Scholarly article on topic 'A Neutron Diffractometer for a Long Pulsed Neutron Source'

A Neutron Diffractometer for a Long Pulsed Neutron Source Academic research paper on "Physical sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Physics Procedia
OECD Field of science
Keywords
{"Neutron Diffraction" / "Long-pulse neutron sources" / "wavelength multiplexing"}

Abstract of research paper on Physical sciences, author of scientific article — Paul Sokol, Cailin Wang

Abstract Long pulsed neutron sources are being actively developed as small university based sources and are being considered for the next generation of high powered sources, such as the European Neutron Source (ESS) and the Spallation Neutron Source (SNS) second target station. New instrumentation concepts will be required to effectively utilize the full spectrum of neutrons generated by these sources. Neutron diffractometers, which utilize time-of-flight (TOF) techniques for wavelength resolution, are particularly problematic. We describe an instrument for a long pulsed source that provides resolution comparable to that obtained on short pulsed sources without the need of long incident flight paths. We accomplish this by utilizing high speed choppers to impose a time structure on the spectrum of incident neutrons. By strategically positioning these choppers the response matrix assumes a convenient form that can be deconvoluted from the measured TOF spectrum to produce the diffraction pattern of the sample. We compare the performance of this instrument to other possible diffraction instruments that could be utilized on a long pulsed source.

Academic research paper on topic "A Neutron Diffractometer for a Long Pulsed Neutron Source"

Physics

Procedía

Union of Compact Accelerator-Driven Neutron Sources I & II

A Neutron Diffractometer for a Long Pulsed Neutron Source

Paul Sokola'b'* and Cailin Wangb

aDepartment of Physics, Indiana University, Bloomington, IN 47405 USA bCenter for the Exploration of Energy and Matter, Indiana University, Bloomington, IN 47408, USA

Abstract

Long pulsed neutron sources are being actively developed as small university based sources and are being considered for the next generation of high powered sources, such as the European Neutron Source (ESS) and the Spallation Neutron Source (SNS) second target station. New instrumentation concepts will be required to effectively utilize the full spectrum of neutrons generated by these sources. Neutron diffractometers, which utilize time-of-flight (TOF) techniques for wavelength resolution, are particularly problematic. We describe an instrument for a long pulsed source that provides resolution comparable to that obtained on short pulsed sources without the need of long incident flight paths. We accomplish this by utilizing high speed choppers to impose a time structure on the spectrum of incident neutrons. By strategically positioning these choppers the response matrix assumes a convenient form that can be deconvoluted from the measured TOF spectrum to produce the diffraction pattern of the sample. We compare the performance of this instrument to other possible diffraction instruments that could be utilized on a long pulsed source.

© 2012 Published by Elsevier B.V. Selection and/or peer-review under responsibility of UCANS

Keywords: Neutron Diffraction; Long-pulse neutron sources; wavelength multiplexing

Available online at www.sciencedirect.cor

SciVerse ScienceDirect

Physics Procedía 26 (2012) 178 - 187

Corresponding author. Tel.: 1-812-856-1458.

E-mail address: pesokol@indiana.edu.

1875-3892 © 2012 Published by Elsevier B.V. Selection and/or peer-review under responsibility of UCANS doi:10.1016/j.phpro.2012.03.023

1. Introductiom

Structure is one of the most fundamental properties of materials in a broad range of disciplines. Our understanding of modern materials and systems such as superconductors, magnetic materials, liquid crystals, layered materials, materials for renewable energy, polymers, rocks and zeolites - to name just a few - depends on a detailed understanding of the structure at the atomic and molecular level [1-3]. Diffraction techniques are the standard tool for structural determination. X-ray diffraction is excellent for studying heavy elements, but not for light elements due to the similarities in the x-ray cross section [1]. Neutron diffraction nicely compliments X-ray scattering since neutrons are ideal to study compounds with light elements or those that are comprised of neighboring elements. Unfortunately, neutron diffraction capabilities are rare in academic settings and exist at only a few universities with research reactors.

