1. Presentation of the instruments

a) Historical

At the origin, ILL was equipped with four "powder" diffractometers : D1A (high resolution) and D1B (the first position sensitive linear detector), installed on a thermal neutron guide H22 in the guide hall; D2 (high flux) on the thermal neutron beam H11 in the reactor hall, and D4 (liquids) on the "hot" neutron beam H8 in the reactor hall. At the beginning of the 80's ("Second Souffle"), it was decided to replace D2 by two instruments : D2B (high resolution) and D20 (high flux). D2B was commissioned in 1984, but the construction of D20, which involved the new "microstrip" technique of position sensitive detectors, took a long way round. D4 was also upgraded (D4B, with two position sensitive detectors).

After the ILL shut-down, the reduction in budget led to transform some instruments in CRG. This was the case for half of D1A, which was then managed by the Swiss community up to 1998 (start of the swiss national source SINQ).

In 1997, when D20 went in operation, D1B was transformed in CRG-B (France + Spain).

b) Present situation of the 5 instruments

D2B - This high resolution powder diffractometer (2q_FWHM = 0.4° in the usual high flux mode, 0.1-0.2° in the very high resolution mode), equipped with a moving set of 64 ³He detectors, is scheduled for many short experiments (more than any other ILL machine last year). It is mostly devoted to the study of phase transitions, precise positionning of light atoms, determination of Debye-Waller factors, by Rietveld refinements. D2B produced the ILL's most cited publication (in 1990 on charge transfer in oxide superconductors), and also the most cited ILL publication since the reactor refurbishment (a 1995 Phys. Rev. Letters on Giant Magneto-Resistive materials, see highlights § 4a). An important upgrade (super-D2B project, see § 5c) is foreseen within the ILL millenium development programme.

D20 - This machine uses the world's first large-area microstrip position sensitive detector (1600 cells, Dq = 160°), and can collect complete diffraction patterns in less than a second, making it faster than synchrotron machines for many chemical kinetics experiments (the neutron machine can use much larger samples, e.g. typical of real batteries and chemical cells). The new possibilities offered by D20 are now only being appreciated, and the machine will be even more powerful with the new monochromators at present under construction. Unfortunately, D20 is still very much a prototype instrument, and the continued deterioration of it's microstrip elements during the year 1998, probably due to insufficient cleaning during assembly, required a long (˜ one year) shut-down and a major repair (see § 5a). It is hoped that D20 will operate again in spring 2000.

D4 - This instrument, installed on the hot source, is a liquids diffractometer, rather than a powder machine. It is being modernized with a bank of 9 high pressure He³ microstrip detectors (D4C project, see § 5a), which will increase its efficiency by more than an order of magnitude, making it a unique instrument for the study of small amorphous samples, such as experiments using isotope contrast. It is generally recognized that, even before this upgrade, D4 is the best machine in its category in the world (this is largely due to its exceptionnally high stability). Although one of the most heavily demanded of the ILL instruments, D4 will however continue to share a beam tube with IN1, and remain only 50 % scheduled.

D1A - This "half CRG" is at present 50 % scheduled for ILL users (and, up to 1998, 50 % CRG with the swiss community). It is the only ILL diffractometer available for neutron strain scanning, a field of increasing interest. Several industrialists (e.g. British Aerospace, Volkswagen, EDF, etc...) have conducted paid beam time experiments on D1A. There is currently a proposal by a consortium group of five UK University groups to rebuild a dedicated strain scanning instrument at ILL : this proposal is discussed within the present review (see § 5b). D1A has produced some of the ILL's most cited publications, but most high resolution Rietveld refinement experiments are now made on D2B.

D1B - This machine is now a 100 % CRG, run by a consortium associating the Laboratoire de Cristallographie (CNRS, Grenoble, France) and the CSIC/CICYT (Spain). D1B has produced more publications than any other ILL machine (according to library records), and has been heavily demanded for work in magnetism (structures and phase transitions) and chemical kinetics. Crystallographic textures are also studied on D1B (an Euler craddle is available). However, it is supposed that the most demanding experiments in these fields will be done in future on D20. D1B has been recently equipped by CNRS with in situ  thermogravimetry, allowing to measure simultaneously diffraction diagrams and mass variations under gas pressure (H₂, N₂, O₂) at elevated temperature.

c) Staff

During the reactor cycles, the instruments are generally run 24 hours a day by two ILL scientists and one technician; exceptions are : D4, which has only one attached ILL scientist (D4 shares a beam with IN1); D1A, which also has only one attached ILL scientist devoted to the strain scanning activity (the second responsible left at the ending of the swiss CRG); D1B, where the staff is supplied by the CRG consortium (one scientist and one technician from CNRS, Grenoble, and one scientist from Spain).

