1. PLATON and STRUCTURE VALIDATION Ton Spek National Single Crystal Service Facility, Utrecht University, The Netherlands. Goettingen, 13-Oct-2007

2. Overview of the Talk 1.What is PLATON 2.Structure Validation 3.Concluding Remarks Copy: http://cryst.chem.uu.nl

3. What is PLATON PLATON is a collection of tools for single crystal structure analysis bundled within a single SHELX compatible program. The tools are either extended versions of existing tools or unique to the program. The program was/is developed over of period of more than 25 years in the context of our National Single Crystal Service Facility in the Netherlands.

4. PLATON USAGE Today, PLATON is most widely used implicitly in its validation incarnation for all single crystal structures that are validated with the IUCr CHECKCIF utility. Tools are available in PLATON to analyze and solve the reported issues that need attention.

5. OTHER PLATON USAGE PLATON also offers guided/automatic structure determination and refinement tools for routine structure analyses from scratch (i.e. the ‘Unix-only’ SYSTEM S tool and the new FLIPPER/STRUCTURE tool that is based on the Charge Flipping Ab initio phasing method). Next Slide: Main Function Menu 

7. Selected Tools ADDSYM – Detection and Handling of Missed (Pseudo)Symmetry TwinRotMat – Detection of Twinning SOLV – Report of Solvent Accessible Voids SQUEEZE – Handling of Disordered Solvents in Least Squares Refinement (Easy to use Alternative for Clever Disorder Modelling) BijvoetPair – Post-refinement Absolute Structure Determination (Alternative for Flack x) VALIDATION – PART of IUCr CHECKCIF

8. ADDSYM Often, a structure solves only in a space group with lower symmetry than the correct space group. The structure should subsequently be checked for higher symmetry. About 1% of the 2006 & 2007 entries in the CSD need a change of space group. E.g. A structure solves only in P1. ADDSYM is a tool to come up with the proper space group and to carry out the transformation (  new.res) Next slide: Recent example of missed symmetry

9. Organic Letters (2006) 8, 3175 P1, Z’ = 8 CCoCCo Correct Symmetry ?

10. Test for Higher Symmetry Start PLATON with a.ins or.cif Click on ADDSYM on the main menu Analyse automatically generated result  Display next

12. After Transformation to P2 1 2 1 2 1, Z’ = 2

13. (Pseudo)Merohedral Twinning Options to handle twinning in L.S. refinement available in SHELXL, CRYSTALS etc. Problem: Determination of the Twin Law that is in effect. Partial solution: coset decomposition, try all possibilities (I.e. all symmetry operations of the lattice but not of the structure) ROTAX (S.Parson et al. (2002) J. Appl. Cryst., 35, 168. (Based on the analysis of poorly fitting reflections of the type F(obs) >> F(calc) ) TwinRotMat Automatic Twinning Analysis as implemented in PLATON (Based on a similar analysis but implemented differently)

14. TwinRotMat Example Originally published as disordered in P3. Solution and Refinement in the trigonal space group P-3  R= 20%. Run PLATON/TwinRotMat on CIF/FCF Result: Twin law with an the estimate of the twinning fraction and the estimated drop in R-value Example of a Merohedral Twin 

16. Ideas behind the Algorithm Reflections effected by twinning show-up in the least-squares refinement with F(obs) >> F(calc) Overlapping reflections necessarily have the same Theta value within a tolerance. Generate a list of implied possible twin axes based on the above observations. Test each proposed twin law for its effect on R.

17. Possible Twin Axis H H’ H” = H + H’ Strong reflection H’ with theta close to theta of reflection H Candidate twinning axis (Normalize !) Reflection with F(obs) >> F(calc)

19. Solvent Accessible Voids A typical crystal structure has only in the order of 65% of the available space filled. The remainder volume is in voids (cusps) in-between atoms (too small to accommodate an H-atom) Solvent accessible voids can be defined as regions in the structure that can accommodate at least a sphere with radius 1.2 Angstrom without intersecting with any of the van der Waals spheres assigned to each atom in the structure. Next Slide: Void Algorithm: Cartoon Style 

20. STEP #1 – EXCLUDE VOLUME INSIDE THE VAN DER WAALS SPHERE DEFINE SOLVENT ACCESSIBLE VOID

21. STEP # 2 – EXCLUDE AN ACCESS RADIAL VOLUME TO FIND THE LOCATION OF ATOMS WITH THEIR CENTRE AT LEAST 1.2 ANGSTROM AWAY

