Ultramicroscopy North-Holland
38 (1991) 41-45
Electron crystallography at atomic resolution: ab initio structure analysis of copper perchlorophthalocyanine D.L. Dorset Electron Diffraction Department, Medical Foundation of Buffalo, Inc., 73 High Street, Buffalo, NY 14203-1196,
USA
W.F. Tivol and J.N. Turner Wadsworth Center for Laboratories and Research, New York State Department of Health and School of Public Health, The University at Albany, Albany, NY 12201-0509, USA Received
24 April 1991
are High-voltage (1200 kV) electron diffraction intensities from - 100 A thick crystals of copper perchlorophthalocyanine used to determine the molecular packing at atomic resolution, thus greatly exceeding the structure detail observed by electron microscopy. Initial crystallographic phases were determined by direct methods often used in X-ray crystallography, i.e., locating the positions of heavy (Cl and Cu) atoms in the structure. All other atom positions were found in subsequent Fourier refinement (final R = 0.28). Calculated bond distances and angles are similar to those found in the earlier X-ray crystal structure of the unchlorinated parent compound.
1. Introduction In recent years, the electron microscope has been used to visualize molecular packing in thin organic crystals for a variety of materials including relatively radiation-stable aromatics, e.g. pigments [l], as well as the more electron-beam-sensitive aliphatics, such as n-paraffins [2] and linear polymers [3]. At conventional electron accelerating voltages (e.g. 100 kV), the typical image resolution is limited to about 3 A, largely owing to the phase transfer characteristics of the microscope objective lens, which are a function of spherical and chromatic aberration and the electron wavelength [4]. However, the stated goal of these studies [5] has been to observe structure that can be interpreted directly in terms of atomic positions, necessitating the use of instruments with higher accelerating voltages and/or better objective lens characteristics. The ambitious goal of atomic-resolution electron microscopy of organic crystals has been approached, but has not been fully realized. Most 0304-3991/91/$03.50
0 1991 - Elsevier Science Publishers
progress has been made with copper perchlorophthalocyanine (fig. l), for which the pioneering studies of Uyeda et al. [6] produced images of the molecular quatrefoil from epitaxially oriented crystals tilted 26.5 o around the monoclinic b-axis. After Uyeda et al. determined the effect of nonoptimal lens defocus [4] and dynamical electron scattering [7] on the experimental images, the best micrographs of this compound were obtained with
Cl
I
43 N-cNh Cl
Cl Cl
c’
C’ c,
I
Cl
Cl Cl
N
Cl
Cl @??+ Fig. 1. Molecular
B.V. All rights reserved
structure
Cl Cl
Cl
of copper
perchlorophthalocyanine.
42
D. L. Dorset et al. / Electron crystallography
a 500 kV electron microscope in which the positions of the heavier copper and chlorine atoms could be clearly discerned, especially after averaging by superpositioning [8]. The lighter-atom positions, however, were not well defined, and a need to minimize the dynamical scattering of electrons through the crystal by means of thinner specimens was suggested [8], even for this relatively high electron accelerating voltage. Since image-averaging techniques have been employed to seek an atomic-resolution structure from electron micrographs, it can be argued that this is equivalent to carrying out a crystal structure analysis using the electron diffraction intensity data from the same projection. Indeed, since lens aberrations are expressed as phase terms in the back focal plane of the microscope objective lens [9], their effect is lost when the diffraction intensities are recorded. Moreover, given recent success with the use of direct methods to determine undistorted crystallographic phases for electron diffraction structure analysis [lo], such a
at atomic resolution
determination may be the best prospect for taining an atomic-resolution crystal structure. validity of this approach is demonstrated by analysis of experimental data described in paper.
obThe the this
2. Materials and methods As described by Uyeda et al. [6], thin microcrystals of copper perchlorophthalocyanine for the present study were grown by evaporation onto KC1 (001) crystal faces that had first been cleaved and then outgassed in vacua just before the deposition of the organic material. (The compound was a gift from Dr. J.R. Fryer, University of Glasgow, who also showed us how to epitaxially orient the samples.) Typical thicknesses of these microcrystals are about 100 A. After evaporating a thin film of carbon on this deposited film, the organic layer was separated from the salt substrate by flotation on a water surface, and bare 300-mesh
Fig. 2. Experimental electron diffraction pattern from a thin crystal of copper perchlorophthalocyanine (CuCltsC,,N,). As discussed by Uyeda et al. [8], the unit cell dimensions are a =19.62 A,, b = 26.08 A, c = 3.76 A, /j = 116.5 O. The space group is C2/c. In the projection down the molecular columns (26.5 o tilt around the b axis), the apparent a’ = a sin p axis is 17.60 A. Thus, the structure determination is carried out with these two-dimensional diffraction data in rectangular plane group cmm, which is centrosymmetric.