A few months ago, the staff situation in powder and liquid diffraction was rather bad, because the very experienced instrument responsibles of D2B (Emmanuelle Suard) and D4 (Henry Fischer) were leaving ILL. The situation could be stabilized for D2B by a staff scientist contract for E. Suard, but not for D4. The departure of Henry Fischer in August caused some difficulties for the D4C project (see § 5a); the latter is now supervised by Pierre Palleau, newly appointed as an engineer, but with long experience on D4. It was also possible to have an overlap between the departure of Henry Fischer and the arrival of the new scientific responsible, Gabriel Cuello, owing to the temporary help of Miguel Gonzalez (both are experienced D4 users).

2. Statistics

a) Experimental request and realization

The number of scheduled experiments and the required and allocated beam time for each of the reviewed instrument are given in Appendix 4 (the CRG beam time is not included for D1A, and for D1B in 1997 and 1998).

Around 200 experiments are performed each year on the ILL powder instruments, which corresponds to more than 25 % of the total number of ILL experiments. Indeed, the average duration of an experiment on a powder instrument is short, particularly on D2B : 2.5 days; it is larger on D4 (4.5 days); it has increased on D1A (from 2.7 days in 1996 to 4.4 days in 1998) because more stress experiments are performed.

The ratio of requested to allocated beam time varies from instrument to instrument; on the average, it is 1.8, close to the average value for ILL (1.9). It is particularly high (2.2) for the diffractometer for liquids, D4, which operates only 50 % of the year; D2B is also very demanded (2.0).

The loss of beam time is small, generally between 1 and 6 % (exceptions : D20 in 1998 : 10 %, D4 in 1996 : 11 %).

ILL receives approximately 300 individual users per year for experiments on powder and liquid diffraction (see § 3).

b) Publications

The number of publications arising from the 5 reviewed instruments are summarized from 1996 in Table 3 of Appendix 4.

An analysis of a set of 231 publications made in 1996 and 1997 showed that somewhat more than half (132) are in reviews which may be considered in the domain of physics (30 Phys. Rev.B, 8 PRL, 9 JMMM, 47 Physica B+C, 18 J. Phys. Condensed Matter, 3 Europhysics Letters, etc...), and ˜ 40 % (86) in reviews related to chemistry in a broad sense (including crystallography, metallurgy and materials science : 11 J. Solid State Chem., 33 J. Alloys & Compounds, 7 Chemistry of Materials, 6 Acta Cryst. + J. Appl. Cryst., 3 J. Non Cryst. Solids, etc...).

The comparison of the overall ILL publications over the past years shows that the rate of publications in 1997 recovered the one before refurbishment (516 in 1997 compared to 559 in 1990); for the powder and liquid diffraction, recovery was already practically attained in 1996 (122 publications in 1996 compared to 130 in 1990).

The delay between experiment and publication varies considerably from one case to the other; for the whole ILL, this is estimated to be 2 years. The delay is usually shorter for powder experiments, with many published within one year, but precise statistics are not available.

The average "success rate" of a powder diffraction experiment is similar to the case of single crystal diffraction : about 50 %, with some publications requiring more than one experiment (and some experiments leading to several publications).

At the request of the Scientific Council, a measurement of the impact of ILL publications was made in 1998. The reference period is 1981-1997 inclusive, corresponding to 5085 papers, including work by both ILL scientists and users.

63 publications were found to be cited at least 100 times (for comparison, "normal" papers are cited only 10-15 times), from which 11 were related to the powder diffraction technique (see list in Appendix 3)

These include the ILL's most cited publication (in 1990, by Cava et al, a Physica C on charge transfer in oxide superconductors, 646 citations), and also the most cited ILL publication since the reactor refurbishment (in 1995, by Hwang et al, a Phys. Rev. Letters on Giant Magnetoresistive materials, 218 citations) (see highlights, § 4a and Appendix 3).

In fact, 7 of these 11 papers are related to structutal studies on the high T_c cuprates, one on giant magnetoresistive perovskites, one on the localization of benzene molecules in zeolithes, and two review papers on powder neutron diffraction and on the structure of liquid semiconductors. 6 of the 11 most cited papers correspond to work on D2B, 3 on D1A, 1 on D4 and 1 on D1B.