22. DEFINE SOLVENT ACCESSIBLE VOID STEP # 3 – EXTEND INNER VOLUME WITH POINTS WITHIN 1.2 ANGSTROM FROM ITS OUTER BOUNDS

23. Cg Listing of all voids in the triclinic unit cell

25. VOID APPLICATIONS Calculation of Kitaigorodskii Packing Index As part of the SQUEEZE routine to handle the contribution of disordered solvents in crystal structure refinement Determination of the available space in solid state reactions (Ohashi) Determination of pore volumes, pore shapes and migration paths in microporous crystals

26. SQUEEZE Takes the contribution of disordered solvents to the calculated structure factors into account by back-Fourier transformation of density found in the ‘solvent accessible volume’ outside the ordered part of the structure (iterated). Filter: Input shelxl.res & shelxl.hkl Output: ‘solvent free’ shelxl.hkl Refine with SHELXL or Crystals Note:SHELXL lacks option for fixed contribution to Structure Factor Calculation.

27. SQUEEZE Algorithm 1.Calculate difference map (FFT) 2.Use the VOID-map as a mask on the FFT-map to set all density outside the VOID’s to zero. 3.FFT -1 this masked Difference map -> contribution of the disordered solvent to the structure factors 4.Calculate an improved difference map with F(obs) phases based on F(calc) including the recovered solvent contribution and F(calc) without the solvent contribution. 5.Recycle to 2 until convergence.

28. SQUEEZE In the Complex Plane Fc(model) Fc(solvent) Fc(total) Fobs Solvent Free Fobs Black: Split Fc into a discrete and solvent contribution Red: For SHELX refinement, temporarily substract recovered solvent contribution from Fobs.

29. Comment The Void-map can also be used to count the number of electrons in the masked volume. A complete dataset is required for this feature. Ideally, the solvent contribution is taken into account as a fixed contribution in the Structure Factor calculation (CRYSTALS) otherwise it is substracted temporarily from F(obs)^2 (SHELXL) and re-instated afterwards with info saved beyond column 80 for the final Fo/Fc list.

30. Publication Note Always give the details of the use of SQUEEZE in the comment section Append the small CIF file produced by PLATON to the main CIF Use essentially complete data sets with sufficient resolution only. Make sure that there is no unresolved charge balance problem.

31. Post-Refinement Absolute Structure Determination Generally done as part of the least squares refinement with a ‘twinning’ parameter.(Flack x) Determine Flack x parameter + su Validity Analysis following the Flack & Bernardinelli criteria. Often indeterminate conclusions obtained in the case of light atom structures Alternative approaches offered by PLATON 

32. Scatter Plot of Bijvoet Differences Plot of the Observed Bijvoet Differences against the Calculated Differences. A Least-Squares line is calculated The Green least squares line should run from the lower left to the upper right corner for the correct absolute structure. Vertical bars on data points indicate the su on the Bijvoet Difference. Example 

33. Excellent Correlation

34. Practical Aspects of Flack x The structure should contain atoms with sufficiently strong anomalous dispersion contributions for the radiation used (generally  in the experiment (e.g. Br). Preferably, but not nesessarily, a full set of Friedel pairs is needed. (danger: correlation !) Unfortunately, many relevant pharmaceuticals contain in their native form only light atoms that at best have only weak anomalous scattering power and thus fail the strict Flack conditions.

35. Light Atom Targets Options for the Absolute Structure Determination of Light Atom Compounds Add HBr in case of tertiary N. Co-crystallize with e.g. CBr4. Co-crystallize with compound with known. absolute configuration. Alternative: Statistical analysis of Bijvoet pair differences.