D. L. Dorset et al. / Electron crystallography
copper electron microscope grids were used to pick up the remaining film, which was then airdried. Initially, selected-area electron diffraction experiments had been carried out at 100 kV with a JEOL JEM-lOOB7 electron microscope equipped with a side-entry goniometer stage to ascertain that the proper crystal orientation could be found via specimen rotation when the grid was tilted 26S” to the electron beam. In this preliminary work, however, it was clear that the observed diffraction intensity data were so perturbed by multiple-beam dynamical scattering that they were not useful for ab initio structure analysis. To remedy this situation, diffraction patterns were obtained at 1200 kV in an AEI EM7 high-voltage electron microscope equipped with a tilt-rotation specimen holder [ll]. As seen in fig. 2 (especially when compared with previously published 100 kV data [S]), the electron diffraction patterns from this compound clearly show the intensity variations due to the unit-cell Fourier transform to 0.91 A-’ resolution. The diffraction films were scanned with a Joyce-Loebl MkIIIC flat-bed microdensitometer to record the intensities of 198 unique reflections. A Wilson plot [12] indicated that an overall temperature factor correction was not required for the atomic scattering factors [13] before the normalized structure factor values 1E,, 1 were calculated in the usual way [14]. In order to assign phase values to the experimental diffraction amplitudes, the reflections with calculated 1E,, ) > 0.6 were used to determine Z2 three-phase invariants [14] in plane group cmm: 4 = &, + %%*+ +/I,,
(I)
where the Miller indices h, + h, + h, = 0 and, in general, h, # h, Z h,. These were arranged in sequence according to values of A = k
I +‘&A,
I>
(2)
where k is a scaling term and the triple invariants with the highest A values taken to be most probably correct. Since the zonaI projection in cmm is centered, the phase of only one reflection can be chosen to define the origin [14] and, as usual, h = (hk0) f gg0, where g is an even integer. The
43
at atomic resolution
values of three other reflections, which occur often in the triple invariant relationships, were assigned symbolic values a, b, c. (Thus 23 = 8 Fourier maps must be computed by permuting values 0, ~7 for these three unknowns.) From the 31 triples with highest A values, the phases of 27 reflections were found, 8 with unequivocal assignments and the others with single or summed symbolic values.
3. Results From the eight Fourier maps calculated by permutation of the centrosymmetric values possible for the symbolic phases, there was only one for which the positions of the central copper atom and peripheral chlorines appeared at chemically reasonable locations. These atomic positions were used to calculate an initial phase set for all 198 reflections, using the kinematical structure factor calculation: Fh = cf,
cos(27rh
l
q),
where F,, = 1Fh I exp(i$) and I#I= 0, 7~ and fj is the atomic scattering factor. The reverse Fourier transform to the first electrostatic potential map,
~(~)=~~lF,l‘=~~
cos27r(h*r-cp’),
(4)
h
where 9’ = @/2a, using all reflections began to show light-atom positions near the central copper atom, especially when a difference Fourier synthesis based on I F, I - I F, I was used [15]. With only one carbon atom missing, the calculated crystallographic R-factor was 0.50. After location of the remaining carbon position and refining the structure by Fourier techniques, the crystallographic R-value falls to 0.32 for all data or 0.28 for the 150 structure factors which correspond to I E,, I > 0.6. In the final electrostatic potential map, the positions of all atoms are clearly resolved (fig. 3). Using an assumed 26.5” tilt of the molecular plane, one can calculate bond distances and angles [15]. These are given in fig. 4. The values for the light-atom framework are quite similar to the ones tabulated in the crystal structure of copper
44
D.L. Dorset et al. / Electron ctystallography
at atomic resolution
Table 1 Atomic coordinates for copper (molecular tilt of 26.5 o assumed)
Fig. 3. Electrostatic potential map for copper perchlorophthalocyanine after Fourier refinement. All atoms of the structure are clearly resolved.