3. Users

a) General characteristics of the user community

The number of different users on each instrument, from October 1994 to April 1998, are given below :

The 4 powder diffractometers, especially D1B and D2B, serve a large community, about 3 times as great as on more specialised instruments. D4 only operates half time.

A more detailed analysis of the user community for the period 1995-1998, and a list of most frequent users for each instrument, are given in Appendix 4.

The percentage of "new" users in 1998 (i.e. which did not come since the refurbishment in 1995) is ˜ 20-25 %.

b) Users Survey

A questionnaire was sent to 237 users. The 46 answers received (˜ 20 %) are summarized in Appendix 2.

37 answers were received from regular users (performing at least one experiment per year), and 9 from occasional users.  Generally, they classify their work as fundamental research (31), but also as materials characterization (20), applied research (7) and technical development (7).

Surprisingly, nearly half (20) did not work on other neutron sources. From the others, many comments concern the complementarity between time-of-flight and constant wavelength diffractometers (see § 6b).

The majority of users found the instrument performance good or very good (30); but several improvements are clearly required, in particular :

- of course solve the detector problem on D20;

- install sample changers at high or low T;

- improve the electronics and data acquisition of D1B;

- install oscillating radial collimators on D20 and D1B;

- on all instruments, improve the software / user interface and in particular the immediate visualisation of data.

All users consider ILL powder diffraction instruments as essential for their future research, especially if special sample environments are developed. Super high flux is increasingly important (small / short lived / diluted samples). The ultimate resolution of the high resolution instrument must be improved (upgraded or new D2B).

A requirement of the Solid State Chemistry community is that ILL should give more time for short standard (at room temperature, in air) test or characterization experiments, out of Scientific Council sessions.

4. Science

a) General

Neutron powder diffraction (taken in its broad sense, i.e. including disordered systems) covers an extremely wide range of scientific domains, from very fundamental "academic" physics to industry related problems. All five ILL instruments were at the origin of important scientific achievements. A list of selected highlights is given in Appendix 3.

D2B is mostly involved in the detailed study of crystal and magnetic structures, and their dependence with temperature, pressure, etc..., by high resolution Rietveld refinements :

- nuclear and magnetic structures, lattice-magnetic interactions (Jahn-Teller distortions) in new oxides, including high T_c superconductors and related compounds, giant magnetoresistive oxides (see highlights), spin ladders, spin Peierls compounds;

- commensurate and incommensurate magnetic structures, magnetic frustration in rare earth, actinide and transition metal compounds;

- superionic conductors : order-transitions, in-situ study of insertion, ...

- crystal structure of clathrate systems, of fullerides, new metastable high pressure forms of ice (see highlights),...;

- detailed crystal structure, subtle phase transitions and H-bond studies in molecular systems (hydroxides, sulfates, oxalates,.);

- insertion in zeolites, site occupation in cements, interstitial solid solutions and compounds, localisation of hydrogen or deuterium, ...

- structural studies of quasi-crystals and approximants, decagonal phases.

D1A has two domains of scientific activity : (i) powder structure determinations and refinements, similar to those made on D2B, and (ii) internal stress studies on materials of technological interest, with the aim of predicting fatigue threshold and/or validating finite element models.

Internal stress studies in bulk materials include residual stress determination in two-phase materials (e.g. metal matrix composites reinforced by carbon or SiC fibers), in gradient materials, stress relaxation during deformation, relation between residual stress, grain shape and texture, ...

Many studies are also made of residual stresses in interfaces (e.g. weldings) and near surfaces (coatings, shot-peened materials, laser treated surfaces,...).

D20 (and to a less extent D1B) allows to study the same type of problems than D2B when a very high counting rate is required (but one is limited to medium resolution) : small samples, weak reflection lines, fast measurements (required for example when studying an unstable phase). But it gives also access to new possibilities, in particular structural studies of thin films or adsorbed layers, kinetic measurements and stroboscopic study of cyclic phenomena.

The in situ  investigation and optimisation of synthesis or material processing (e.g. industrial process for sintered magnets Nd₂Fe₁₄B), and the study of kinetic reactions (dehydration or phase transformations in cements, nitridisation, hydrogen absorption-desorption in intermetallics,...) are new fields, initiated on D1B, and where a breakthrough  is now made possible by the high intensity and the large position sensitive detector of D20.

