. Teflon has various formulations but do not contain hydrogen. However, fluorine has a nominal 4 barn scattering cross section. The Teflon sheets were ≈0.49 mm thick. Again, individual disks 25 mm in diameter were stacked in known but random thicknesses centered behind the control sample. Spectra were collected to detect correlations in reaction rate to the applied mass. In addition, masses of other materials including paraffin, low density polyethylene (LDPE), adhesive tape, silicon, and a lithium rich polymer were evaluated as neutron scatterers in geometrically equivalent NDP experiments. A third dataset was collected by positioning a single 50 mm diameter, 8 mm thick disk of Nylon-6 downstream of the sample between spectrum acquisition. The spectra were taken in a sequence of increasing distances to evaluate the geometry of scatterer upon the sample reaction rate. Similar to the previous datasets, the scatterer was kept aligned perpendicular to the neutron beam and concentric with sample center. In a final measurement set, the series of Nylon-6 disc thicknesses, described above, were centered and attached to the same side of the sample as the approaching neutron beam. That is, the scatterer intercepted the neutron beam and the neutrons must pass through the scatterer prior to reaching the sample. In this scenario the sample holder (Figure 2) was simply reversed 180◦ and mounted in the NDP chamber so the BSG side of the control sample faced an antipode detector. The detector is equal in distance from the sample as the upstream detector in the previous three sets of measurements. The spectra obtained from all four measurement sets are described below. RESULTS
As downstream masses of hydrogenous material adjacent to the thin control sample were increased, the reaction rate of 10 B(n,α)7 Li increased rapidly first, then slowed and asymptotically approached a maximum enhancement value. The change in reaction rate is plotted in Figure 3 as a function of scatterer thickness for various materials. These data are corrected where necessary for spectral background noise, variations in count times, neutron fluence variations, and then normalized to the observed reaction rate of the hydrogen-free 10 B control sample. Fitted lines to the datasets illustrate the trend in reaction rate with increasing scattering mass. Counting uncertainties for the data points are negligibly small; however, slight variations in the spacing of the disk stack, up to 20 disks, account for some variations observed among repeat analyses. The considerable hydrogen density (≈7.2 × 1022 atoms/cm3 ) in Nylon-6 is comparable to the hydrogen density of paraffin wax, polyethylene, and natural rubber. As
Rev. Sci. Instrum. 85, 045109 (2014) 1.6
Nylon Teflon Various (Labeled) Adhesive Tape
1.5
Reaction Enhancement
045109-3
Paraffin
LDPE
Nylon-6
1.4
1.3
1.2
1.1
Silicon 1.0 0
1000
2000
3000
4000
5000
6000
Thickness of Scattering Material (μm) FIG. 3. Graph of reaction enhancement as a function of increasing neutron scatterer thickness placed behind sample and normalized to the hydrogenfree control sample. Statistical counting error for each point is less than 1%. Lines are fitted to the data using the equation in Table I.
expected, the reaction rate correlates with mass, hydrogen density, and neutron scattering cross section. The relative reaction rate for nylon approaches a maximum ratio near 1.5 for the experimental conditions at the NIST NDP facility. Measurements were also conducted using Teflon and intrinsic single crystalline silicon, both shown in Figure 3. Again, these later two materials do not contain hydrogen; however, the scattering cross section of fluorine is about 4 barns. The 2 mm thick silicon wafer exhibited no measureable reaction enhancement. Not shown are two other materials evaluated for reaction enhancement: (1) a series of polycrystalline AlN plates that produced a reaction enhancement slightly greater than that seen for Teflon, and (2) Lithoflex, a Li-enriched polymer typically used for gamma-free neutron shielding [Ref. 2 - page 31] which gave no measureable enhancement. Strips of common transparent adhesive tape are occasionally applied to secure small samples to the mounting hardware for NDP analysis. A simple test was conducted to evaluate the potential for reaction enhancement due to the presence of hydrogenous tape during a measurement. Strips of tape 18 mm wide and cut to 50 mm lengths were progressively applied to the reverse side of the control sample between spectrum acquisitions. The thickness of a single strip, including the adhesive on one side, is ≈0.59 μm. Of course this may vary by composition, brand, batch, or even the location along the length of a roll. With increasing thickness (number of strips), the rapid rise in reaction rate was observed as with other hydrogenous materials. The enhancement curve seen in Figure 3 trends just below that observed for nylon of similar thickness. In this experiment 0, 3, 6, 10, 20, 25, and 30 strips of tape were applied. Variations in the number and size of sub-millimeter diameter air pockets trapped in the adhesive layer between successive tape layers could foreseeably introduce variation in the results. It is important to note that even a single layer of tape was sufficient to increase the reaction rate by over 3% and then decreases to an average of about 2% enhancement per layer as 10 layers of tape were applied.
