My image of RIETAN-2000 is a white dewy flower of lotus floating quiet on a muddy pond evolving poisonous air.
This is the reason why the present page has a background of pure white.

I am proud to announce that a versatile pattern-fitting system, RIETAN-2000, for angle-dispersive X-ray and neutron powder diffraction was released on August 1, 2000 and is now available for download. It is a descendant of RIETAN-94/97beta, working on multiple architectures. Numerous new features were injected into RIETAN-94 to change it into a heavy-duty pattern-fitting system of high performance.

The major upgrade of RIETAN-2000 underlines my continued commitment to the powder diffraction community and my intent to further push the limits in structure refinement by powder diffraction.

RIETAN-2000 resulted from our struggle for more than three years, which were necessary to develop and polish up it. I apologize to users of RIETAN-94 and RIETAN-97beta for the long delay in releasing it.

RIETAN-2000 is currently copyrighted but distributed free of charge with its source code not open to the public; we wish to control the development and future of RIETAN by ourselves.

Whenever original results acquired with RIETAN-2000 are published in journals, proceedings, facility reports, etc. or reported in scientific meetings with abstracts, the program name "RIETAN-2000" should explicitly be stated. Furthermore, cite the following paper:

• F. Izumi and T. Ikeda, Mater. Sci. Forum, 321-324 (2000) 198-203.

Note that "321-324" is volume numbers while "198-203" is the range of pages. Only giving credit to RIETAN-2000 for Rietveld or Le Bail analysis is fine in the cases of abstracts, short reports, and so forth with limited spaces, e.g., as follows:

Structure parameters of CaTiO₃ were refined by the Rietveld method from the X-ray diffraction data with RIETAN-2000.

Although RIETAN-2000 has exhaustively been tested with intensity data of many compounds, it may still contain a fair number of bugs. RIETAN-2000 is distributed in the hope that it will be useful and reliable. It is, however, provided "as is'' without any warranty since I have a number of commitments on my time. There is no charge for the program, per se, whereas technical service concerning it is not free.

3.1 Four kinds of pattern fitting

RIETAN-2000 is a software package applicable to the following five purposes:

1. Simulation of X-ray and neutron diffraction patterns
2. Pattern decomposition by Le Bail refinement
   1. with initial integrated intensities estimated by Wilson's statistics
   2. with the same initial integrated intensities (=100)
   3. with initial integrated intensities input from a file *.ffi
      • and varied during least-squares fitting
      • and kept constant during least-squares fitting (strictly speaking, different from Le Bail analysis)
   4. into which a partial structure is introduced
3. Refinement of lattice and structure parameters by the Rietveld method
4. Whole-pattern fitting based on the maximum-entropy method (MEM)
5. Local pattern decomposition by individual profile fitting (a rather outdated technique)

In pattern fitting 2-5, a premier and robust engine for nonlinear least squares by Gauss-Newton, Fletcher's modified Marquardt, and Powell's conjugate-direction methods can be utilized, permitting stable convergence in most refinements. The use of the conjugate-direction method, also known as the direction set method in multi-dimensions, often makes it possible to escape from local minima easily and automatically. It is relatively slow but excels at analyzing complex structures such as organic compounds and micro-porous materials in final stages.

Since the Rietveld method is now very popular, it is unnecessary to explain it here. Refer to the following books if necessary:

• "The Rietveld Method," ed. by R. A. Young, Oxford University Press, Oxford (1995).
• F. Izumi, "Applications of Synchrotron Radiation to Materials Analysis," ed. by H. Saisho and Y. Gohshi, Elsevier, Amsterdam (1996), Chap. 7.

The Le Bail method provides the opportunity for estimation of integrated intensities in the absence of any structural model with a simple procedure proposed by Rietveld. The resulting integrated intensities serve for ab initio structure analysis by the heavy-atom (Patterson), direct, and Monte Carlo methods. RIETAN-2000 is capable of outputting files for (1) improving integrated intensities for overlapping reflections and determining three-dimensional Patterson functions by the maximum-entropy Patterson method with ALBA and (2) construction of structural models with EXPO2004. To refine lattice parameters, this method excels the Rietveld method when the effect of preferred orientation or coarse particles is pronounced. This is a newborn function that needs to be tested and improved further. Nevertheless, we confirmed that RIETAN-2000 usually yield lower R factors than EXPO2004. Our first application of the Le Bail method to the ab initio structure analysis of HLS was reported in:

• T. Ikeda, Y. Akiyama, F. Izumi, Y. Kiyozumi, F. Mizukami, and T. Kodaira, Chem. Mater., 13 (2001) 1286-1295.

MEM-based Pattern Fitting, which will be abbreviated to MPF here, will be described briefly in Sect. 5.

Individual profile fitting deals with intensity data within a limited 2θ range. No constraints are imposed upon peak positions and integrated intensities of reflections in that range. Both of them are directly refined by a method of nonlinear least squares together with profile parameters.

3.2 ORFFE

The RIETAN-2000 package also includes ORFFE for calculating geometric structure parameters such as interatomic distances and bond angles:

• W. R. Busing, K. O. Martin, and H. A. Levy, Report ORNL-TM-306, Oak Ridge National Laboratory, Tennessee (1964).

In addition, it serves to impose restraints on geometric structure parameters in Rietveld refinements of compounds containing many atoms in their asymmetric units.

3.3 Other new features in RIETAN-2000

Other improvements in RIETAN-2000 are listed below:

1. Correction of surface roughness on the basis of models of Sparks et al., Suortti, Pitschke et al., and combination of Sparks et al. and Suortti models.
2. Split profile functions: pseudo-Voigt and Pearson VII functions of Toraya (1990).
3. Introduction of anisotropic profile broadening into the split profile functions.
4. Modified split pseudo-Voigt function applicable only to 'relaxed' reflections.
5. Representation of anisotropic profile broadening in the Thompson-Cox-Hastings pseudo-Voigt function (1987) by the procedure of Finger et al. (1994).
6. Partial profile relaxation (see Sect. 6).
7. Profile cutoff equal to (the peak intensity) multiplied by a constant given by the user.
8. Two peak-shift functions with forms of Legendre polynomials.
9. Composite background function: (background intensities input from a file) multiplied by Legendre polynomials.
10. Dynamic allocation of parameters included in the model function.
11. Analysis of intensity data measured with variable step widths and counting times.
12. Analysis of X-ray diffraction data measured with a goniometer equipped with a divergence slit to keep the irradiation width constant.
13. Use of else if and else blocks, Go to + *labels, and local integer variables in input files, *.ins (new Tink).
14. Improvement in the accuracy of 'observed' and calculated integrated intensities.
15. Correction for microabsorption (absorption contrast) in the quantitative X-ray analysis of mixtures using Brindley's procedure.
16. Output of binary files, *.fos, for the MPF and MEM/Rietveld methods.
17. Imposing nonlinear restraints on geometric structure parameters using outputs, *.ffe, from the extended version of ORFFE.
18. Assignment of five columns to each of a pair, A and 1000*C + S, for an atom designation in ORFFE.
19. Automatic generation of instructions 2 to calculate bond angles using a list of interatomic distances obtained with instructions 201.
20. Input of various parameters associated with displaying refinement/simulation results with Igor Pro.
21. Output of *.itx files for displaying Rietveld-refinement/simulation results with a command-driven interactive plotting program gnuplot.
22. Creating a CIF through an output file, *.lst with lst2cif (a separate program included in the RIETAN-2000 package).
23. Introducing lattice and structure parameters in *.res (output by EXPO2004) with the SHELX format into *.ins.
24. Automatic updates in lattice and structure parameters in *.vcs with the VICS format.
25. Elaborate batch files, DD.bat and DD2.bat, written for Windows NT/2000/XP + Command Prompt to execute related programs (ORFFE, Igor Pro, a text editor, tee, and less) sequentially after pattern fitting.