Within its 2 years of operation, D20 has already allowed important scientific achievements, such as :

- in situ study of the crystallization of intermetallic amorphous ribbons, showing the detailed sequence of phase transformations which involve several metastable intermediate stages (at this occasion, new intermetallic phases were discovered);

- extremely weak incommensurate magnetic transitions;

- surface corrosion studies of zircaloy (where the oxide film thickness was as small as 1 µm);

- melting/freezing of adsorbed water;

- orientational alignment of plate-like particles under flow or electric field and study of the relaxation processes by a stroboscopic technique.

D4 is devoted to the determination ol local atomic arrangements and structure factors in liquids and amorphous systems :

- study of the many-body contribution to the interatomic potential in simple monoatomic liquids;

- structure of binary liquids, in particular under pressure, near the critical point or in the supercritical conditions, in order to validate numerical simulation models;

- structure of complex liquids (aqueous and non-aqueous electrolyte solutions, molten metal-salt mixtures,...);

- structure of liquids in confined geometry (e.g. water in porous matrix, interlayer water in clays);

- local order in molten quasicrystals;

- local order in amorphous solids and glasses (fast ion conductors, borate, silicate, selenite or oxide glasses, ...), by the method of isotopic substitution;

- structural studies of nanocrystalline materials (carbon nanotubes, nanocrystalline oxides,...).

Part of the neutron beam time of D4 (˜ 10 %) is used in classical powder diffraction, on Gd or other strongly absorbing rare earths, to use the small wavelength, in particular for magnetic structure studies.

b) Evolution of subjects and scientific areas for future

Scientific subjects studied by powder diffraction are evolving continuously and rapidly.

In particular, a large increase of residual stress studies, made essentially on D1A, has been observed since the ILL  refurbishment : from 28 days beam time (6 experiments) in 1995 to 85 days beam time (16 experiments) in 1998. Studies on GMR and charge-ordered perovskites have "exploded" these last years and, for example, occupied ˜ 30 % of the beam time on D2B in 1997 and 1998. On the other hand, neutron diffraction studies on surfaces have decreased, people in this field go now preferentially to the synchrotron.

As stated by one user within his answer to the questionnaire, "whatever the hot topic is in solid state chemistry, it is likely that powder (neutron) diffraction will play a vital role in the (next) 20 years; one has only to think of the topics that have come in and out of fashion : ionic conductors, zeolites, high T_C superconductors, colossal magnetoresistive materials,....".

Scientific areas for the future mentionned in the User's Survey are :

- in situ preparation and kinetics,

- glaciology, geology, palaeontology,

- structure and phase transitions in molecular compounds relevant to pharmaceutical research,

- defects in structures,

- high pressures (at least up to 60-70 GPa),

- materials of industrial interest,

- nanostructured systems : carbon nanotubes, liquids in confined geometries,...

5. Instrumentation

a) Microstrip detectors

The rebuild of the D20 microstrip detector is proceeding according to plan, with users expected next May. Although the reasons for the destruction of the previous detector are not entirely known, it is clear that arcing due to poor contacts, interactions between high tension supplies to adjacent plates, and insufficient clean glass and gas, could all have contributed.

All three of these problems have been adressed in the new detector, with good plated contacts, modified electrode geometry, and reduced high tension interaction between plates. The new plates have been tested in a small prototype detector, and the complete set of plates for D20 ordered.

These specific problems are less important for the new D4C microstrip detectors, since the modular design uses only one plate for each of the 9 detectors. Some delay has however been experienced because a few of the plates delivered for D4 were below specification, and had to be reordered.

Another complication was the departure of the instrument responsible in August; these staff problems have now been solved (see § 1c). The new mechanics of D4C is at present being assembled, and start-up of the new instrument is planned for May 2000.

b) The high precision Strain Scanner project

D1A presents unique advantages for strain scanning measurements : high flux and high precision; the latter is due to the high angular resolution of D1A, and to the coupled use of a radial collimator and a position sensitive detector. But the present instrument has also several drawbacks, in particular lack of space around the instrument and fixed take-off angle.

Therefore a consortium group of seven UK university groups, led by P. Withers (Manchester Univ.), proposed to rebuild a new strain scanner which should be one of the world leading instruments. The new instrument would be installed on the thermal guide H22, behind D1B.

The major scientific aim is to validate finite element calculations on industrial devices. One will concentrate on the case of non-textured samples of high crystalline symmetry, where one diffraction line per phase gives the required information (more complex cases should be studied on a time-of-flight instrument).

The eventual opening of the instrument to crystallographic studies (for which 64 days beam time were allocated by ILL Scientific Council Subcommittees in 1998) has not been decided yet.