045109-4
R. Gregory Downing
Rev. Sci. Instrum. 85, 045109 (2014) 1.2
1.6
1.0
Reaction Reduction
Reaction Enhancement
1.5
1.4
1.3
1.2
0.8
0.6
0.4
0.2
0.0
1.1
0
2000
4000
6000
8000
10000
Distance Separating Sample from Scatterer (μm)
1.0 0.1
1
10
100
Distance between Sample Surface and Nylon Surface (mm) FIG. 4. Graph of the measured enhancement to the reaction rate as an 8 mm thick disk of Nylon was moved away from the back of the control sample. The counting error at each point is less than 1%. The dashed line uses the fit equation and coefficients found in Table I, labeled “Scatterer Separation”.
Measurement results are graphically presented in Figure 4 for the condition of physically withdrawing a 5 mm thick, single mass of nylon from sample surface opposite of the impinging neutron beam. At discrete distance increments, spectra were collected to determine count rate. The reduction in rate enhancement follows the formula given in Table I and approaches no enhancement by the distance of 40 mm and beyond. It is important to note that the scattering masses approach the BSG layer on the control sample surface no closer than 0.3 mm because of thinnest silicon wafer thickness. When extrapolated to zero distance, the fitted line implies that if the boron was located in the hydrogenous material, the reaction enhancement would increase over that observed in this work. The final measurement set in this series of scattering experiments was obtained by stepwise increasing the hydrogenous mass placed between the control sample and the approaching neutron beam. The neutron beam must pass through the series of nylon masses prior to entering the adjacent control sample. As before, the change in reaction rate is normalized to the bare control sample with the counting uncertainty smaller than the displayed data points. Data obtained are shown in Figure 5. Variation in the stacking closeness of fit for the nylon disks is expected to produce minor variations TABLE I. The equation and the coefficients obtained by fitting the thickness values of nylon, Teflon, or the separation distances found in Figures 3 and 4 to the observed reaction rates from a control sample. The independent variable “x” is the material thickness or separation distance of the scatterer from the control sample surface. The dependent variable “y” is the relative reaction rate. This equation is similar to that of equation 5-48 by Duderstadt and Hamilton for neutron albedo.10 y = (a + cx)/(1 + bx) Material Nylon-6 Teflon Scatterer separation
a
b
c
1 1 1.5832
0.0015397 0.30610 0.51589
0.0023860 0.36951 0.49965
FIG. 5. Graph showing the reduction in neutron reaction rate with increasing mass of hydrogenous scatterer placed in the neutron beam path immediately prior to the sample. Two materials were evaluated: nylon and polypropylene.