In this paper, we propose a novel neutron diffractometer for the Low Energy Neutron Source (LENS) located at Indiana University [4]. LENS is the only university-based long-pulse, cold neutron source in the US and offers unique capabilities for research, and instrumentation development in neutron scattering. This proposed instrument will be an excellent tool for studying the large scale structures that are increasingly important in materials science. It utilizes a novel wavelength multiplexing scheme to deliver the resolution obtainable on a short pulsed source without resorting to extremely long incident flight paths. It would serve as a prototype for future instruments on proposed long pulsed neutron sources such as the second target station at the Spallation Neutron Source and the European Spallation Source.

In many respects the instrument we are proposing will have performance characteristics similar to the Special Environment Powder Diffractometer (SEPD) that was located at the Intense Pulsed Neutron Source (IPNS) [4,5]. SEPD utilized 0.5 to 10 Â neutrons which allowed it to study structures with d-spacing from ~ 0.25 to 5 Â. Our instrument will utilize 0.5-20 Â neutrons, due to the lower source repetition rate and cold moderator, approximately doubling the size of unit cells that can be studied. SEPD utilized a conventional (non-guide) incident flight path to provide a 1 cm x 0.5 cm beam at the sample position and delivered a flux of 5x106 n/cm2/s. Our instrument will utilize guides which give an increase in flux. However, the choppers needed to provide adequate resolution with a modest flight path counteract some of this gain but we will have a factor of 5-10 more flux on sample than SEPD for comparable resolution (Ad/d ~5x10-3). Samples with masses from 0.05 g to 20g can be studied in reasonable times (from minutes to hours).

2. Short Pulsed Source Diffraction Instruments

Powder diffraction instruments at pulsed neutron sources rely on TOF techniques to determine neutron wavelength. A typical neutron powder diffractometer utilizes a long initial flight path, which primarily determines the wavelength resolution, between the neutron source and the sample. A shorter final flight path separates the sample from neutron detectors that surround it. The de Broglie relation gives

h t X = — = (3955s / m)-

mv l (1)

where A, m and v are the wavelength, mass and velocity of the neutron, h is Planck's constant and l and t are the flight path length from the source to the detector and the TOF of the neutron. Since the neutron energy (velocity) does not change during the diffraction process the neutron wavelength is directly proportional to its TOF. Bragg's law can then be written as

d = ^ = (3955s / m)—--

2 sin 6 2l sin 6 (2)

where d is the spacing of a plane in the sample and 26 is the scattering angle. Thus, at a fixed scattering angle the neutron time of flight is directly proportional to d. This highlights one of the advantages of TOF techniques - the entire diffraction pattern can, in principal, be recorded by a single fixed detector. This feature is particularly valuable for studies in special sample environments (e.g. low and high temperatures, high pressures, etc.)

The layout of a short pulse TOF diffractometer and its operation is shown graphically in Fig.1 [4,5]. Neutrons generated at the source in a short pulse (-microseconds) have a broad range of wavelengths. They propagate towards the sample, and scatter from the sample, and those with the appropriate wavelength will be diffracted into the detector. The measured TOF spectrum is directly proportional to the diffraction pattern for the sample. Standard techniques, such as Rietveld refinement, are carried out for structural determinations.

Fig. 1 Schematic layout and time-distance diagram for a short-pulse neutron diffractometer (SEPD).

^ One can express these relationships mathematically. If D is an array of a different value of ^-spacing, T is an array of a different value of the neutron TOF then their relation is expressed via a response matrix fiby

T = rD (3)

For an ideal TOF diffractometer, with perfect instrumental resolution, the response function will have the form

Ry. = I(-. )S(dl - at j); a = (3955s / m) / 2l sin 6 (4)

which simply represents the proportionality between the d-spacing and the neutron TOF. In ideal cases, the response function will have non-zero elements only along a diagonal. The source intensity I(-j) in eq. (4) is transformed from the intensity I(X) at neutron wavelength X through eq. (1), where I(X) is obtained from intensity spectrum I(E) expressed as a sum of Maxwellian distribution and an epithermal term [6].