The strain scanner project was classified as high priority by the Powder Diffraction workshop and the ILL Scientific Council. The funding application to UK authorities is in progress.

c) Super-D2B

The ILL's high resolution powder diffractometer D2B has now completed 15 years service; it is still very demanded (requested-to-allocated beam time ratio ˜ 2), has a high scientific production, but is now competed by new instruments in other neutron sources, in particular from ISIS. Moreover, the use of D2B in its highest resolution mode, which is increasingly necessary with new problems and small samples of new materials under extreme conditions (e.g. subtle changes in the splitting of superstructure lines), is limited by the relatively low flux and long duration of experiments (typically 8 to 10 hours for a scan).

Super-D2B is therefore a proposal to improve the efficiency of this instrument by an order of magnitude. The project, presented in the frame of the ILL Millenium Development Programme, contains two parts : (i) a new high resolution set of detectors, and (ii) a new double-focussing (horizontal and vertical) composite monochromator. High resolution will be achieved in the horizontal plane with an array of 128 mylar collimators. The necessary (lower) resolution in the vertical direction will be obtained by charge localisation on 128 high pressure 300 mm linear wire ³He detectors. This vertical resolution will be used to measure a large part of the diffraction cone, rather than the usual small section in the horizontal plane. Compared to the present instrument, the detector solid angle will be increased by a factor 6.

The D2B upgrading was given first priority by the Powder Diffraction Workshop and the Subcommittees 5a and 5b (structures). Nevertheless, the Scientific Council of april 1999 requested more precision on the scientific case, and the Instrument Subcommittee raised questions on the technical choice. Therefore, the decision has not been taken yet.

d) Data acquisition and software

The ILL Powder Diffraction Workshop dedicated a session to data treatment and software development (see Appendix 1). The main statements are summarized below :

- In powder diffraction, the software for data treatment is very much developed. The reason is that the concerned community is quite large and many people (including commercial companies) contribute to that development. In what constant-wavelength neutron diffraction is concerned, most of the structural work is largely satisfied by the existing programmes.

- The specificity of neutron diffraction makes that only some programmes, using the Rietveld method (CCSL, FullProf, GSAS and RIETAN), have enough features to handle complicated situations (in addition to the conventional crystallographic problems) : anisotropic peak broadening due to defects, spin correlation lengths or crystallite size distributions, incommensurate magnetic structures, rigid body and soft constraints refinements, multi-pattern (treatment of heterogeneous data) capabilities, etc... Progress is needed in more sophisticated cases : peak-shifts and asymmetry due to some kind of defects, partial treatment of diffuse scattering, incommensurate/composite crystal structures, complex form factors of plastic crystals, etc...

- The free distribution of the executable codes is very important. User-developer feedback is the only way for correcting bugs and to compare the relative performances for each particular problem. The role of ILL in this respect is to contribute and support this kind of activity.

- The development of software for treating disordered materials is not so advanced as those handling conventional crystallographic problems. The concerned community may improve this situation.

- The improvement of the present software should be concentrated in : documentation, Graphical User Interfaces, simplifying the input files using a command-oriented language and totally free format.

- An important task that should be undertaken by the ILL is the development of the software for handling the instruments and for in situ preliminary data treatment : visualisation, integration, and generation of data in appropriate formats. This is specially important for new instruments producing a great amount of data (D20 for instance).

- The improvement and/or the development of new diffraction software by scientists take always longer than what was expected as starting. The establishment of free Collaborative Diffraction Software Development Groups (CDSDG) working with public source codes are needed to improve and to increase the capabilities of present software for data analysis.

6. Comparison with other neutron and with synchrotron instruments

a) Comparison with other constant wavelength neutron diffractometers

During the workshop, instruments on present or future medium flux neutron sources were presented and compared with ILL powder diffractometers.

At LLB/Saclay (flux 5 times less than ILL) :

- G4.1 is analogous to D1B, but on a cold neutron source : it has a comparable flux with 800 cells instead of 400,  and a slightly better resolution than D1B at l = 2.4 Å; but the instrument lineshape is not so good, and short wavelengths (1 - 1.5 Å) are not available.

- 3T2 is analogous to D2B, but of course with less flux; the resolution is comparable to that of the "conventional" D2B, but less than the "high resolution" D2B.

- G6.1 is a long wavelength (l = 4.8 Å) instrument, largely devoted to high pressures, and has no equivalent at ILL.

On SINQ (where the flux is presently 20 times less than ILL) :

- DMC is a cold neutron 2-axis instrument, with position sensitive multidetector; it has a much lower flux than D1B, but a better background (due to oscillating radial collimator).