in the fit to the curve and for reproducibility of measurements at a given thickness value. DISCUSSION
A series of measurements were conducted to investigate the influence of neutron scattering upon the effective reaction rates from the sample itself and the environment surrounding sample during NDP measurements. The effect has potential value or may cause significant inaccuracies in quantitative NDP analysis if uncorrected. Neutron research facilities designs vary and supply characteristic neutron energy spectra which affect the neutron capture and scatter probabilities in a sample. The average neutron energy of the NDP instrument at NIST is ≈5.2 meV. In contrast, typical thermal neutron beam provides an average energy of slightly greater than 25.8 meV. When the lower energy neutron backscatters from a room temperature sample, it gains energy from the interaction. The upshift of neutron energy correspondingly lowers the neutron capture probability, following a predictable inverse relationship. The temperature differential of a thermal neutron beam and a thermal sample is small; therefore, the neutron energy change is negligible and facilities that use of thermal neutron beams benefits more so from backscattered neutrons. Dr. Sacit Cetiner,11 using NDP at the Breazeale Nuclear Reactor at the Radiation Science and Engineering Center, The Pennsylvania State University, reports a greater reaction rate in a thermal neutron beam NDP measurement than reported here with cold neutron beams. In unpublished work, he used a portion of the BSG control sample and the nylon used in this study for a series of back scattering measurements. Nonetheless, a 5 meV energy neutron beam, equal in fluence rate to a thermal beam, will produce the higher reaction rate because the capture cross section is initially twice greater before incurring gains from the scattered neutrons. Several caveats are important to consider before applying the reaction enhancement technique in analytical NDP work. Quantitative determinations are frequently established by comparing a standard of known analyte abundance to a sample of unknown abundance in a comparable geometry. The usual standard of comparison at the NIST NDP facility is
045109-5
R. Gregory Downing
the control standard used in this study. The potential for error is nontrivial when samples containing differing amounts of neutron scatterers, i.e., hydrogenous polymers, are compared to a standard not matched with similar scattering properties. Or, if an adjacent scattering mass is in the neutron beam and in close proximity to the region of sample being measured, additional error may be introduced. Another confounding factor involves the use of adhesive tape to secure small samples in the NDP chamber. A single tape strip produces 3% reaction enhancement when placed across the back of the thin sample. Minimizing the use of tape to the edge of the sample or use of alternative methods of mounting small samples is recommended for the highest quality measurements. A Teflon sheet is low background material that compressively holds the sample in place when tape is applied outside the neutron beam. However, as shown in Figure 3, the use of Teflon only reduces scattering enhancement. Where practical, aluminum foil or silicon wafer would serve as a better support. Several questions arose during this work that could serve as basis for future studies. First is whether the analyte concentration profile is disproportionally enhanced with respect to depth when a measurement is performed within or in close proximity to a hydrogenous material? Reaction rate losses and gains over the typical escape depth of a charged particle appear as only a few percent considering the values obtained in Figures 3–5. More probably, the blurring from charged particle energy straggling over the profile range likely exceeds non-uniform reaction rates taking place in the sample. A second issue is when the sample area is small compared to the beam size and the scatterer environment is larger than the sample. As the lateral adjacent scatterer remains in the beam and in close proximity to the sample, the normalized neutron fluence rate at the sample would seemingly increase proportional to the decreasing sample area for a fixed size scattering mass. The same enhancing conditions are met by the use of a masking aperture on a large sample containing a neutron scatterer. Another consideration is when the neutron fluence is nonuniform across the beam area striking the sample within a scattering environment. Scattering tends to spatially redistribute neutrons uniformly into a spherical geometry. For instance, the scattering of Gaussian intensity