36. Statistical Analysis of Bijvoet Pairs Many experimentalists have the feeling that the official Flack x method is too conservative. This Experience is based on multiple carefully executed experiments with compounds with known absolute structure. The feeling is that also in light atom structures the average of thousands of small Bijvoet differences will point in the direction of the correct enantiomorph. Example: The Nonius CAD4 test crystal 

37. Example: Ammonium Bitartrate Test MoKa, P212121

38. Ammonium BiTartrate (MoK 

39. Bayesian Approach Rob Hooft (Bruker) has developed an alternative approach for the analyses of Bijvoet differences that is based on Bayesian statistics. (Paper under review) Under the assumption that the material is enantiopure, the probability that the assumed absolute structure is correct, given the set of observed Bijvoet Pair Differences, is calculated. An extension of the method also offers the Fleq y (Hooft y) parameter to be compared with the Flack x. Example: Ascorbic Acid, P21, MoKa data 

40. Natural Vitamin C, L-(+)Ascorbic Acid MoKa

41. L-(+) Ascorbic Acid

42. Hooft y Proper Procedure Collect data with an essentially complete set of Bijvoet Pairs Refine in the usual way (preferably) with BASF and TWIN instructions (SHELXL) Structure Factors to be used in the analysis are recalculated in PLATON from the parameters in the CIF (No Flack x contribution).

43. Do we need Validation ? Some Statistics Validation CSD Entries 2006 + 2007 Number of entries: 35760 # of likely Space Group Changes: 384 # of structures with voids: 3354 Numerous problems with H, O, OH, H 2 O etc.

44. Structure Validation Some Examples of Recently Published Structures with a Problem that Apparently Escaped the Attention of the Referees. Promote CheckCif as the Current (Partial) IUCr Solution to this problem. Details of what PLATON can do in this Context. Next: An Example of a strange structure in the CSD 

45. Structure of an Interesting CH 3 Bridged Zr Dimer Paper has been cited 47 times ! So can we believe this structure? The Referees did …! H.. H = 1.32 Ang. ! But …

46. Comment The methyl hydrogen atoms are expected outside the Zr 2 C 2 ring (and indeed have been found in similar structures) Referees likely had no access to (or did not access) the primary data other than the ORTEP illustration in the paper. General problem: A limited number of experts is available to referee too many structural papers that offer only limited primary (deposited) data.

47. Dalton Trans. (2001), 729-735 Next Slide: ORTEP with downloaded CIF data 

48. From CSD

49. Organometallics (2006) 25, 1511-1516 Next Slide: This is why the reported density is low and the R and Rw high 

50. Solvent Accessible Void of 235 Ang3 out of 1123 Ang 3 Not Accounted for in the Refinement Model

51. SOLUTION A solution for the structure validation problem was pioneered by the International Union of Crystallography -Provide and archive crystallographic data in the computer readable CIF standard format. -Offer Automated validation, with a computer generated report for authors and referees. -Have journals enforce a structure validation protocol. - The IUCr journals and most major journals now indeed implement some form of validation procedure.

52. THE CIF DATA STANDARD -Driving Force: Syd Hall (IUCr/ Acta Cryst C) -Early Adopted by XTAL & SHELX(T)L. -Currently: WinGX,Crystals,Texsan, Maxus etc. -Acta Cryst. C/E – Electronic Submission -Acta Cryst.:Automatic Validation at the Gate -CIF data available for referees for detailed inspection (and optional calculations). -Data retrieval from the WEB for published papers -CCDC – Deposition in CIF-FORMAT.

53. VALIDATION QUESTIONS Single crystal validation addresses three simple but important questions: 1 – Is the reported information complete? 2 – What is the quality of the analysis? 3 – Is the Structure Correct?

54. IUCr CHECKCIF WEB-Service http://checkcif.iucr.org reports the outcome of: IUCr standard tests Consistency, Missing Data, Proper Procedure, Quality etc. + Additional PLATON based tests Missed Symmetry, Twinning, Voids, Geometry, Displacement Parameters, Absolute Structure etc.

55. ALERT LEVELS ALERT A – Serious Problem ALERT B – Potentially Serious Problem ALERT C – Check & Explain ALERT G – Verify or Take Notice

56. ALERT TYPES 1 - CIF Construction/Syntax errors, Missing or Inconsistent Data. 2 - Indicators that the Structure Model may be Wrong or Deficient. 3 - Indicators that the quality of the results may be low. 4 - Cosmetic Improvements, Queries and Suggestions.

57. In-House Validation with PLATON -Details: www.cryst.chem.uu.nl/platon -Available for UNIX/LINUX, Windows, Mac-OSX -Driven by the file CHECK.DEF with criteria, ALERT messages and advice. -Unix: platon –u structure.cif -Result on file: structure.chk -Applicable on CIF’s and CCDC-FDAT

58. EXAMPLE OF PLATON GENERATED ALERTS FOR A RECENT PAPER PUBLISHED IN J.Amer.Chem.Soc. (2007) Attracted special attention in Chemical and Engineering News Properly Validated ?