- HRPT, a new high resolution powder diffractometer, presently commissioning, has a new ³He 1600 cells multidetector, and a high resolution Dd / d = 2.5 10^-3. It will be competitive with D1A and D1B.

On FRM-2 (Munich), where the flux should be near one half of that of the HFR-ILL, an instrument of the type of D2B, with supermirror guide and 80 detectors, is foreseen.

In conclusion, the best powder instruments on medium flux sources are far from D2B and D20 in flux, but, owing to the progress in neutron optics, become competitive with D1A and D1B. For liquids, the only instrument on reactor source is 7C2 at LLB/Saclay, also on a hot source; but D4 remains better, in flux and in scattering vector range available (DQ_max = 30 Å^-1 on D4, 17 Å^-1 on 7C2). These statements were confirmed by the User's Review.

b) Comparison with Time-of-Flight instruments

Powder diffraction is one of the great successes of pulsed neutron sources. The comparison between ILL and ISIS powder and liquid diffraction instruments, in particular between D2B and HRPD, and between D4 and SANDALS, was largely debated in the workshop and in the answers to the questionnaire.

Generally, it was found that time-of-flight (TOF) and constant wavelength diffractometers (CWD) are complementary : each type of instrument has advantages in certain circumstances, and both are often needed for a full scientific coverage.

High resolution diffractometers. Both leading instruments (D2B and HRPD) are found excellent by users, with, in the average, similar resolution and intensity. This can be understood because the counting rate is proportional to the product of the detector solid angle by the source solid angle; in CWD (compared to TOF diffractometers), the latter is ˜ 10 times larger because of the efficient focusing, but the former is ˜ 10 times smaller because of mechanical constraints. HRPD gives access to a larger q range (therefore to smaller d spacings) and has better resolution than D2B at high q. Also, unlike D2B, HRPD has a constant, synchrotron-like resolution Dq /q in all the q-range accessible in back-scattering. However, obtaining low-q data and merging them with high-q data is still a cumbersome and unreliable procedure on HRPD. Therefore, D2B remains better for magnetic studies, also in virtue of somewhat superior flux. D2B has also a simpler peak shape and a better background. At ISIS, a new high resolution diffractometer, OSIRIS, installed on the cold source, is starting to operate : it should be much better than HRPD for magnetic scattering and might compete with D2B.

Data corrections and analysis are considered by most users as much more difficult on spallation sources, because of complicated asymmetrical peak shapes, difficulties in scaling overlapping parts of the pattern (obtained on the different detector banks), and the necessity to divide raw data by the incident spectrum, which must be determined very precisely. But these difficulties can be overcome by carefull work. Improvements to existing Rietveld-TOF software (GSAS, CCSL) and the addition of the TOF option to the popular FULLPROF package are expected to give a major contribution to the TOF data analysis capabilities.

High flux diffractometers. POLARIS (at ISIS) has higher resolution, but lower counts and higher background than D20. But the new ISIS instrument, GEM, equipped with 7 detector banks (10 times more detecting angle than D20), should have a flux comparable to that of D20. Because of the fixed geometry, TOF instruments are much better adapted for high pressure crystallographic studies on powders; at ISIS, pressures of 30 GPa at room temperature and of 15 GPa at 90 K have been obtained. (But magnetic structure studies under pressure are performed on CWD, e.g. G6.1 in Saclay).

Diffractometers for liquids. Although instruments on pulsed spallation sources allow higher q values to be obtained (e.g. 40 Å^-1 on SANDALS), CWD instruments and especially D4 are generally preferred for diffraction studies on liquids, because of the much better stability of source and detector (a precision of 5 10^-4 on the measured cross-sections is required for studies involving isotopic substitution) and the more easy data treatment. In two cases, SANDALS was preferred to D4 : (i) larger flux in a specific q range (0.3 - 3 Å^-1), and (ii) better for H substitution. But TOF diffractometers for liquids cannot be used for rare earths (because of resonances), and more generally are more difficult for absorbing systems. A lot of work has still to be done to improve data treatment software for disordered materials studied on pulsed spallation sources.

c) Neutrons versus Synchrotron Radiation in Powder Diffraction

A session was devoted to this subject during the workshop.

The characteristics of high energy synchrotron radiation (parallel beam optics, vertical focusing) allow considerable advantages in powder diffraction : very small instrumental linewidth (e.g. D2q_FWHM = 0.006° on BM16 at the ESRF) and very accurate scattering angles (± 1 second on BM16), well defined lineshapes, high statistics, weak absorption,... Kinetic experiments can be performed at the time scale of ˜ 0.1 s, but the demand is not very high. Damaging of the sample by the beam is generally not a big problem in the case of high energy X-rays, because the absorption is small.