phthalocyanine [16] and the carbon-chlorine distances are also typical for chloro-aromatic compounds [17]. Although the crystallographic R-factor is somewhat higher than the values normally
perchlorophthalocyanine
Atom
x/a
Y/b
s/c
cu Cl1 Cl2 Cl3 Cl4 Nl N2 N3 Cl c2 c3 c4 c5 C6 c7 C8
0.000 0.081 0.157 0.271 0.405 0.000 0.093 0.117 0.136 0.203 0.265 0.327 0.055 0.036 0.074 0.036
0.000 0.302 0.202 0.120 0.057 0.070 0.000 0.090 0.042 0.025 0.055 0.025 0.106 0.158 0.202 0.246
0.000 0.142 0.275 0.475 0.709 0.000 0.163 0.205 0.238 0.355 0.464 0.573 0.096 0.063 0.130 0.063
a) The coordinates are referred to an orthorhombic axes: a = 17.60 A, b = 26.08 A,, c = 5.00 A.
‘)
cell with
accepted in X-ray crystal structure analyses, the ab initio direct phase determination followed by Fourier refinement proceeds as it would in X-ray crystallography with a lowering of this figure of merit to a minimum value as the atoms are refined to their optimal positions, which, moreover, correspond to chemically reasonable valence parameters. Final atomic coordinates are listed in table 1.
4. Discussion
Fig. 4. Bond distance and angles for copper perchlorophthalocyanine. The values correspond closely to the values found in X-ray crystal structures of similar compounds, as is shown by comparison of average bond lengths to published values (in parentheses): aromati: C-C 1.39 +O.OS A (1.39 + 0.02 A); pyrrole C-C 1.40 * 0.01 A (1.45 f 0.02 A); pyrrok C-N 1.41 f 0.04 A (1.37f0.01 A); azo-linkage C-N 1.30+0.01 A (1.33*0.004 A); C-Cl 1.70+0.05 A (1.72 kO.01 A). On the other hand, the Cu-N distance is somewhat shorter, 1.83 fO.OO A,, than the spacing found for the unchlorinated copper phthalocyanine (1.93 kO.01 A) [16]. (Standard deviations are found by averaging over equivalent bonds.)
It is interesting to note that this analysis could have been performed equally well by extracting structural information from a 500 kV electron microscope image for use in the phasing of electron diffraction intensities. Copper and chlorine positions in the images published by Uyeda et al. [5] are very close to the values found from the best direct phase determination described above. Thus, even though Uyeda’s laboratory also proposed the use of direct phase determination for filling in phase values where the objective lens transfer function changes contrast, as well as for resolution extension [18], their present images are also sufficient to initiate a total structure determination to
D.L. Dorser et al. / Electron crystallography
atomic resolution, given electron diffraction data obtained at high voltage. From this study and a number of other ones carried out recently with other representative compounds [lo], it is clear that the ab initio electron crystallographic determination of organic structure is a viable technique. In some cases, direct phase determinations by themselves are sufficient for the structure analysis. In others, high-resolution images are desirable so that additional phase information can be obtained to facilitate the use of direct methods for resolution enhancement. This study demonstrates that electron diffraction in the high-voltage electron microscope should make an important contribution to this emerging branch of crystallography by producing data that correspond closely to the kinematical scattering approximation, especially for organic compounds containing heavy atoms. Thus, for materials which can be prepared only as microcrystals, electron scattering can be used to find atom positions and, hence, molecular architecture.