In general, the allowed d-spacings for the sample can then be recovered from the measured scattering intensities in eq. (3) by inverting the response matrix

where matrix algebra can be directly employed to extract information on the crystal structure. In reality, the discrete nature of matrix expression in eq. (5) can easily lead to some diagonal elements becoming

zero and R matrix becomes singular, making direct inversion unrealistic. In addition, a direct inversion without any constraint will lead to the solution is unstable and even negative when the noise is presented

in the reflection data T [7]. Indirect inversion techniques, such constrained least-square or regularization methods, are useful to reconstruct diffraction pattern I(d) from noisy experimental data.

Note that the response function is determined only by the source performance and the instrument geometry. In practice, the response function will include instrumental effects which can give slightly offdiagonal elements. Due to finite moderator pulse width, sample size and detector size, a simple expression for the resolution of a TOF diffractometer is approximately given by

M _ ~d

+(cotOAO)2 +

where Ad, At, AO and Al are the uncertainties in the plane spacing, neutron TOF, Bragg angle and total flight path length, respectively [2]. The primary contribution to At is the width (in time) of the initial neutron pulse at the moderator. For an optimized instrument design, the time and geometrical contributions to the resolution are made equal by appropriate design.

For short pulsed neutron sources the final flight path and detector size determine the geometric components while the moderator pulse width and incident flight path length determine the time contribution. For example, SEPD had an incident flight path of 14 m, a final flight path of 1.5 m, and 1.5 cm diameter detectors. This gives an instrumental resolution of Ad/d~5x10-3 for a 90o scattering angle. It is sufficient for structural determinations in moderately complex structures.

Due to the limited resolution (Adid) in lattice spacing, the delta-function in eq. (4) needs to be replaced

by a realistic Gaussian function (2ji)_1 2exp[-(i/1-a/j)2/2o/], where aj=Ad. An example of R matrix is

■*—i

shown in Fig. 2. Obviously the non-zero elements are in the vicinity of a diagonal in the R matrix.

Fig. 2 Example of £ -matrix in 3D and its contour plot on d-TOF plane for the short-pulse diffractometer (SEPD). R -matrix has a

dimension of 500 500. The parameters used for calculating R are: time-channel width-27.35 |is, pulse width-5 |is, moderator-to-sample distance-14 m, Ad/d=3x10-3, and diffraction angle 9=75°

3. Long Pulsed Neutron Diffractometer

Simply translating the design of short-pulsed instruments to long-pulsed ones is problematic. To maintain comparable resolution each of the terms in eq. (6) must have the same value when an instrument design is translated from a short pulsed source to a long pulsed source. Since the geometrical contributions are determined primarily by the final flight path and detector design, this is easily accomplished. The timing contribution, on the other hand is primarily determined by the source pulse length and the incident flight path (which sets the neutron TOF). In the case of SEPD considered above the source pulse length was ~ 2|is and the total flight path was 15.5 m, giving a TOF for 1 Â neutrons of ~4 ms. This gives At/t ~ 5x10-4 which is comparable to the instrumental resolution, as expected.

Obtaining similar performance for an instrument on a long pulsed source would require maintaining a similar value of At/t. Typical values for At at long pulsed sources are on the order of 1-2 ms. The pulse length at LENS will be 1.6 ms and both the SNS second target station and the ESS use similar values. Thus, to obtain the same resolution as SEPD would require a total flight path length of approximately 10 km! Such a long flight path is clearly not feasible. Other solutions, including shortening the pulse width or bandwidth with choppers, are equally unpalatable since they result in a severe loss of intensity.