In conclusion, neutrons and synchrotron radiation are complementary : very generally speaking, if X-rays are better than neutrons for structure determination, neutrons are better than X-rays for structure refinement, in particular for obtaining precise site occupancies and precise atomic positions of light elements in the presence of heavy elements (the statistics of published structural work with Rietveld refinement give 3276 cases with neutrons, 2557 with conventional X-rays and 126 with synchrotron X-rays). This is due to several factors : the weak neutron-matter interaction, the easiness of corrections, the independence of neutron scattering length versus q.

X-rays are better for small samples (although the modelisation of granularity and of preferred orientation may be difficult), and neutrons are better for large samples as required for example in real-time diffraction studies of chemical or electro-chemical processes. A number of scientific problems call for a simultaneous use of both radiations, and we expect to see more and more coupled neutron - synchrotron radiation structural studies.

Neutrons are of course unique for magnetic structure determination. Synchrotron radiation is leading in the field of high pressures, because of the smaller samples : phase diagrams and structures could be studied at pressures as high as 180 GPa (in comparison to P_max = 40 GPa with neutrons in Saclay), but some aspects cannot be studied with X-rays : magnetic structures, precise structural informations on H and D, many aspects of diffuse scattering.

Because of the weak absorption of high energy X-rays, and the high accuracy on scattering angles, synchrotron radiation offers very attracting perspectives in the field of strain scanning; but many difficult aspects have to be overcome, due in particular to the needle shape of the gauge volume.

Concerning diffraction by liquids, X-rays allow studies in a large q domain for elements ranging from Z = 30 to 50. But there is never enough contrast to obtain partial structure factors in diatomic liquids by using the anomalous effect : the only technique at present is neutron scattering with isotopic substitution.

7. Conclusion : the powder diffraction instrument profile of ILL

Powder diffraction (including conventional X-rays, synchrotron radiation and neutrons) is a technique with increasing applications in many fields of Science. It is one of the first techniques to be applied to new materials, which are often available as impure samples and in small quantities.

If the powder and liquid ILL diffractometers were the best in their categories since the origin, the progress in neutron optics has allowed some medium source instruments (present or future) to become competitive with D1A and D1B. Moreover, the great success of spallation sources in the field of powder neutron diffraction (see § 6b) has challenged the supremacy of ILL in this field (with the exception, surprisingly, of the diffractometers for liquids). Therefore, the upgrading and modernization of the ILL powder diffraction instruments is a priority, in order to optimize the use of the high flux source.

The ILL powder diffraction instrument profile for the future was largely discussed during the workshop. It was generally agreed that the minimum required is :

- a "fast" instrument for kinetics (with fixed multidetector, medium / good resolution) : it was D1B, it is now (and it will be) D20;

- a "high resolution" instrument for structures : it was D1A, it is now D2B, it should be in the future super-D2B (or a new instrument) with very high resolution.

They should be accompanied by two more specialized instruments :

- a diffractometer for liquids, with fixed short wavelength neutrons, installed on a hot  source : it was D4, it will be D4C;

- a high resolution strain scanner.

With this port-folio, it may be remarked that the number of "true" powder diffractometers at ILL is smaller than at ISIS (4) and at LLB (4).

Although their achievements are now catched up by some medium source diffractometers, it was generally agreed that D1A and D1B will still remain usefull instruments for the future, especially for demanding sample environments (low and high temperatures, high pressures, in situ electrochemical experiments,...). In particular, it is recommended that the 50 % scheduled D1A instrument should remain operational at least up to the repair of D20 and the upgrade of D2B.

According to several users and speakers at the Workshop, there is a gap in the instrument suite described above : a high intensity instrument for structure resolution, with good Rietveld refinement (i.e. intermediate between D20 and D2B) is missing; this requires a high resolution option on D20, with focusing Ge monochromator and large take-off angle.

Other demands are :

- a powder diffractometer on cold source for magnetic structure studies; but D16 could partially fulfill this role;

- a high resolution powder diffractometer on hot source with fixed short wavelength, for good structural refinements at high q, precise determination of Debye-Waller factors, study of disorder (diffuse scattering), etc... ; but in the present state of the art, it would be difficult to have a good monochromator with high take-off angle.