Acknowledgements The authors would like to thank Dr. John R. Fryer for his gift of the copper perchlorophthalocyanine and instruction on how to epitaxially orient this sample. The high-voltage electron microscope facility is funded as a national resource in part by the NIH National Center for Research Resources under PHS grant RR01219.
at atomic resolution
45
References [l] J.R. Fryer, J. Electron
Microsc. Tech. 11 (1989) 310. [2] F. Zemlin, E. Reuber, E. Beckmann, E. Zeitler and D.L. Dorset, Science 229 (1985) 461. [3] J.-F. Revel and R. St J. Manley, J. Mater. Sci. Lett. 5 (1986) 249. [4] N. Uyeda and K. Ishizuka, J. Electron Microsc. 23 (1974) 79. [5] N. Uyeda, in: Electron Crystallography of Organic Molecules, Eds. J.R. Fryer and D.L. Dorset (Kluwer, Dordrecht, 1991) p. 147. [6] N. Uyeda, T. Kobayashi, E. Suito, Y. Harada and M. Watanabe, J. Appl. Phys. 43 (1972) 5181. [7] K. Ishizuka and N. Uyeda, Acta Cryst. A 33 (1977) 740. [8] N. Uyeda, T. Kobayashi, K. Ishizuka and Y. Fujiyoshi, Chem. Ser. 44 (1979) 4761. [9] K.J. Hanszen, Adv. Opt. Electron Microsc. 4 (1971) 1. [lo] D.L. Dorset and F. Zemlin, Ultramicroscopy 33 (1990) 227; D.L. Dorset, Proc. Natl. Acad. Sci. USA 87 (1990) 8541; D.L. Dorset, Macromolecules 24 (1991) 1175. [ll] J.N. Turner, D.P. Barnard, P. McCauley and W.F. Tivol, in: Electron Crystallography of Organic Molecules, Eds. J.R. Fryer and D.L. Dorset (Khtwer, Dordrecht, 1991) p. 55. [12] A.J.C. Wilson, Nature 150 (1942) 151. [13] P.A. Doyle and P.S. Turner, Acta Cryst. A 24 (1968) 390. [14] H.A. Hauptman, Crystal Structure Determination: The Role of the Cosine Seminvariants (Plenum, New York, 1972). [15] G.H. Stout and L.H. Jensen, X-Ray Structure Determination: A Practical Guide (Macmillan, New York, 1968) pp. 261, 416-419. [16] C.J. Brown, J. Chem. Sot. A (1968) 2489. [17] L.E. Sutton, Ed., Tables of Interatomic Distances and Configuration in Molecules and Ions (Chemical Society, London, 1958). [18] K. Ishizuka, M. Miyazaki and N. Uyeda, Acta Cryst. A 38 (1982) 408.
38 (1991) 41-45
Electron crystallography at atomic resolution: ab initio structure analysis of copper perchlorophthalocyanine D.L. Dorset Electron Diffraction Department, Medical Foundation of Buffalo, Inc., 73 High Street, Buffalo, NY 14203-1196,
USA
W.F. Tivol and J.N. Turner Wadsworth Center for Laboratories and Research, New York State Department of Health and School of Public Health, The University at Albany, Albany, NY 12201-0509, USA Received
24 April 1991
are High-voltage (1200 kV) electron diffraction intensities from - 100 A thick crystals of copper perchlorophthalocyanine used to determine the molecular packing at atomic resolution, thus greatly exceeding the structure detail observed by electron microscopy. Initial crystallographic phases were determined by direct methods often used in X-ray crystallography, i.e., locating the positions of heavy (Cl and Cu) atoms in the structure. All other atom positions were found in subsequent Fourier refinement (final R = 0.28). Calculated bond distances and angles are similar to those found in the earlier X-ray crystal structure of the unchlorinated parent compound.
1. Introduction In recent years, the electron microscope has been used to visualize molecular packing in thin organic crystals for a variety of materials including relatively radiation-stable aromatics, e.g. pigments [l], as well as the more electron-beam-sensitive aliphatics, such as n-paraffins [2] and linear polymers [3]. At conventional electron accelerating voltages (e.g. 100 kV), the typical image resolution is limited to about 3 A, largely owing to the phase transfer characteristics of the microscope objective lens, which are a function of spherical and chromatic aberration and the electron wavelength [4]. However, the stated goal of these studies [5] has been to observe structure that can be interpreted directly in terms of atomic positions, necessitating the use of instruments with higher accelerating voltages and/or better objective lens characteristics. The ambitious goal of atomic-resolution electron microscopy of organic crystals has been approached, but has not been fully realized. Most 0304-3991/91/$03.50
0 1991 - Elsevier Science Publishers
progress has been made with copper perchlorophthalocyanine (fig. l), for which the pioneering studies of Uyeda et al. [6] produced images of the molecular quatrefoil from epitaxially oriented crystals tilted 26.5 o around the monoclinic b-axis. After Uyeda et al. determined the effect of nonoptimal lens defocus [4] and dynamical electron scattering [7] on the experimental images, the best micrographs of this compound were obtained with
Cl
I
43 N-cNh Cl
Cl Cl
c’
C’ c,
I
Cl
Cl Cl
N
Cl
Cl @??+ Fig. 1. Molecular
B.V. All rights reserved
structure
Cl Cl
Cl
of copper
perchlorophthalocyanine.