We propose a new instrument, as shown in Fig. 3, more suited to a long pulsed source. It provides a resolution comparable to that of short pulsed sources without using long incident flight paths or sacrificing intensity. It consists of a 14 m incident flight path and a 1.5 m final flight path, and utilizes a m=2 guide for the incident flight path. Two high speed double disc choppers are located at 5.5 and 10.5 m, and a bandwidth limiting chopper is located at 1.5 m, with all distances relative to the moderator surface. The first high speed chopper splits the long macro-pulse from the source into micro-pulses. For our instrument, the 1.6 ms long macro-pulse would be split into approximately 200 micro-pulses comprised of a 4|is neutron pulse followed by a 4 |is space. Most of micro-pulses still contain a broad spectrum of neutron wavelengths.

Moderator

t Chopper

Detector

Target

Fig. 3 Schematic layout and time-distance diagram of new TOF neutron diffractometer

Detector Sample

Utilizing only a single high speed chopper to introduce a time structure on the macro-pulse does not provide any advantage. The observed TOF signal would be the sum of the signals from each micro-pulse. Thus, we would observe the same pattern, but with different time offsets. Determination of the original diffraction pattern from this signal would be difficult since each TOF channel would, in principle, have a contribution from every wavelength in the spectrum.

The second high speed chopper with the same period can be introduced to select a series of discrete wavelengths from each of the micro-pulses. The second chopper actually selects out the same discrete wavelength range from each of the micro-pulses. However, each of the wavelengths selected corresponds to a different emission time at the source. Finally, the intensity in each wavelength band will depend on the source intensity corresponding to that wavelength and emission time.

An appropriate spacing of the choppers leads to a considerable simplification of the system. By choosing the distance between the two choppers equal to the distance from the second chopper to the sample all neutrons, independent of either emission time or wavelength, will arrive at the detector in a discrete set of time windows. If we assume that the choppers have a period of At and characterize the opening of the first (second) chopper by the integer n(m) then the neutron wavelength and arrival time are given by

A = (3955s / m) — (m - n); T = (2m - n)At

where L is the spacing between the two choppers and between the final chopper and the detector. By fixing the chopper-chopper and chopper-sample distance at the same value, Bragg's law can then be written as

(3955s

2L sin 6 2m - n (8)

which is similar to SEPD result. The main difference is that each reflection with a particular d-value can, in principle, contribute to several different time channels. Thus, we no longer have the simple one-to-one correspondence of the short pulsed instrument. However, we still retain a well defined relationship between d and t and can still define a response matrix:

Rj = ZZ7(Km, te-atj(m -n)/(2m -n));

a = (3955s / m)/ 2 L sin# (:1)

where te is the emission time for the neutron that makes it through the nth opening of the first chopper and the mth opening of the second chopper.

We have simulated the response of the incident flight path for this system using VITESS software [19]. Choppers with a 4 |is window and a 4 |is closed period where utilized. The results are shown in Fig. 4. The close spacing of the pulses simulates a continuous, if noisy, spectrum. However, closer examination the insert in Fig. 7 shows that the neutrons arrive at the detector at evenly spaced time intervals. Furthermore the width of the pulses at the detector agrees well with the chopper opening. This gives an effective width of approximately 5 |is. For 10 A neutrons this corresponds to a At/t of 1x10-4. Thus, with proper choice of the final flight path distance and the detector size we can easily achieve the desired resolution of 5x10-3.

Tllltu

Fig. 4 Incident spectrum at sample position simulated by VITESS software.

The intensity of the incident spectrum was also evaluated using VITESS. Developing suitable moderator kernels and ensuring they accurately represent the source intensity is always problematic. Therefore, we carried out three simulations to compare relative intensities: No guide or chopper (i.e. SEPD), guide only (i.e. SEPD with a guide), and guide and chopper (i.e. this instrument). We found that the addition of a guide increased the flux on SEPD by approximately a factor of 100 at the longer wavelengths that will be of most interest on this instrument. This is in reasonable agreement with simple estimates based on 1/r2 losses.

