A story of success or how to solve a structure?

Back to the Crunch page

Newsgroups: sci.techniques.xtallography
From: huffman@indiana.edu (John C. Huffman)
Subject: Want to solve a structure?
Organization: Indiana University Molecular Structure Center
Date: Tue, 26 Apr 1994 16:34:52 GMT

I have an interesting problem if someone wants a challenge. A colleague has an organic compound (C28H37O6N) which he has been able to get fairly decent crystals of. Examination of the crystals (thin triangular plates, looking much like an arrowhead) revealed twinning, as evidenced by omega scans. I was finally able to determine that the twinning was parallel to the thin dimension of the crystal and was able to split it into two fragments, one of which no longer shows evidence of twinning. A comprehensive search of reciprocal space locates a set of diffraction maxima indexable as triclinic, with cell parameters suggesting Z=4. Since the material is enantimerically pure (according to the student) there must be four independent molecules present (ca 140 non-hydrogen atoms). I have spent some time trying MULTAN and various options in SHELX to no avail. The data are collected at low temperatures (-173C) and appear to be well-behaved. My lack of success may simply be due to my inexperience with "large" equal atom problems, or there may still be a twinning problem. If someone wants to take a crack at the solution, I have placed SHELX .hkl and .ins files in directory /pub/datasets on our anonymous ftp server. The server is:

Since the best you can get is probably an acknowledgment for your efforts, I am willing to offer one low-temperature data set (up to 4000 reflections, air-stable compound) for the first solution.

If I have any takers and a successful solution, I will post details in case anyone else has problems of interest they would like to share.

=============================================
John C. Huffman Senior Scientist
Indiana University Molecular Structure Center
Bloomington, IN 47408-4001 USA
---------------------------------------------
huffman@indiana.edu (812) 855-6742
=============================================
Newsgroups: sci.techniques.xtallography
From: a4160007@horus.sara.nl (Milco Numan)
Subject: Re: Want to solve a structure?
Date: 29 Apr 94 13:59:12 +0100
Organization: Lab. for Crystallography, Univ. of Amsterdam

Dear Dr Huffman,

We have propably solved your problem structure 94112. Could you please send us some additional information about the positions of the N and O atoms so we can check our solution properly ?

FYI : We have used the program CRUNCH (De Gelder, De Graaff & Schenk, Acta Cryst A49, 287-293, 1993) using defaults (R2=0.093 based on E's).

Yours sincerely,

Kees Goubitz (fz@crys.chem.uva.nl) & Milco Numan (milcon@crys.chem.uva.nl)

Lab. voor kristallografie
Universiteit van Amsterdam
Nieuwe Achtergracht 166
1018 WV Amsterdam
Tel. + 31 (20) 525 7038/7041
Fax. + 31 (20) 525 5698

Newsgroups: sci.techniques.xtallography
From: huffman@indiana.edu (John C. Huffman)
Subject: Re: Want to solve a structure?
Organization: Indiana University Molecular Structure Center
Date: Tue, 26 Apr 1994 16:34:52 GMT

And the winner is........

Milco Numan and Kees Goubitz were able to solve the strucuture I posted earlier. Several different groups had a crack at it-- thanks to all! I would appreciate hearing from others who were trying, and I will post a summary of this rather unusual experiment! (Instead of saying solved by MULTAN... etc., can I say "solved by Internet....?).

Miloc and Kees are from the Laboratory for Crystallography, Universtiy of Amssterdam, The Netherlands.

Summary

The sample was submitted by Jason Speake of Prof. Paul Grieco's research group, Indiana University Chemistry Department. The sample consisted of diamond shaped plates of varying thicknesses. Several crystals were examined on the goniostat, and severe twinning was present in all samples, as evidenced by doubled peaks in several zones. A careful search of reflections in one hemisphere suggested that the twinning was parallel to the plate. A rather large crystal (0.3 mm in thickness) was cleaved along the thin dimension and the two fragments examined. While one fragment revealed some residual twinning, the other appeared to be single. All crystals were examined at -174oC using a gaseous nitrogen cooling system. A search of a limited hemisphere of reciprocal space located a set of diffraction maxima with no systematic absences or symmetry, indicating a triclinic space group. Based on the proposed formulation of C28H37O6N, a reasonable density was obtained for Z = 4. Since the structure was enantimerically pure, this indicated that the space group was probably P1, with four independent molecules present.

Data were collected using a standard moving crystal-moving detector mode with fixed background counts at each extreme of the scan. Data were collected to 45o in two-theta initially, and then to 55o when difficulty was encountered in the solution process (see below!). After correction for Lorentz and polarization terms, equivalent data were averaged (R = 0.063) to yield a set of 11,555 unique intensities. Initial attempts using MULTAN78 and SHELXL-PC failed to reveal the structure.

At this stage an nmr was examined closely (JS) and revealed that the crystals had apparently incorporated an earlier solvent, and indications were that it was present in all the crystals. This solvent was readily removed from the sample and the material was again recrystallized from hexane. The resulting crystals consisted of exceedingly thin plates of varying sizes. Careful examination revealed that the new form was, in fact, identical to the earlier phase, but not twinned. Thus, the presence of a small percentage of impurity led to twinning in the system. Examination of numerous crystals of the untwinned material failed to locate a crystal with diffracting qualities better than the original specimen cleaved from a twin.

Various "brute force" approaches were then employed using both SHELX (PC and unix versions) and MULTAN, but to no avail (in one case, TREF was set and the program was allowed to run over a two day period on a Silicon Graphics Indigo Elan). Part of the difficulty may have been the original investigators (JCH) inexperience with large non-centric equal-atom problems. As a last(?) resort, the data were placed in the IUMSC ftp.chem.indiana.edu archive and a message sent via the sci.techniques.xtallography newsgroup asking for assistance. Several crystallography laboratories obtained the data, but in general no solutions were obtained using the "common" direct methods software. At this point the data were examined at Universiteit van Amsterdam (Lab. voor kristallografie) and, with considerable effort, a solution was obtained using program CRUNCH (De Gelder, De Graaff & Schenk, Acta Cryst A49, 287-293 (1993)) on an IBM RS/6000 system. The resulting solution was then refined at the IUMSC using the GENLES least squares program of A. C. Larsen, (LANL) and other utility programs in the XTEL library.

The structure does indeed consist of four independent molecules in the P1 space group. After isotropic refinement of the non-hydrogen atoms, nearly all hydrogen atoms were visible. No attempt was made to refine the hydrogen atoms due to the number of variables involved. With all non-hydrogen atoms anisotropic, the data to parameter ratio is 6:1. All molecular parameters are within normally accepted values, and the four molecules are essentially identical in conformation (see BMFIT tables in report).

A final difference Fourier was featureless, the largest peak being 0.27e/A3.

Back to the Crunch page