42
D. L. Dorset et al. / Electron crystallography
a 500 kV electron microscope in which the positions of the heavier copper and chlorine atoms could be clearly discerned, especially after averaging by superpositioning [8]. The lighter-atom positions, however, were not well defined, and a need to minimize the dynamical scattering of electrons through the crystal by means of thinner specimens was suggested [8], even for this relatively high electron accelerating voltage. Since image-averaging techniques have been employed to seek an atomic-resolution structure from electron micrographs, it can be argued that this is equivalent to carrying out a crystal structure analysis using the electron diffraction intensity data from the same projection. Indeed, since lens aberrations are expressed as phase terms in the back focal plane of the microscope objective lens [9], their effect is lost when the diffraction intensities are recorded. Moreover, given recent success with the use of direct methods to determine undistorted crystallographic phases for electron diffraction structure analysis [lo], such a
at atomic resolution
determination may be the best prospect for taining an atomic-resolution crystal structure. validity of this approach is demonstrated by analysis of experimental data described in paper.
obThe the this
2. Materials and methods As described by Uyeda et al. [6], thin microcrystals of copper perchlorophthalocyanine for the present study were grown by evaporation onto KC1 (001) crystal faces that had first been cleaved and then outgassed in vacua just before the deposition of the organic material. (The compound was a gift from Dr. J.R. Fryer, University of Glasgow, who also showed us how to epitaxially orient the samples.) Typical thicknesses of these microcrystals are about 100 A. After evaporating a thin film of carbon on this deposited film, the organic layer was separated from the salt substrate by flotation on a water surface, and bare 300-mesh
Fig. 2. Experimental electron diffraction pattern from a thin crystal of copper perchlorophthalocyanine (CuCltsC,,N,). As discussed by Uyeda et al. [8], the unit cell dimensions are a =19.62 A,, b = 26.08 A, c = 3.76 A, /j = 116.5 O. The space group is C2/c. In the projection down the molecular columns (26.5 o tilt around the b axis), the apparent a’ = a sin p axis is 17.60 A. Thus, the structure determination is carried out with these two-dimensional diffraction data in rectangular plane group cmm, which is centrosymmetric.
D. L. Dorset et al. / Electron crystallography
copper electron microscope grids were used to pick up the remaining film, which was then airdried. Initially, selected-area electron diffraction experiments had been carried out at 100 kV with a JEOL JEM-lOOB7 electron microscope equipped with a side-entry goniometer stage to ascertain that the proper crystal orientation could be found via specimen rotation when the grid was tilted 26S” to the electron beam. In this preliminary work, however, it was clear that the observed diffraction intensity data were so perturbed by multiple-beam dynamical scattering that they were not useful for ab initio structure analysis. To remedy this situation, diffraction patterns were obtained at 1200 kV in an AEI EM7 high-voltage electron microscope equipped with a tilt-rotation specimen holder [ll]. As seen in fig. 2 (especially when compared with previously published 100 kV data [S]), the electron diffraction patterns from this compound clearly show the intensity variations due to the unit-cell Fourier transform to 0.91 A-’ resolution. The diffraction films were scanned with a Joyce-Loebl MkIIIC flat-bed microdensitometer to record the intensities of 198 unique reflections. A Wilson plot [12] indicated that an overall temperature factor correction was not required for the atomic scattering factors [13] before the normalized structure factor values 1E,, 1 were calculated in the usual way [14]. In order to assign phase values to the experimental diffraction amplitudes, the reflections with calculated 1E,, ) > 0.6 were used to determine Z2 three-phase invariants [14] in plane group cmm: 4 = &, + %%*+ +/I,,
(I)
where the Miller indices h, + h, + h, = 0 and, in general, h, # h, Z h,. These were arranged in sequence according to values of A = k
I +‘&A,
I>
(2)
where k is a scaling term and the triple invariants with the highest A values taken to be most probably correct. Since the zonaI projection in cmm is centered, the phase of only one reflection can be chosen to define the origin [14] and, as usual, h = (hk0) f gg0, where g is an even integer. The
43
at atomic resolution
values of three other reflections, which occur often in the triple invariant relationships, were assigned symbolic values a, b, c. (Thus 23 = 8 Fourier maps must be computed by permuting values 0, ~7 for these three unknowns.) From the 31 triples with highest A values, the phases of 27 reflections were found, 8 with unequivocal assignments and the others with single or summed symbolic values.