The introduction of choppers leads to a reduction in intensity of approximately 90%. One would expect a reduction of 75% due to the introduction of two choppers both with 50% transmission. The additional reduction results from the fact that the choppers have a finite rotational speed leading to a triangular, rather than a square, transmission function. However, even so, this instrument would be expected to have a flux roughly a factor of 5-10 greater than on SEPD. We should also note that no attempt was made to optimize performance in these comparisons so slightly larger gains may be possible.

We calculated the system response function, where the long-pulses of neutrons are assumed to come from a moderator at temperature 7=3(M<. and pass through a pulse shaping chopper. The pulse-length is assumed to be t= 1.0-1.6 ms [6]. The fl-matrix is calculated by replacing the delta function in eq. (11) with a Gaussian function, whose standard deviation Ad is assumed constant for all d. For simplicity, the instrument resolution function in this paper is defined as Ad/d at d=1 A. Fig. 5 shows an example of non-

square R matrix as function of TOF and ¿/-space. Obviously there is a wide band of non-zero elements, resulting from pulse overlapping effect as mentioned above.

Fig.5 Example of J?-matrix in 3D and its contour plot on c/-TOF plane for the new diffractometer. R -matrix has a dimension of 500x3919. The parameters used for calculating

R are: time-channel width-4 ^s, pulse width-1.6 ms, chopper period-8 ^s, slit width in chopper-4 |is, moderator-to-sample distance-14 m, moderator-to-first-chopper distance is 4 m, and diffraction angle 0=85°.

4. Diffraction Pattern Reconstruction

The capability of reconstructing I(d) from raw TOF data is demonstrated for both short-pulse and longpulse diffractometers. The I(d) pattern can be recovered from the simulated raw data using a constrained least-square method with a criterion as follows

Ra 0h ((modeled) -1{ Ij(d)(modeled) > 0

(raw)]2 minimum

The sum in eq. (12) is over /'=1,2,...N grids on the time t axis, and j=1,2,...M grids on the lattice-space

d axis. The variance of reconstruction (% ) is defined as

- — V'V

[Jf ( Q (top d e ! b d) -It (. Q (raw)]'1

can be used to characterize the quality of reconstruction. A good solution that fits with experimental data would have =0.9-1.2.

A series of random noise is added in the TOF data I(t) (raw), evaluated from eq. (4), to simulate the real data collected by the detector. Because of that, there is a fluctuation of the reconstructed I(d) at sampling points. To reduce this random error, we simulated 5 raw TOF spectra and calculated the average solution over the individual reconstructed I(d). This protocol is used for both short and long pulse

diffractometers.

We generated the fl-matrix for a SEPD instrument using the input parameters resemble the original setup of SEPD [12], i.e., total neutron path length is 14 m, the scattering angle 0=75o, and the corresponded resolution Ad/d=5x10-3, moderator temperature 7=100 K, and the emission time distribution is assumed to be a Gaussian with full-width at half-maximum (FWHM) of 5 ^s to approximate the asymmetric function. We reconstructed I(d) using constrained least-square method, the good fit between

modeled curve and raw TOF curve and the unbiased residuals are shown in Fig. 6(a). The average result from 5 raw spectra (each has 3x106 counts) is shown in Fig. 6 (b). There is a very good agreement between original and the averaged reconstructed 7(d).

Fig. 6 (a) Comparison between the raw TOF spectrum of single crystal Al simulated for SEPD and the modeled spectrum obtained after reconstruction. The insert is the logarithm of TOF spectrum to clarify two small peaks near 13.8 ms and 15.9 ms. The residual spectrum is shown in the lower panel, with the variance of reconstruction /2=1 043. (b) The reconstructed I(d) is compared with original I(d). The insert is the I(d) in the region of d=0.5-1.05 A

In the case of the new diffractometer, the pulse-length is assumed to be i=1.0-1.6 ms, closely resemble to the source performance figures of LENS [4]. A simulated raw TOF spectrum for Al single crystal is shown in Fig. 7. The TOF spectrum loses the featured peaks of I(d) curve, due to the overlapped contribution from reflected neutrons at different wavelengths or emitted at different times from moderator.