With all of these new functions and capabilities, RIETAN-2000 delivers superior application performance on the four pattern-fitting methods. I am confident that it will contribute fruitful results toward the completion of creative scientific work.

In the above title, 'Tink' (not 'Think' but 'Tink') should be regarded as a verb. I have found mouse-driven Graphical User Interface (GUI) to be even slower and more unwieldy to use for power users. I will, therefore, never adopt a GUI for RIETAN-2000. Instead, a powerful preprocessor called Tink was implemented as antithesis to the GUI. Tink converts *.ins, which is sufficiently self-explanatory, into a scratch file that contains only integer-, real-, and character-type variables input and dealt with actually by RIETAN-2000.

Tink was upgraded to New Tink for input files (*.ins) in RIETAN-2000. With New Tink, a straightforward and flat Character User Interface (CUI) has been realized. You can shortly learn how to use it. You must be deeply impressed at the resulting CUI, particularly, Fortran 77-like If, else, end if, and else if blocks, which may include parameters to calculate the model function. The unique method of inputting data is now a kind of a cultural heritage for RIETAN users.

5.1 Advanced technology of MPF

RIETAN-2000 offers a state-of-the-art structure-refinement technique alternative to the classical Rietveld method: MEM-based Pattern Fitting (MPF).

The revolutionary MPF technology was implemented in RIETAN-2000 to overcome limitations of Rietveld analysis. RIETAN-2000 is distinguished from other programs for classical Rietveld analysis by this elegant methodology. In MPF, PRIMA is required for MEM analysis in addition to RIETAN-2000. PRIMA is distributed as a part of the VENUS system. Whole-pattern fitting and MEM analysis are alternately repeated in an iterative procedure called REMEDY cycles, where the bias imposed by a structural model upon final electron/nuclear densities is minimized. The total number of electrons (X-ray diffraction) or the total coherent-scattering lengths (neutron diffraction) in the unit cell is fixed at that input by the user during the REMEDY cycles. Electron/nuclear-density distribution changes noticeably with dramatic decreases in R_B and R_F during the REMEDY cycles, which is a strong piece of evidence for the reduction in the bias of the structural model.

5.2 Serious flaw of the MEM/Rietveld method

The so-called MEM/Rietveld method necessarily suffers from errors due to the above bias because it simply deals with one set of 'observed' structure factors estimated on the basis of final profile and structure parameters determined by Rietveld analysis. Even if the structural model in the Rietveld analysis is essentially correct, chemical bonding cannot perfectly be represented only with structure parameters such as fractional coordinates and atomic displacement parameters. This is a reason why D syntheses are performed in studies of electron densities by single-crystal X-ray diffraction. Refinement of isotropic atomic displacement parameters in Rietveld analysis (anisotropic atomic displacement parameters cannot practically be refined even from synchrotron X-ray powder diffraction data) lowers the accuracy of the 'observed' structure factors. In addition, standard deviations of the 'observed' structure factors are estimated with an equation containing calculated structure factors, which gives a further bias toward the structural model in Rietveld analysis.

In conclusion, MEM is, per se, model-free whereas the MEM/Rietveld method is more or less model-dependent. The MEM/Rietveld method should never be confounded with MEM. Electron/nuclear densities can be determined only semiquantitatively with the MEM/Rietveld method. This expedient method is, therefore, unsound on its application to determination of electron/nuclear-density distribution though it is certainly effective for rough visualization of electron/nuclear densities.

I agree that the MEM/Rietveld method is more effective in modifying imperfect structural models than Fourier/D synthesis thanks to higher S/N ratios and non-negative densities obtained by MEM. However, it cannot give accurate electron/nuclear densities particularly when reflection overlap is serious because of low symmetry and insufficient instrumental resolution. Frankly speaking, the MEM/Rietveld method is nothing but iterative conventional Rietveld refinements where structural models are modified, if necessary, after visualizing electron/nuclear densities by MEM.

5.3 Predominance of the MPF method over the MEM/Rietveld method

In our more sophisticated structure-refinement technique, the REMEDY cycles partially overcome the defect in the MEM/Rietveld method, and crystal structures are represented with three-dimensional electron/nuclear densities within the unit cell. In other words, powder diffraction data are visualized by MEM with high S/N ratios like HRTEM images. This method is, hence, capable of modeling disordered atomic arrangements, chemical bonds, nonlocalized electrons, and anharmonic thermal motion more adequately than the conventional Rietveld method, where crystal structures are expressed with structure parameters such as fractional coordinates, occupation factors, and isotropic/anisotropic atomic displacement parameters. I hope that the appearance of the free software, RIETAN-2000 as well as the 3D visualization system VENUS, will give incentives to study electron/nuclear-density distribution in a variety of compounds.

5.4 Related publications

Applications of MPF to synchrotron X-ray powder diffraction data as well as my severe criticism again the MEM/Rietveld method have recently been presented in:

• F. Izumi and T. Ikeda, Commission on Powder Diffraction, IUCr Newslett., No. 26 (2001) 7-9.

• M. Merli, D. Capsoni, M. Bini, and V. Massarotti, Solid State Commun., 121 (2002) 193-198.
• J. B. Claridge, A. J. Fowkes, M. J. Rosseinsky, and I. D. Watts, Chem. Mater., 15 (2003) 1497-1504.
• J. B. Claridge, Y. Kubozono, and M. J. Rosseinski, Chem. Mater., 15 (2003) 1830-1839.
• M. Yashima and D. Ishimura, Chem. Phys. Lett., 378 (2003) 395-399.
• M. Yashima, K. Nomura, H. Kageyama, Y. Miyazaki, N. Chitose, and K. Adachi, Chem. Phys. Lett., 380 (2003) 391-396.
• T. Ikeda, Y. Akiyama, Y. Oumi, A. Kawai, and F. Mizukami, Angew. Chem. Int. Ed., 43 (2004) 4892-4896.