3. Results From the eight Fourier maps calculated by permutation of the centrosymmetric values possible for the symbolic phases, there was only one for which the positions of the central copper atom and peripheral chlorines appeared at chemically reasonable locations. These atomic positions were used to calculate an initial phase set for all 198 reflections, using the kinematical structure factor calculation: Fh = cf,
cos(27rh
l
q),
where F,, = 1Fh I exp(i$) and I#I= 0, 7~ and fj is the atomic scattering factor. The reverse Fourier transform to the first electrostatic potential map,
~(~)=~~lF,l‘=~~
cos27r(h*r-cp’),
(4)
h
where 9’ = @/2a, using all reflections began to show light-atom positions near the central copper atom, especially when a difference Fourier synthesis based on I F, I - I F, I was used [15]. With only one carbon atom missing, the calculated crystallographic R-factor was 0.50. After location of the remaining carbon position and refining the structure by Fourier techniques, the crystallographic R-value falls to 0.32 for all data or 0.28 for the 150 structure factors which correspond to I E,, I > 0.6. In the final electrostatic potential map, the positions of all atoms are clearly resolved (fig. 3). Using an assumed 26.5” tilt of the molecular plane, one can calculate bond distances and angles [15]. These are given in fig. 4. The values for the light-atom framework are quite similar to the ones tabulated in the crystal structure of copper
44
D.L. Dorset et al. / Electron ctystallography
at atomic resolution
Table 1 Atomic coordinates for copper (molecular tilt of 26.5 o assumed)
Fig. 3. Electrostatic potential map for copper perchlorophthalocyanine after Fourier refinement. All atoms of the structure are clearly resolved.
phthalocyanine [16] and the carbon-chlorine distances are also typical for chloro-aromatic compounds [17]. Although the crystallographic R-factor is somewhat higher than the values normally
perchlorophthalocyanine
Atom
x/a
Y/b
s/c
cu Cl1 Cl2 Cl3 Cl4 Nl N2 N3 Cl c2 c3 c4 c5 C6 c7 C8
0.000 0.081 0.157 0.271 0.405 0.000 0.093 0.117 0.136 0.203 0.265 0.327 0.055 0.036 0.074 0.036
0.000 0.302 0.202 0.120 0.057 0.070 0.000 0.090 0.042 0.025 0.055 0.025 0.106 0.158 0.202 0.246
0.000 0.142 0.275 0.475 0.709 0.000 0.163 0.205 0.238 0.355 0.464 0.573 0.096 0.063 0.130 0.063
a) The coordinates are referred to an orthorhombic axes: a = 17.60 A, b = 26.08 A,, c = 5.00 A.
‘)
cell with
accepted in X-ray crystal structure analyses, the ab initio direct phase determination followed by Fourier refinement proceeds as it would in X-ray crystallography with a lowering of this figure of merit to a minimum value as the atoms are refined to their optimal positions, which, moreover, correspond to chemically reasonable valence parameters. Final atomic coordinates are listed in table 1.
4. Discussion
Fig. 4. Bond distance and angles for copper perchlorophthalocyanine. The values correspond closely to the values found in X-ray crystal structures of similar compounds, as is shown by comparison of average bond lengths to published values (in parentheses): aromati: C-C 1.39 +O.OS A (1.39 + 0.02 A); pyrrole C-C 1.40 * 0.01 A (1.45 f 0.02 A); pyrrok C-N 1.41 f 0.04 A (1.37f0.01 A); azo-linkage C-N 1.30+0.01 A (1.33*0.004 A); C-Cl 1.70+0.05 A (1.72 kO.01 A). On the other hand, the Cu-N distance is somewhat shorter, 1.83 fO.OO A,, than the spacing found for the unchlorinated copper phthalocyanine (1.93 kO.01 A) [16]. (Standard deviations are found by averaging over equivalent bonds.)