M 4 % 2 E 0

t (m s

Fig.7 a) A simulated TOF spectrum (raw data) and a modeled TOF spectrum based on reconstructed I(d) for single crystal Al. The residual spectrum, defined as [7i(t)(modeled)-7i (t)(raw)]/[I;(t)(raw)]1/2, is also shown in the lower panel. The total number of counts is 3x106; a background level of 30 and a series of random noise are added in the simulated spectrum. The variance of reconstruction X2=1.014. b). Reconstructed I(d) compared with the originally input I(d). The reconstructed I(d) is the average obtained from 5 raw TOF spectra (each has 3x106 counts) with different random noise. The setup parameters are: #=85o, pulse length=1.6 ms, resolution

<4d/d=5x10-4. The time channel width is 4 |^s. The total number of counts is 3M. The insert is the logarithm of TOF spectrum.

5. Conclusions

The need for new types of sources of cold and very cold neutrons has been recognized previously. One of the most promising operational modes envisions low-frequency (several Hz) pulsed sources with cold moderators and long (~ms) pulse lengths coupled to new types of scattering instruments. The Spallation Neutron Source (SNS), the premiere pulsed source in the world, is planning a long-pulsed cold source for its second target station. European Spallation Source (ESS) will have only one long-pulsed target station. LENS is currently the only source in the world operating in a long-pulsed mode and thus is an invaluable resource for the evaluation of new instrument concepts.

The proposed instrument will have an impact in two ways. Firstly, it allows to research larger unit cells than could be studied on conventional instruments, even those at major national facilities. Secondly, it will serve as a prototype for future instruments located on long pulsed sources. SNS second target station and the ESS will need proven concepts for the instruments that they support. The results presented in this paper demonstrate the feasibility of utilizing the long-pulsed neutron source to construct a highly efficient, highly resolved TOF neutron diffractometer for advanced materials study.

Acknowledgements

This report was prepared by Indiana University under partial support from award 70NANB10H255 from the National Institute of Standards and Technology, U.S. Department of Commerce. The statements, findings, conclusions, and recommendations are those of the authors and do not necessarily reflect the views of the National Institute of Standards and Technology or the U.S. Department of Commerce. Construction of LENS was supported by the National Science Foundation grants DMR-0220560 and DMR-0320627, the 21st Century Science and Technology fund of Indiana, Indiana University, and the Department of Defense. Operations of LENS is supported by Indiana University.

References

[1] L. Liang, R. Rinaldi, and H. Schober (Ed.), Neutron Applications in Earth, Energy, and Environmental Sciences, Springer, New York, 2009.

[2] E. H. Kisi and C.J. Howard, Applications of Neutron Powder Diffraction, Oxford University Press, New York, 2008.

[3] MRS Bulletin, Harnessing Materials for Energy, 33 (2008).

[4] C.M. Lavelle, D.V. Baxter, M.A. Lone et al., Nucl. Instrum. Meth. A 587 (2008) 324.

[5] J.D. Jorgensen, J. Faber Jr, J.M. Carpenter, R.K. Crawford, J.R. Haumann, R.L. Hitterman, R. Kleb, G.E. Ostrowski, F.J. Rotella, T.G. Worlton, J. Appl. Cryst. 22 (1989) 321.

[6] http://www.pns.anl.gov/instruments/sepd.

[7] I. Baek, J.M. Carpenter, and E. Iverson, Nucl. Instrum. Meth. A 490 (2002) 522.

[8] S. Ikeda and J. M. Carpenter, Nucl. Instrum. Meth. A 239 (1985) 536.

[9] V. Dose, Rep. Prog. Phys. 66 (2003) 1421.

[10] VITESS web site: http://www.hmi.de/projects/ess/vitess