A very high level of goodness-of-fit, S, less than, say, 1.3 is indispensable to MPF and Le Bail refinement to derive integrated intensities for ab initio structure analysis.

Introduction of our original technique called partial profile relaxation into MPF more or less improves fits between observed and calculated intensities, particularly when coping with extremely asymmetric reflections in low-angle regions below 20 degrees. In addition, the finesse of profile relaxation enables us to analyze intensity data measured with a variety of geometries using characteristic X-ray, synchrotron radiation (SR), and neutron beams flexibly and simply using one of split profile functions implemented in RIETAN-2000.

Partial profile relaxation is introducible into Le Bail and Rietveld refinements, and MPF, where you will enjoy pronounced drops in R factors. The idea of profile relaxation is simple, but its effect is amazing!

The following operating systems are currently supported: Mac OS X, Microsoft Windows 98/Me/NT/2000/XP (hereafter abbreviated to Windows), and Linux.

RIETAN-2000 for Mac OS X is a Terminal (console) application for Darwin (= UNIX), which means that all the text files used when running it should have the UNIX format (end-of-line character: LF). A Carbon application is no longer distributed because it is slow in calculating.

RIETAN-2000 may run just fine on Windows 98/Me. Nevertheless, Windows NT/2000/XP is preferred to Windows 98/Me for decent performance. Windows NT/2000/XP is robust, supporting the Command Prompt that is much inferior to various shells in UNIX but not so bad. On the other hand, Windows 98 and, particularly, Windows Me have some serious problems including the absence of memory protection, memory leak, and MS-DOS Prompt of less features.

The Windows version is currently optimized for Pentium processors. Be careful not to run it on a PC equipped with a 80486 processor (obsolete!).

A couple of elaborate and cool batch files, DD.bat and DD2.bat, are supplied for Windows NT/2000/XP. They serve to learn how to combine two or more programs on Windows, respectively. One Fortran program and two short scripts written in VBScript are further called in DD.bat and DD2.bat. DD.bat and DD2.bat are much more convenient than another simpler batch file, RIETAN-2000.bat, for Windows 98/Me.

8.1 Archivers for Windows

A distribution file for the Windows version described in Sect. 10 has been compressed using an archiver. Prior to installing it, download one of archivers described below if necessary.

For Japanese, a representative free archiver is

• +Lhaca, deluxe version 1.20

which does not use any DLLs. Those who cannot read Japanese may use the following program:

• tbz.exe

8.2 Text editors

Text editors are indispensable to editing input files (*.ins). They also serve to browse/print outputs from RIETAN-2000. A number of powerful editors are available on the Web. Here, I introduce excellent free software for Mac OS X and Windows.

(a) Mac OS X

• TextEdit (multilingual, included in Mac OS X)
• mi (Japanese & English)
• iText (Japanese)
• HUNDOSHI-EDIT (Japanese)

Note that the standard text mode should be specified on the use of TextEdit.

(b) Windows

• iText (Japanese)
• K2Editor (Japanese)
• MKEditor for Windows (Japanese)
• NoEditor (Japanese)
• TeraPad (Japanese)
• xyzzy (Japanese)
• NTEmacs (multilingual)

Both character codes (Shift-JIS, JIS, and EUC) and ends of line (CR, LF, and CR+LF) can simultaneously be converted with an editor.

(c) Linux

• emacs
• Nano

Changes in RIETAN-2000 since January in 2004 are listed in what follows:

• Rev. 2.5 (2006/03/15): Fixed a bug giving incorrect dispersion corrections for characteristic radiation other than Cu Kα and Cu Kβ.
• Rev. 2.4.1 (2005/12/28): Corrected two bugs with Tink where lines containing only end-of-line characters give illegal ranges for character data. Corrected a bug with calculation of partial derivatives with respect to peak-shift parameters without checking their ID's.
• Rev. 2.4 (2005/12/09): Changed 'displacement parameter' in parameter lists of *.lst into 'atomic displacement parameter'. Removed parts of trailing blanks in *.lst. Added a 64-bit executable of the Mac OS X version.
• Rev. 2.3.9 (2005/10/05): Fixed a bug with calculation of the partial derivative of the March-Dollase function with respect to a preferred-orientation parameter, r, when analyzing X-ray diffraction data of compounds belonging to noncentrosymmetric space groups. Added error messages for incorrect end-of-line characters in *.ins and *.int. Changed a cryptic error message 'Too large NREGION' into 'Too large NREGION. Is the first line in *.int 'GENERAL$'?'. Modified the source code in such a way that a 64-bit executable (not distributed at present) can be generated with Absoft Pro Fortran for Mac OS X v9.2.
• Rev. 2.3.8 (2005/09/21): Corrected a typographical error in the standard output, i.e., from 'diatance' to 'distance'.
• Rev. 2.3.7 (2005/08/10): Supported automatic updates in lattice and structure parameters in *.vcs with the VICS format after Rietveld analysis in the Mac OS X version, where CrystalMaker text files, *.cmt, are no longer updated. Changed '2-th' in the first line in *.ffo into '2-theta'.
• Rev. 2.3.6 (2005/07/20): Changed the first line in *.ffo: from |F|^2, Area, and I/I0 to |Fo|^2, I, and I/I1, respectively. From now on, the Mac OS X version (Terminal application) will be built for G4 and G5 processors with Absoft Pro Fortran for Mac OS X v9.2. Corrected a careless mistake in *E.ins.
• Rev. 2.3.5 (2005/06/04): Fixed a bug that causes an error when NTARG is set at 7 (Cr Kα radiation). Modified lines for inputting R12 in all the template files, *.ins.
• Rev. 2.3.4 (2005/05/28): Corrected a misspelled word in *.lst from 'exclusing' to 'excluding'. Updated tee.exe and less.exe for the Windows version; their latest versions were obtained from GnuWin32 together with three DLLs (libiconv2.dll, libintl3.dll, and pcre.dll).
• Rev. 2.3.3 (2005/05/25): Fixed a problem with simulation of X-ray powder diffraction patterns of mixtures. Modified lines related to correction for microabsorption in *.lst; mean linear attenuation coefficients of a mixture are output during cycles of optimizing mass fractions. Updated the part relevant to correction for microabsorption in a template file, Cu3Fe4P6E.ins.
• Rev. 2.3.2 (2005/05/22): Added the mass attenuation coefficients of all the elements for Ag, Fe, and Cr Kα radiation. Fixed a bug where no effective radii of phases are initially set at zero. Updated the part relevant to correction for microabsorption in a template file, Cu3Fe4P6E.ins.
• Rev. 2.3.1 (2005/05/20): Added the mass attenuation coefficients of all the elements for Mo Kα and Cu Kβ ('Kb') radiation. Updated the part relevant to correction for microabsorption in a template file, Cu3Fe4P6E.ins.
• Rev. 2.3 (2005/05/18): Fixed a bug causing failure in outputting crystal data for each phase in *.lst when dealing with single-phase samples or neutron diffraction data in Rietveld refinements. Increased the length of the absolute path for a directory where the executable binary file of RIETAN-2000 is located from 50 to 256.
• Rev. 2.2.6 (2005/05/14): Fixed a bug causing the absence of a phase name after "Symmetry conditions in" when setting NPRINT at 2. Updated all the template files, *.ins, to avoid trouble on the analysis of neutron powder diffraction data.
• Rev. 2.2.5 (2005/05/13): Changed input data for the kind of characteristic radiation from two characters ('Ag', 'Mo', 'Kb', 'Cu', 'Co', 'Fe', and 'Cr') to an integer (1-7). Updated all the template files, *.ins.
• Rev. 2.2.4 (2005/05/11): Fixed a bug with the input of anomalous dispersion corrections for Cu Kβ radiation; they are now input by the user for all the elements contained in the sample. Updated all the template files, *.ins. Increased columns for dispersion corrections in *.lst.
• Rev. 2.2.3 (2005/05/07): Updated the wavelengths of characteristic radiation (Ag, Mo, Kb, Cu, Co, Fe, and Cr) described in "International Tables for Crystallography," Vol. C, 3rd ed., ed. by E. Prince, Kluwer, Dordrecht (2004), pp. 206-208, where α₁ should be α₂, and vice versa. Updated atomic weights compiled in asfdc in conformity to "Standard Atomic Weights 2004". Changed extension 'pat' into 'itx' in the Mac OS version, considering that most users of this version must prefer Igor Pro to other plotting programs; read "Igor Pro Version 5.0 User's Guide," Part 1, WaveMetrics, Inc., Lake Oswego (2004), II-187.
• Rev. 2.2.2 (2005/05/06): Added mass attenuation coefficients (mu/d) of phases to the list of mass fractions resulting from quantitative analysis correcting for microabsorption.
• Rev. 2.2.1 (2005/05/05): Improved the algorithm of determining mass fractions in a mixture by Brindley's method in such a way that calculations of mass fractions from current mean linear attenuation coeffcients, mu(mean), are repeated until the relative change in mu(mean), RDIF, decreases below 0.0001. The maximum number of iterations is 20. Increased NAP (maximum number of atoms in the asymmetric unit) from 150 to 300.
• Rev. 2.2 (2005/05/03): Added correction for microabsorption (absorption contrast) in the quantitative X-ray analysis of mixtures using Brindley's procedure. For details in this correction method, read G. W. Brindley, Philos. Mag., 36 (1945) 347-369; "International Tables for X-Ray Crystallography," Vol. III, Kynoch Press, Birmingham (1962), pp. 194-200; H. P. Klug and L. E. Alexander, "X-Ray Diffraction Procedures for Polycrystalline and Amorphous Materials," 2nd ed., Wiley, New York (1974), pp. 541-543; J. C. Taylor and C. E. Matulis, J. Appl. Crystallogr., 24 (1991) 14-17. The program incorporates the mass attenuation coeffcients of all the elements for Cu and Co Kα radiation; they are compiled in "International Tables for Crystallography," Vol. C, 3rd ed., ed. by E. Prince, Kluwer, Dordrecht (2004), pp. 230-236. Increased NT (maximum number of parameters contained in the model function) from 999 to 2000 and NPH (maximum number of phases) from 8 to 16. Assigned four columns to parameter numbers output after refinement in *.lst.
• Rev. 2.1.1 (2005/04/14): Fixed a bug appending extra and meaningless characters to updated lines in *.ins when NUPDT was set at 1.
• Rev. 2.1 (2005/04/08): Increased the maximum line length in *.ins to 150. Note that only comments may be described in columns 81-150 and that '#', '!', and '{' (the last character in a line) in comment lines must be located in columns 1-80. Modified *.ins in such a way that they contain no 'hankaku-katakana' characters. Fixed a minor bug where the first argument of a UNIX-compatible subroutine GETARG was not explicitly declared INTEGER*2 in the Windows version (partly remained unchanged in Rev. 1.8.3). Rewrote three shell scripts (x, ffe, cif) and Readme_Darwin for the Mac OS X version to use bash as a login shell.
• Rev. 2.0.2 (2005/03/15): Fixed a bug causing failure to update refinable structure parameters in *.vcs with the VICS format after Rietveld analysis in the Windows version. Removed the last vacant lines in spgra and spgri as in VICS v5.5.2.
• Rev. 2.0.1 (2005/02/16): Changed DENSTY in *.ins into ABSORP, which may be positive (sample density), zero (neglect absorption), and negative (-μR) in neutron powder diffraction. Removed unnecessary lines for an alternative setting of space group P3₁21 (No. 152) in a database file spgri.
• Rev. 2.0 (2004/11/16): Corrected some lines in a database file, spgra, storing information about 230 space groups. Improved the feature of Le Bail analysis considerably. First, integrated intensities are updated only after refining variable parameters in each cycle as is usual in Le Bail analysis. Second, arrays A (parameters contained in the model function) and YINC (integrated intensities) giving the minimum objective function are restored if the objective function increases after a refinement cycle. Third, the manner of checking convergence in Le Bail analysis is now different from that in Rietveld analysis. These improvements prevent the divergence of the solution, enable Le Bail analysis by the conjugate direction method, and speed up Le Bail analysis. Corrected minor errors in comment lines in user input files, *E.ins, and batch files, *.bat (Windows). From now on, the Mac OS X version will be built for G4 processors with Absoft Pro Fortran for Mac OS X v9.0.
• Rev. 1.8.5 (2004/07/01): Rewrote SUBROUTINE SF and SUBROUTINE ANALD so as to remove all the GO TO statements. Optimized the Darwin version for G4 processors.
• Rev. 1.8.4 (2004/05/13): Fixed a bug with the extension of a file for MEM analysis using PRIMA in the Mac OS version: 'mem' --> 'fos'. Rewrote SUBROUTINE DERPHI with SELECT CASE statements.
• Rev. 1.8.3 (2004/05/11): Deleted unused arguments and type declaration statements relevant to them. Fixed a minor bug where the first argument of a UNIX-compatible subroutine GETARG was not explicitly declared INTEGER*2 in the Windows version (no actual harm). From now on, the Windows version will be built with Visual Fortran for Windows v6.6C3.
• Rev. 1.8.2 (2004/04/23): Fixed a problem with 'beta23' denoting anisotropic atomic displacement parameters in linear constraints.
• Rev. 1.8.1 (2004/04/21): Following Rev. 1.8, rewrote several parts of source code in line with the grammar of Fortran 90. Fixed a bug where nonlinear least-squares refinement by the Gauss-Newton method may not proceed properly when part of refinable lattice parameters diverge.
• Rev. 1.8 (2004/04/15): Rewrote many lines in conformity with the grammar of Fortran 90. There are no feature enhancements or incompatibilities with Rev. 1.7.X.
• Rev. 1.7.1 (2004/04/10): Fixed a bug with outputting two identical lines containing only refinement identifiers, ID(I), when updating *.ins including such a line. This problem, which occurs only in the Mac OS version because of a bug of Absoft Pro Fortran, was solved by changing 'DO WHILE (.TRUE.)' into 'DO'.
• Rev. 1.7 (2004/04/09): More detailed information about profile parameters is output when initial profile parameters give invalid primary profile parameters.
• Rev. 1.6.1 (2004/04/07): Fixed a problem with reading in background intensities from *.bkg (NRANGE = 2) and excluding part of diffraction data (NEXC = 1). Increased the number of digits for offsets in Igor text files, *.pat, from 6 to 7. Changed the format for 2θ(min) and 2θ(max) from F8.3 to F9.4.
• Rev. 1.6 (2004/04/03): Rewritten part of source programs in conformity with Fortran 90/95; an allocatable array was used for the first time. Increased NB and NGEN (total number of reflections) from 7000 to 20000. Changed the format of NFUNCT (NLESQ = 2) from I4 to I6. At the tail of the standard output, not 'Copyleft' but 'Copyright' is written; further, a CPU time is output in place of an elapsed time.
• Rev. 1.5.3 (2004/03/12): Fixed a problem with outputting estimated standard deviations (zero) for isotropic atomic displacement parameters, U, even if B's are fixed.
• Rev. 1.5.2 (2004/02/20): Optimized rietan.exe for Intel Pentium III or later.
• Rev. 1.5.1 (2004/02/19): Changed 'Bacgground' into 'Background' in the standard output.
• Rev. 1.5 (2004/01/15): A binary file, *.fos, is input for MEM analysis instead of a text file, *.mem. A Lagrangian multiplier is no longer input in *.ins.
• Rev. 1.4 (2004/01/09): A coefficient, SCIO, to adjust the estimated standard deviations of |F_o|'s is no longer input in RIETAN-2000, which led to changes in data recorded in *.mem.
• Rev. 1.3 (2004/01/06): Made it possible to read in binary files for MPF analysis, *.fba, compatible with PRIMA v3.1. Note that text files, *.fba, output by PRIMA v3.01 or earlier can no longer input by RIETAN-2000 Rev. 1.3. Added observed integrated intensities, I_o, to the tails of lines in *.ffo resulting from Le Bail analysis; they are used to estimate standard deviations of |F_o|'s on the basis of counting statistics. Modified a line in *.mem for MEM analysis. Fixed a bug related to identifiers, ID, for structure parameters in MPF analysis (NMODE = 2 and 3). As of January 6 in 2004, no source code of RIETAN-2000 will be distributed.