59. Problems Addressed by PLATON/CIF-CHECK -Missed Higher Space Group Symmetry -Solvent Accessible Voids in the Structure -Unusual Displacement Parameters -Hirshfeld Rigid Bond test -Misassigned Atom Type -Population/Occupancy Parameters -Mono Coordinated/Bonded Metals -Isolated Atoms (e.g. O, H, Transition Metals)

60. More Problems Addressed by PLATON -Too Many Hydrogen Atoms on an Atom -Missing Hydrogen Atoms -Valence & Hybridization -Short Intra/Inter-Molecular Contacts -O-H without Acceptor -Unusual Bond Length/Angle -CH3 Moiety Geometry -To be extended with tests for new problems ‘invented’ by authors.

61. Additional Problems Addressed by PLATON/FCF-CHECK Information from.cif and.fcf files Report on the resolution of the data Report about randomly missing data Check the completeness of the data (e.g. for missing cusps of data Report on Missed (Pseudo) Merohedral Twinning Report on Friedel Pairs and Absolute Structure Next Slide: ASYM VIEW Display for the inspection of the data completeness 

62. Missing cusp of data Section in reciprocal space

63. The Missed Symmetry Problem -Up to 10 % of the structures in space groups such as Cc have higher symmetry (e.g. C2/c, R-3c, Fdd2 etc.) than was originally reported. (To be Marshed is not good for your scientific reputation as a crystallographer). -MISSYM (Y. LePage) & PLATON/ADDSYM algorithm addresses the problem. - Next Slide: Example 

64. Organometallics (2004) 23,2310

65. Change of Space Group ALERT

66. Things to be Checked when ADDSYM suggests a new Space Group Consistency of the new cell parameters with the new crystal system (90.2 = 90 ?) Proposed new symmetry consistent with the reflection data ? Analyse the new implied systematic absences Case of Pseudo-symmetry ? Analyze potential disorder (real/artifact) Successful re-refinement

67. Incorrectly Oriented O-H The O-H moiety is generally, with very few exceptions, part of a D-H..A system. An investigation of structures in the CSD brings up many ‘exceptions’. Closer analysis shows that misplacement of the O-H hydrogen atom is in general the cause. Molecules have an environment in the crystal ! Example 

68. Example of a PLATON/Check Validation Report: Two ALERTS related to the misplaced Hydrogen Atom

69. Unsatisfactory Hydrogen Bond Network Validation Looks at inter-molecular contacts ALERT ! Correct !

70. Wrong Structures Sometimes (semi) automatic structure determination procedures can come up with ‘reasonably looking’ but wrong structures. Structure validation software should send out proper ALERTS to the investigators (e.g. IUCr Checkcif)

71. Structure Determination Artifacts Pseudo-symmetry easily results in false structures (often requiring a disorder model). Example: An Organometallic-AuCl compound from the CSD with the Cl in the wrong position  Very Short C-H..Cl ?! ALERTED by validation (C..Cl = 2.19 Ang) Moving Cl to the correct position drops R from 4 to 2 % (  see next two slides).

74. Consult the CSD It is a good idea to always consult the CSD for previous reports on structures related to the one at hand (in particular when the result looks unique). The statistics provided by VISTA (CCDC) can be very helpful for this. However, be aware, such an analysis often shows outliers. Many of those appear to be errors. Example: A search for short S..S contacts gave:

75. Entry from the CSD H S

76. But with Space Group Symmetry => Different structure with S-S Bond !

77. THE MESSAGE -Validation should not be postponed to the publication phase. All validation issues should be taken care of during the analysis. -Everything unusual in a structure is suspect, mostly incorrect (artifact) and should be investigated and discussed in great detail and supported by additional independent evidence. -The CSD can be very helpful when looking for possible precedents (but be careful  )

78. CONCLUSION Validation Procedures are excellent Tools to: -Set Quality Standards (Not just on R-Value) -Save a lot of Time in Checking, both by the Investigators and the Journals (referees) - Point at Interesting Features (Pseudo-Symmetry, short Interactions etc.) to be discussed. -Surface a problem that only an experienced Crystallographer might be able to Address -Proof of a GOOD structure.

79. Additional Info http://www.cryst.chem.uu.nl (including a copy of this powerpoint presentation) Thanks for your attention !!