It is interesting to note that this analysis could have been performed equally well by extracting structural information from a 500 kV electron microscope image for use in the phasing of electron diffraction intensities. Copper and chlorine positions in the images published by Uyeda et al. [5] are very close to the values found from the best direct phase determination described above. Thus, even though Uyeda’s laboratory also proposed the use of direct phase determination for filling in phase values where the objective lens transfer function changes contrast, as well as for resolution extension [18], their present images are also sufficient to initiate a total structure determination to
D.L. Dorser et al. / Electron crystallography
atomic resolution, given electron diffraction data obtained at high voltage. From this study and a number of other ones carried out recently with other representative compounds [lo], it is clear that the ab initio electron crystallographic determination of organic structure is a viable technique. In some cases, direct phase determinations by themselves are sufficient for the structure analysis. In others, high-resolution images are desirable so that additional phase information can be obtained to facilitate the use of direct methods for resolution enhancement. This study demonstrates that electron diffraction in the high-voltage electron microscope should make an important contribution to this emerging branch of crystallography by producing data that correspond closely to the kinematical scattering approximation, especially for organic compounds containing heavy atoms. Thus, for materials which can be prepared only as microcrystals, electron scattering can be used to find atom positions and, hence, molecular architecture.
Acknowledgements The authors would like to thank Dr. John R. Fryer for his gift of the copper perchlorophthalocyanine and instruction on how to epitaxially orient this sample. The high-voltage electron microscope facility is funded as a national resource in part by the NIH National Center for Research Resources under PHS grant RR01219.
at atomic resolution
45
References [l] J.R. Fryer, J. Electron
Microsc. Tech. 11 (1989) 310. [2] F. Zemlin, E. Reuber, E. Beckmann, E. Zeitler and D.L. Dorset, Science 229 (1985) 461. [3] J.-F. Revel and R. St J. Manley, J. Mater. Sci. Lett. 5 (1986) 249. [4] N. Uyeda and K. Ishizuka, J. Electron Microsc. 23 (1974) 79. [5] N. Uyeda, in: Electron Crystallography of Organic Molecules, Eds. J.R. Fryer and D.L. Dorset (Kluwer, Dordrecht, 1991) p. 147. [6] N. Uyeda, T. Kobayashi, E. Suito, Y. Harada and M. Watanabe, J. Appl. Phys. 43 (1972) 5181. [7] K. Ishizuka and N. Uyeda, Acta Cryst. A 33 (1977) 740. [8] N. Uyeda, T. Kobayashi, K. Ishizuka and Y. Fujiyoshi, Chem. Ser. 44 (1979) 4761. [9] K.J. Hanszen, Adv. Opt. Electron Microsc. 4 (1971) 1. [lo] D.L. Dorset and F. Zemlin, Ultramicroscopy 33 (1990) 227; D.L. Dorset, Proc. Natl. Acad. Sci. USA 87 (1990) 8541; D.L. Dorset, Macromolecules 24 (1991) 1175. [ll] J.N. Turner, D.P. Barnard, P. McCauley and W.F. Tivol, in: Electron Crystallography of Organic Molecules, Eds. J.R. Fryer and D.L. Dorset (Khtwer, Dordrecht, 1991) p. 55. [12] A.J.C. Wilson, Nature 150 (1942) 151. [13] P.A. Doyle and P.S. Turner, Acta Cryst. A 24 (1968) 390. [14] H.A. Hauptman, Crystal Structure Determination: The Role of the Cosine Seminvariants (Plenum, New York, 1972). [15] G.H. Stout and L.H. Jensen, X-Ray Structure Determination: A Practical Guide (Macmillan, New York, 1968) pp. 261, 416-419. [16] C.J. Brown, J. Chem. Sot. A (1968) 2489. [17] L.E. Sutton, Ed., Tables of Interatomic Distances and Configuration in Molecules and Ions (Chemical Society, London, 1958). [18] K. Ishizuka, M. Miyazaki and N. Uyeda, Acta Cryst. A 38 (1982) 408.