Download one of three binary archives for three different operating systems to mount or uncompress/extract it. Note that an older version of RIETAN-2000 has to be deleted prior to the installation of the latest version.

The archive files will sometimes be updated; please regard revision numbers or dates of updates. Changes in the latest versions are described in Sect. 9.

Only executable programs are included in the distribution files; the source code of RIETAN-2000 is given only to collaborators.

10. 1 RIETAN-2000 for Mac OS X

A disk image file containing Terminal applications for Darwin (UNIX) is distributed here. To grab it, click (ctrl + click in the case of Mac OS X) the mouse button on the following archive file name

• rietan2000m.dmg (Rev. 2.5; 2006/03/15; 2.2 MB)

to display a pop-up menu and choose "Download Link to Disk", "Save this Link As...", "Save Target As...", or a corresponding item (depending on your browser). Double-click the downloaded disk image file, rietan2000m.dmg, on Mac OS X to mount a virtual disk 'RIETAN-2000 Folder'.

Read a text file, RIETAN-2000 Folder:Readme_mac.txt, thoroughly to obtain further information.

10. 2 RIETAN-2000 for Windows

Position the pointer over the the following archive file name

• rietan2000w.tbz (Rev. 2.5; 2006/03/15; 4.9 MB)

Right-click the link to display a pop-up menu and choose "Save Link As...", "Save Target As...", or a corresponding item (depending on your browser). Drag and drop the downloaded file, rietan2000w.tbz, onto the icon of an archiver (refer to Archivers for Windows) to get a folder called RIETAN-2000 Folder. For details in the uncompressing/extraction process, refer to instructions of the archiver carefully.

Read Readme_win.txt in RIETAN-2000 Folder thoroughly to acquire further information. Gnuplot v4.0 for Windows is included in the above archive file to save time for those who want to plot results of Rietveld refinements with it.

10.3 RIETAN-2000 for Linux

With my permission, Hiroshi Ohki of Shinshu University has been distributing

• A Linux version of RIETAN-2000

An executable file built for Intel CPUs is included in this version.

You can use gnuplot script files, *.plt, included in the distribution files for Windows with 'set terminal x11'.

RIETAN-2000 was coded so that they can be built with the following Fortran compilers:

1) Mac OS X: Absoft Pro Fortran for Mac OS X v9.2
2) Windows: Visual Fortran for Windows v6.6C3

Porting RIETAN-2000 to various platforms are easy thanks to the separation of OS-dependent parts.

The archive files involve three kinds of template files, *.ins, written in both English and Japanese for neutron, characteristic X-ray, and SR powder diffraction. They also include another template file, Cu3Fe4P6E.ins, written in English for multi-phase Rietveld analysis of X-ray powder diffraction data. Here, please browse a couple of template files for Rietveld analysis of X-ray powder diffraction data for fluorapatite:

• FapatiteE.ins (English)
• FapatiteJ.ins (Japanese)

You will feel overwhelmed with their volumes resulting from the many new capabilities implemented in RIETAN-2000. You will also understand that *.ins is regarded as a kind of a reference manual where you can add/modify/delete comments yourself, which is impossible with the GUI using menus and dialog boxes.

13.1 Graphing programs

RIETAN-2000 has an option to output Igor Pro text files to plot results of analysis and simulation on Mac OS and Windows. The advantage of Igor Pro over a self-made plotting routine is its high capabilities including (1) a programming language, (2) exporting files with several different formats, and (3) page layouts. You may complaint of not possessing an enough budget to buy it. However, WaveMetrics extends academic pricing to students and teachers. In the case of on-line ordering, WaveMetrics further permits students with limited financial resources to purchase Igor Pro at an unbelievable price of $85!

Igor Pro for Linux is not sold. However, the combination of Linux and VMware makes it possible to run Igor Pro for Windows on computers compatible with IBM-PC.

If free software is preferred to commercial software, please select

• gnuplot (any platforms)

Features (1) and (2) of Igor Pro described above are also supported in gnuplot, and feature (3) can practically be achieved by importing EPS files created by gnuplot in graphic software such as Adobe Illustrator.