beam profile would redistribute more evenly which may not change the effective flux, but would reduce the need for precise repositioning of sample and standard for comparability, or reduce the influence of analyte non-uniformities across the sample on the NDP profile. Finally, the question arises; can sample mounts using a hydrogenous structure further increase the neutron reaction rate? A few attempts to customize a neutron scattering environment for NDP samples were tested. The most involved piece was a roughly 100 mm × 100 mm × 100 mm polyethylene mass. The control sample was mounted at the center of the mass with a 20 mm hole drilled for the neutrons to directly illuminate the sample. A second 20 mm hole entered the mass at 45◦ for a detector to view the sample area illuminated by the neutron beam. No reaction enhancement was measured
Rev. Sci. Instrum. 85, 045109 (2014)
beyond that described in Figure 3 with a simple hydrogenous backing. In summary, analyte reaction rates in samples are enhanced and controllable for NDP measurements through careful and reproducible positioning of neutron scatterers. The reaction gain improves sample throughput or detection limits. Cold neutron NDP at NIST benefits nearly 50% in reaction rate when a 5 mm or greater thickness of hydrogenous material was placed to the back of a thin borated sample. Interference from background noise did not increase because of the geometrical drop in the scattered neutron fluence at typical detector operating distances. Although, radiation safety factors must be considered as neutrons are scattered into a spherical geometry. The reaction enhancement conveniently reaches maxima with a relatively small scattering mass. As most NDP instruments in the world operate using low fluence rate thermal neutron beams and limited operating schedules, exploiting the reaction enhancement is of broad benefit. ACKNOWLEDGMENTS
The author wishes to acknowledge helpful discussions with Dr. Sacit Cetiner, Oak Ridge National Laboratory, Dr. Robert Williams, National Institute of Standards and Technology (NIST), Dr. Jeremy Cook, NIST, and the determination of neutron fluence by Dr. Richard Lindstrom, NIST. Certain commercial products are identified in this paper in order to specify the experimental procedures in adequate detail. This identification does not imply recommendation or endorsement by the authors or by the National Institute of Standards and Technology, nor does it imply that the products identified are necessarily the best available for the purpose. Contributions of the National Institute of Standards and Technology are not subject to copyright. 1 R.
G. Downing, G. P. Lamaze, J. K. Langland, and S. T. Hwang, “Neutron depth profiling: Overview and description of NIST facilities,” NIST J. Res. 98, 109 (1993). 2 Neutron Depth Profiling “Chambers,” accessed February 27, 2014, https://sites.google.com/site/nistndp/home. 3 Handbook of Prompt Gamma Activation Analysis with Neutron Beams, edited by G. L. Molnár, (Chemical Research Centre, Budapest) (Kluwer Academic Publishers, Dordrecht, 2004), p. 424. 4 J. R. D. Copley and C. A. Stone, “Neutron scattering and its effect on reaction rates in neutron absorption experiments,” Nucl. Instrum. Methods., Sec. A 281, 593 (1989). 5 E. A. Mackey, D. L. Anderson, R. M. Lindstrom, and G. E. Gordon, “Effects of target shape and neutron scattering on element sensitivities of neutron capture prompt-gamma ray activation analysis,” Anal. Chem. 63, 288 (1991). 6 R. M. Lindstrom, R. R. Greenberg, D. F. R. Mildner, E. A. Mackey, and R. L. Paul, Neutron scattering and neutron reaction rates, in Proceedings of the Second International k0 Users Workshop, Ljubljana, Slovenia, Jozef Stefan Institute, edited by B. Smodis, pp. 50–53 (1997). 7 D. Thompson, S. J. Parry, and R. Benzing, “Evaluation of nuclear effects in the analysis of plastics by neutron activation analysis,” J. Radioanal. Nucl. Chem., Lett. 187(4), 255 (1994). 8 Jeremy Cook, National Institute of Standards and Technology, personal communication (2012). 9 R. M. Lindstrom, “Neutron fluence monitoring by foil activation at the NBSR,” Trans. Am. Nucl. Soc. 83, 268–269 (2000). 10 J. J. Duderstadt and L. J. Hamilton, Nuclear Reactor Analysis (John Wiley & Sons, Inc., New York/London/Sydney/Toronto, 1976). 11 Dr. Sacit Cetiner, Breazeale Nuclear Reactor, Pennsylvania State University, State College, PA, personal communication (2006).
Review of Scientific Instruments is copyrighted by the American Institute of Physics (AIP). Redistribution of journal material is subject to the AIP online journal license and/or AIP copyright. For more information, see http://ojps.aip.org/rsio/rsicr.jsp