For the future, ILL should think on what could be its best high resolution powder diffractometer. It was in particular proposed (P. Radaelli) to consider the case of a TOF machine installed on a reactor, running in energy dispersive mode; such an instrument would present several advantages, compared to a TOF on spallation source (symmetrical line shape), or to a CWD (fixed geometry, very large detector solid angle, no higher energy contamination).

One should also develop the use of polarized neutrons; satisfactory tests of He³ filter and of CRYOPAD (3-dimensional polarization analysis) have recently been made on D1B.

Charles de NOVION

Chairman of ILL Powder Diffraction Instruments Review

With the collaboration of J.C. GOMEZ-SAL, P. RADAELLI,

J. RODRIGUEZ-CARVAJAL, A. HEWAT and P. CHIEUX

Appendix 3 : HIGHLIGHTS AND IMPACT OF ILL PUBLICATIONS

Most cited powder and liquid diffraction ILL publications

Are indicated the ranking in the overall ILL impact list, the number of citations in the reference period 1981-1997 inclusive, the instrument (in parenthesis), the authors, the year of publication (in parenthesis), the precise reference and the title of the paper. Only publications bearing the  ILL name are considered. (For an analysis of all 5085 ILL papers published between 1981 and 1997, please refer to the ILL citation page http://www.ill.fr/dif/citations/).

N°1 - 646 (D2B) Cava, R.J., Hewat, A.W., Hewat, E.A., Batlogg, B., Marezio, M., Rabe, K.M., Krajewski, J.J., Peck, W.F. and Rupp, L.W. (1990) Physica C. 165, 419. "Structural anomalies oxygen ordering and superconductivity in oxygen deficient Ba₂YCu₃O_x".

N°4 - 466 (D1A) Capponi, J.J., Chaillout, C., Hewat, A.W., Lejay, P., Marezio, M., Nguyen, N., Raveau, B., Soubeyroux, J.L. and Tholence, J.L. (1987) Europhysics Letters. 3, 1301. "Structure of the 100 K superconductor Ba₂YCu₃O₇ between 5-300 K by neutron powder diffraction".

N°10 - 218 (D2B) Hwang, H.Y., Cheong, S.W., Radaelli, P.G., Marezio, M. and Batlogg, B. (1995) Physical Review Letters. 75, 914. "Lattice effects on the magnetoresistance in doped LaMnO_3".

N°18 - 195 (D1A) Fitch, A.N., Jobic, H. and Renouprez, A. (1986) Journal of Chemical Chemistry. 90, 1311. "Localisation of benzene in sodium-Y zeolite by powder neutron diffraction".

N°19 - 191 (D2B) Kaldis, E., Fischer, P., Hewat, A.W., Hewat, E.A., Karpinski, J. and Rusiecki, S. (1989) Physica C. 159, 668. "Low temperature anomalies and pressure effects on the structure and T_c of the superconductor YBa₂Cu₄O₈ (T_c = 80K)".

N°28 - 169 (D2B) François, M., Junod, A., Yvon, K., Hewat, A.W., Capponi, J.J., Strobel, P., Marezio, M. and Fischer, E.W. (1988) Solid State Communications. 66, 1117. "A study of the Cu-O chains in the high T_c superconductor YBa₂Cu₃O₇ by high resolution neutron powder diffraction".

N°45 - 129 (D1A/B) Rodriguez-Carvajal, J. (1993) Physica B. 192, 55. "Recent advances in magnetic structure determination by neutron powder diffraction".

N°48 - 125 (D2B) Bordet, P., Capponi, J.J., Chaillout, C., Chenavas, J., Hewat, A.W., Hewat, E.A., Hodeau, J.L., Marezio, M. and Tholence, J.L. (1988) Physica C. 156, 189. "A note on the symmetry and Bi valence of the superconductor Bi₂Sr₂CaCu₂O₈".

N°49 - 125 (D1A) Roth, G., Heger, G., Renker, B., Pannetier, J., Caignaert, V., Hervieu, M. and Raveau, B. (1988) Physica C. 153-155, 972. "Crystallographic study of the tetragonal high-T_c-superconductor YBa₂(Cu_0.95Fe_0.05)₃O_7".

N°58 - 105 (D4) Enderby, J.E. and Barnes, A.C. (1990) Reports on Progress in Physics. 53, 85. "Liquid semiconductors".

N°60 - 102 (D2B) Hewat, A.W., Capponi, J.J., Chaillout, C., Marezio, M. and Hewat, E.A. (1987) Solid State Communications. 64, 301. "Structures of superconducting Ba₂YCu₃O_7-d and semiconducting Ba₂YCu₃O₆ between 25 degrees C and 750 degrees C".