13.2 Third party utilities for RIETAN-2000

As far as I know, seven programs

• MYPLOT (Hiroyuki Miura, Hokkaido University; Windows)
• PatMill (Zentaro Akase, Tohoku University; Windows and Linux)
• PlotRietRefPat (Akitaka Hoshino, Kyoto University; Macintosh; pro Fit is required)
• PowderPlot (Koichi Momma, Tohoku University; Java Applet)
• RietPlot 2000 (Naoki Ohashi, AML/NIMS; Windows)
• RIETVIEW (Dr. Hell / Dokuta Jigokusai; Windows)
• RietPlot (Yoshihisa Mori, Okayama University of Science; Windows)
• WinPLOTR (T. Roisnel and J. Rodriguez-Carvajal, LLB; Windows)

to plot results of Rietveld refinements with RIETAN-2000 are distributed on the Web. PatMill allows us to input *.pat for Igor Pro. PatMill for Windows requires ActivePerl and Drop on Script in addition to gnuplot included in the RIETAN-2000 package. PatMill also runs just fine under Linux.

Further,

• PIETAN (Mikiko Kato, JFE Techno-Research; Windows)
• RieMenu (Migaku Kobayashi, NCT; Windows)
• DD_NT.pl and DD_9x.pl (Zentaro Akase, Tohoku University; Windows)
• CRietan2000 (Mineo Sato, Niigata University; Windows)
• SeqRun.xls (Masami Tsubota; Windows)
• Files to support RIETAN-2000 (Norihiro Muroyama; Windows)

offer user interface environments for the Windows version of RIETAN. I regret that no program for the Macintosh has yet been developed.

I am really honored to have these utilities developed for RIETAN-2000. Please let me know of any other programs having something to do with RIETAN-2000 if any.

13.3 VENUS system

VENUS is an advanced 3D visualization system fully utilizing the OpenGL API and MEM. The VENUS package consists of the following four programs:

• VICS: VIsualization of Crystal Structures
• VEND: Visualization of Electron/Nuclear Densities
• PRIMA: PRactice Iterative MEM Analyses
• ALBA: After Le Bail Analyses

We designed it, bearing a combination with RIETAN-2000 in mind. For details in VENUS, please visit another Web page 3D Visualization System VENUS, where archive files for VENUS are distributed. It should be emphasized that the VENUS system includes a super-fast MEM analysis program, PRIMA, and ALBA for the maximum-entropy Patterson method.

VICS has a feature to input/output standard input files, *.ins, of RIETAN-2000 while VEND can input 3D data files, *.pri, created by PRIMA and ALBA. PRIMA was designed with MPF in mind, as can be inferred from its name. ALBA allows us to improve integrated intensities for overlapped reflections in a sophisticated way. The VENUS system, therefore, adds value to RIETAN-2000.

13.4 Third party structure-drawing programs

Visualization and manipulation of crystal structures using GUI is indispensable for the analysis of crystal structures by powder diffraction. The feature of creating CIFs is particularly useful for this purpose. Nearly all structure-drawing applications, e.g., ATOMS, CrystalMaker, Diamond, Crystallographica, and Balls & Sticks can import CIFs.

13.5 Versatile data processing system PowderX

Powder diffraction analysis software by Cheng Dong

• PowderX

can input intensity data file (general format), *.int, for RIETAN-2000, which is followed by data processing such as background subtraction, smoothing, α₂ stripping, and indexing of reflections with "inbuilt" TREOR.

I am sorry that no formal manual for RIETAN-2000 has been presented yet. I prefer software development to preparing its documentation. In effect, users of previous versions of RIETAN will be able to manage it without serious problems owing to the presence of the template files including a number of comments.

In addition, the following PDF (Portable Document Format) documents, (a)-(g), and other copyrighted articles are available here, in the library, and elsewhere:

(a) Notes on RIETAN-2000

• Notes.pdf (2005/08/26)

A supplementary guide of RIETAN-2000 for users of RIETAN-94/97beta.

(b) Rietveld-refinement program RIETAN-94 for angle-dispersive X-ray and neutron powder diffraction

• RIETAN-94_manual.pdf

This user's manual for an earlier version of RIETAN is obsolete but serves to learn features that had already been supported in RIETAN-94.

(c) Parameters contained in the model function of RIETAN-2000

• Parameters.pdf (2003/08/30)

This document listing all of the parameters to calculate the diffraction intensity (model function) is indispensable for preparation of *.ins.

(d) Rietveld analysis and MPF under partial profile relaxation

• F. Izumi, Rigaku J., 17, No. 1 (2000) 34-45. (2002/08/10)

This review article deals with two cutting-edge features implemented in RIETAN-2000: partial profile relaxation and MPF. It is also available in a printed form from here.

(e) User guide for the multi-purpose pattern-fitting system RIETAN-2000

• F. Izumi, Rigaku-Denki J., 31, No. 2 (2000) 17-29. (in Japanese; 2005/02/18)

This review describes important information regarding the model function to calculate diffraction intensities, Le Bail refinement, and New Tink (preprocessor of input files). Accordingly, it must particularly benefit Japanese users. Various functions contained in the model function are explained in an understandable manner. Those who are going to challenge Le Bail refinement are strongly recommended to read this since its know-how is described here in detail.

Several mistakes in the published article have been corrected in this PDF file.

(f) Talking about the integrated pattern-fitting system RIETAN-2000

• F. Izumi, Hamon, 10, No. 2 (2000) 21-31. (in Japanese)

What can I say ..... I went into rhapsodies over the development of RIETAN-2000 in an aberrant manner like Don Quixote de la Mancha. Finally, my great endeavors bore fruit; RIETAN-2000 was completed in the spring of 2000. This article is heavy with frivolous rhetoric, but the story is believed to be amusing to most people.

Reading this article, you will understand fully the limitation of the MEM/Rietveld method and the necessity of the MPF method.

(g) Beyond the abilities of Rietveld analysis: MEM-based pattern fitting with synchrotron X-ray powder diffraction data

• F. Izumi and T. Ikeda, Commission on Powder Diffraction, IUCr Newslett., No. 26 (2001) 7-9.

This article reports (1) biting criticism against the MEM/Rietveld method, (2) the principle of MPF, and (3) three examples of applying MPF to synchrotron X-ray powder diffraction data measured at SPring-8.

(h) Structure refinement by Rietveld analysis and MEM-based whole-pattern fitting under partial profile relaxation

• F. Izumi, S. Kumazawa, T. Ikeda, and T. Ida, "Powder Diffraction," Proc. Int. Workshop Powder Diffr. (ISPD '98), ed. by S. P. Sen Gupta, Allied Publ., New Delhi (1998), pp. 24-36. (ISBN 81-7023-881-1)

ISPD '98 was held at the Indian Association for the Cultivation of Science (IACS) in Calcutta during October 7-10, 1998. IACS is the premier research institute that has produced many illustrious scientists including C. V. Raman who earned a Nobel prize in 1930 for the well-known discovery of the Raman effect. ISPD was sponsored by the Commission on Powder Diffraction of the International Union of Crystallography and co-sponsored by the Indian National Science Academy.

(i) A book on X-ray powder diffraction and Rietveld analysis

The following book (ISBN 4-254-14059-2 C3043, Price: 4 800 yens, in Japanese, B5 size, 196 pages) was published on February 10, 2002:

• A Practical Guide to X-Ray Powder Analysis --- Introduction to the Rietveld Method," ed. by I. Nakai and F. Izumi, Asakura, Tokyo (2002). (ISBN 4-254-14059-2 C3043)

We wrote this book, bearing RIETAN-2000 in mind. It must supplement the RIETAN-2000 package with a detailed description of its content: fundamental knowledge of crystallography, all the important functions contained in the model function, R factors, method of editing *.ins, etc. Click here to visit a web page (in Japanese) to offer a service for its readers.

(j) Structure analysis introducing partial profile relaxation and the maximum-entropy method with powder diffraction data

• F. Izumi, T. Ikeda, and S. Kumazawa, J. Mineral. Soc. Jpn., 28 (1999) 57-63. (in Japanese)

This article is similar to Ref. (d) except that it is written in Japanese.

(k) Novel techniques of neutron powder diffraction and their applications to superconducting oxides

• F. Izumi, A. Yamamoto, N. R. Khasanova, S. Kumazawa, W.-Z. Hu, and T. Kamiyama, Proc. 9th Jpn-US Workshop High-T_c Supercond., Physica C (Amsterdam), 335 (2000) 239-244.

The last part of this paper describes determination of nuclear densities in HgBa₂CuO_4+z by the MPF method.

(l) Structure refinement from powder diffraction data

• F. Izumi and N. Ohashi, Ceramics Jpn. (Bull. Ceram. Soc. Jpn.), 36 (2000) 1055-1063. (in Japanese)

We described how to use RIETAN-2000 on Windows and Mac OS as part of "Better Usage of Personal Computers for Ceramists". A revised version of this article was included in a CD-ROM book (ISBN 4-931298-35-4) with the same title.

(m) Multi-purpose pattern-fitting system RIETAN-2000 and its applications to microporous materials

• F. Izumi and T. Ikeda, J. Crystallogr. Soc. Jpn., 42 (2000) 516-521. (in Japanese)

This review article describes (1) main new features of RIETAN-2000, (2) changes in R factors during the MPF of Na-LTA, (3) structure refinements of Na-SOD and Li-FAU by MPF, and (4) ab initio structure analysis of HLS. Reading this, you will recognize RIETAN-2000 to have an outstanding performance as multipurpose pattern-fitting software.

(n) Analysis of unknown and disordered structures by utilizing RIETAN-2000

• F. Izumi and T. Ikeda, J. Crystallogr. Soc. Jpn., 44 (2002) 30-34. (in Japanese)

Synchrotron X-ray powder diffraction and latest structure-analysis software has increased possibilities of solving unknown structures. This article describes (1) estimation of integrated intensities by step-by-step Le Bail refinements, (2) derivation of an initial structural model by the direct method, and (3) analysis of a disordered structure by the MPF method. Examples of analyzing synchrotron powder data are demonstrated for cimetidine in (1), zeolite LTL in (2), and Sr_9.3Ni_1.2(PO₄)₇ in (3).

(o) A guide to full utilization of RIETAN (1) Input and output files

• F. Izumi, J. Crystallogr. Soc. Jpn., 44 (2002) 246-254. (in Japanese)

RIETAN-2000 is connected with several related programs via a variety of text files for cooperative crystallographic computation and visualization. Contents of input and output files of RIETAN are explained for its full utilization. The knowledge of measuring reliable powder diffraction data recorded in *.int and parts of data in the user-input file, *.ins, are described particularly in detail.

(p) A guide to full utilization of RIETAN (2) Information obtained from powder diffraction data

• F. Izumi, J. Crystallogr. Soc. Jpn., 44 (2002) 311-317. (in Japanese)

The most important purpose of Rietveld analysis is the refinement of lattice and structure parameters. In addition, we can obtain geometric data (in turn, bond valence sums), magnetic moments (neutron diffraction), crystallite sizes, microstrains, and mass fractions. They are briefly described so that one can gain a better understanding of the information extractable from powder diffraction data.

(q) A guide to full utilization of RIETAN (3) Three-dimensional visualization of electron- and nuclear-density distribution

• F. Izumi, J. Crystallogr. Soc. Jpn., 44 (2002) 380-388. (in Japanese with four color pages)

The combination of RIETAN-2000 and a super-fast MEM analysis program PRIMA enables us to represent crystal structures with densities of electrons (X-ray diffraction) or coherent-scattering lengths (neutron diffraction) instead of structure parameters in Rietveld analysis. We have devised sophisticated methodology named MEM-based pattern fitting, where whole-pattern fitting and MEM analysis are alternately repeated to minimize the bias of a structural model in Rietveld analysis. Applications of MPF to some compounds will be demonstrated. A 3D visualization program, VENUS, serves to display and manipulate electron and scattering-length densities, electrostatic potentials, wave functions, and so forth.

(r) Fundamentals in crystallography and the powder method

• F. Izumi, "Handbook of Ceramic Technology," 2nd ed., ed. by Ceram. Soc. Jpn., Gihodo, Tokyo (2002), Fundamentals, pp. 531-544. (ISBN 4-7655-0032-2; in Japanese)

Introduction to X-ray crystallography and X-ray powder diffraction for biginners.

(s) Structure refinement based on the maximum-entropy method from powder diffraction data

• F. Izumi and R. A. Dilanian, "Recent Research Developments in Physics," Vol. 3, Part II, Transworld Research Network, Trivandrum (2002), pp. 699-726. (ISBN 81-7895-046-4)

The MEM, MPF method, and applications of the MPF technology to several inorganic compounds are described in detail.

(t) MEM-based structure-refinement system REMEDY and its applications

• F. Izumi, S. Kumazawa, T. Ikeda, W.-Z. Hu, A. Yamamoto, and K. Oikawa, Mater. Sci. Forum, 378-381 (2001) 59-64.

The MPF method is outlined together with its applications to structural studies of three compounds: (1) positional disorder of K⁺ ions interlayered in K_xTi_2-x/3Li_x/3O₄, (2) atomic arrangements of K clusters in a K-type zeolite LTA, and (3) nuclear/electron-density distribution in a high-T_c superconductor HgBa₂CuO_4+z.

(u) Development and applications of the pioneering technology of structure refinement from powder diffraction data

• F. Izumi, J. Ceram. Soc. Jpn., 111 (2003) 617-623. (with one color page. Every member of the CerSJ can download its PDF file.)

I was asked to write a special article in honor of getting the 57th CerSJ Award for Academic Achievements mainly for the development of RIETAN-2000. The history of developing RIETAN for angle-dispersive and TOF neutron powder diffraction is briefly retraced. Then, the two advanced techniques of partial profile relaxation and MPF are introduced with five applications.

(v) Positional and orientational disorder in a solid solution of Sr_9+xNi_1.5-x(PO₄)₇

• A. A. Belik, F. Izumi, T. Ikeda, V. A. Morozov, R. A. Dilanian, S. Torii, E. M. Kopnin, O. I. Lebedev, G. Van Tendeloo, and B. I. Lazoryak, Chem. Mater., 14 (2002) 4464-4472.

This original paper deals with application of MPF to synchrotron X-ray powder diffraction data of the highly disordered material. An outline of the MPF method is described in Introduction.

(w) Beyond the ability of Rietveld analysis: whole-pattern fitting based on the maximum-entropy method

• F. Izumi, "Applied Crystallography," Proc. XIX Conf., ed. by H. Morowiec and D. Stroz, World Scientific, Singapore (2004), pp. 65-70. (ISBN 981-238-761-7)

I gave a keynote lecture at the XIXth Conference on Applied Crystallography. This review article briefly describes its contents: the methodology of MPF and its applications to three inorganic materials.

(x) Beyond the ability of Rietveld analysis: MEM-based pattern fitting

• F. Izumi, Solid State Ionics, 172 (2004) 1-6.

I gave a keynote lecture at the 15th International Symposium on the Reactivity of Solids. This review article describes its contents, i.e., MPF and its applications to five inorganic compounds with highly disordered structures.

(y) Powder diffraction

• F. Izumi, "Spectroscopy and Diffraction III," the Fifth Series of Experimental Chemistry, Vol. 11, ed. by the Chemical Society of Japan, Maruzen, Tokyo (2006), Chap. 4 (ISBN 4-621-07310-9; in Japanese).

Chapter 4 describes the various aspects of powder diffraction including X-ray powder diffractometry, Rietveld analysis, pattern decomposition (Pawley and Le Bail methods), MEM, MEM-based pattern fitting, and ab initio structure analysis (reciprocal- and direct-space methods).

RIETAN-2000 incorporates a KDRREF program to generate diffraction indices (hkl) and multiplicities (m). To my regret, this routine contains bugs; it rarely generates surplus reflections, miss possible reflections, and give incorrect multiplicities.

Because contents of database files, particularly spgra, have not been checked sufficiently, equivalent positions input from them may sometimes be erroneous. They can be checked by referring to "International Tables for Crystallography," Vol. I or Vol. A and changed by editing spgri or spgra.

Provided that parts of hkl and m generated with RIETAN-2000 seem to be doubtful, check them using, e.g., Lazy Pulverix developed at the University of Geneva. This program is contained in ICSD and PowderX. The LMGP suit can also list hkl and m.

The latest version of RIETAN-2000 is capable of reading in a file, hklm.phn (n: phase number), storing hkl and m values for allowed Bragg reflections for the n'th phase. If hklm.phn is located in the same folder as *.ins and *.int, reflection indices and multiplicities for the n'th phase is not generated by RIETAN-2000 but input from hklm.phn. According to a format (in conformity with the grammar of Fortran) placed at the first line in this file, h, k, l, and m (four integers in each line) after the first line are input. For example, '(4I5)' means four right-justified integers each with five columns.

I ported Lazy Pulverix to Mac OS and Windows. This program was modified in such a way that a file called hklm.ph1 (change the extension if necessary) is always created. Developers of Lazy Pulverix, professors K. Yvon and E. Parthe, kindly permitted me to redistribute it as free software. To get this program for PC, please send me an e-mail.

In the next version of RIETAN, Lazy Pulverix or a similar program may be substituted for KDRREF.

The development of RIETAN has been led by user feedback. Its phenomenal success results from the response from around the country. I assume that things are working fine provided that no feedback is received. If its users were so selfish as to behave like "black holes," RIETAN would not be improved very much. Please feel free to inform me of any bugs and your thoughts. Sending me *.ins and *.int would facilitate the debug.

Encouraging responses would present me with driving forces for further development and refinement of RIETAN. Please let me have a postcard of your home town or a city you traveled if you like/love RIETAN-2000 and continue to use it. Is that too much to ask?

17.1 Analysis of neutron diffraction data with RIETAN-2000

Important information on Rietveld analysis of neutron powder diffraction data with RIETAN-2000 is presented at here (Web page of HERMES). At least on Japanese WWW sites, I have never browsed such a detailed explanation of analysis of data measured at synchrotron-radiation and neutron-scattering facilities. This Web page must be of great benefit to those who have measured neutron intensity data on HERMES; Kenji Ohoyama, who is an instrumental scientist responsible for the HERMES diffractometer, has taken great effort to prepare the Web page.

17.2 A Web site for structure analysis using RIETAN-2000 (+VENUS)

Tatsuya Furuya built a Web site on structure analysis using RIETAN-2000 (+VENUS) including Q & A concerning powder diffraction and Rietveld analysis. This site is particularly useful for beginners of Rietveld analysis, offering detailed information about RIETAN-2000 for Windows and Mac OS.

17.3 A Web site offering information about RIETAN-2000

Masami Tsubota provides us with information on RIETAN-2000 (under construction) in his Web site.

18.1 Intensity data measured with unequal steps

On the analysis of intensity data collected with unequal steps, e.g., neutron diffraction data taken on HERMES and HRPD at JRR-3M, the top line of *.int with a general format should be not 'GENERAL' but 'GENERAL$'. The last character, '$', is required to calculate observed integrated intensities in order when step widths with respect to 2θ vary two or more times for one reflection. This tip is generally effective when an error message "Too large NREGION" appears during calculation of R factors.

18.2 Refinement of anomalous-scattering factors

A feature to refine anomalous-scattering factors for each crystallographic site was implemented in an in-house version of RIETAN-2000 by Yanan Xiao (Southern Illinois Univ. and Univ. Chicago):

• Y. Xiao, F. Izumi, T. Graber, P. J. Viccaro, and D. E. Wittmer, Powder Diffr., 18 (2003) 32-35.
• Y. Xiao, D. E. Wittmer, F. Izumi, S. Mini, T. Graber, and P. J. Ciccaro, Appl. Phys. Lett., 85 (2004) 736-738.

19. Acknowledgments

I cordially thank Takuji Ikeda for his dedicated cooperation throughout the development of RIETAN-2000 and the extended version of ORFFE. Without his help and encouragement, RIETAN-2000 would not have been completed in the current form full of professionalism. He also contributed his time and energy to testing alpha/beta versions and making up the template files. The feature of MPF owes its existence to Shintaro Kumazawa's proposal to implement it in RIETAN. Thanks are also due to Ruben A. Dilanian for developing lst2cif, VEND, VICS, PRIMA, and ALBA at my request. Many other users helped me in improving RIETAN-2000 by reporting its bugs and problems. I am much indebted to Power Macintosh 9500/132, who died just after the release of RIETAN-2000, for working diligently during computer programing using Language Systems Fortran. The development of both RIETAN-2000 and VENUS was financially supported by the Multi-Core Project for Superconductivity Research.