International
Tables for
Crystallography
Volume H
Powder diffraction
Edited by C. J. Gilmore, J. A. Kaduk and H. Schenk

International Tables for Crystallography (2018). Vol. H, ch. 3.7, pp. 305-313

Section 3.7.2. Powder Diffraction File (PDF)

J. A. Kaduka,b,c*

aDepartment of Chemistry, Illinois Institute of Technology, 3101 South Dearborn Street, Chicago, IL 60616, USA,bDepartment of Physics, North Central College, 131 South Loomis Street, Naperville, IL 60540, USA, and cPoly Crystallography Inc., 423 East Chicago Avenue, Naperville, IL 60540, USA
Correspondence e-mail: kaduk@polycrystallography.com

3.7.2. Powder Diffraction File (PDF)

| top | pdf |

The PDF is a collection of single-phase X-ray powder patterns in the form of tables of characteristic interplanar spacings and corresponding relative intensities, along with other pertinent physical, chemical and crystallographic properties. The PDF contains various subfiles, which include alkaloids, amino acids, peptides and complexes, battery materials, bioactive compounds, carbohydrates, cement materials, ceramics (bioceramics, ferroelectrics, microwave materials, perovskites and semiconductors), common phases, education, explosives, forensic, hydrogen-storage materials, inorganics, intercalates, ionic conductors, Merck Index compounds, metals and alloys, meso- and microporous (clathrates, metal–organic frameworks and zeo­lites), mineral-related (minerals, gems, natural and synthetic), modulated structures, nucleosides and nucleotides, organics, pharmaceuticals, pigments and dyes, polymers, porphyrins, corrins and complexes, steroids, superconducting materials (conventional superconductors, superconductor reaction products, superconductor-related and high-Tc superconductors), terpenes and thermoelectric materials. There is an educational package for classroom use, and the complete PDF is available for educational use on a time-limited basis. A primary purpose of the subfile system is to limit the size of the search universe by applying prior knowledge of the system being studied. This greatly reduces the number of false positives in a database that contains hundreds of thousands of materials. Field experts are consulted to guide the criteria for subfile selection, allowing novices to use the subfiles without being a subject expert.

3.7.2.1. Sources and formats of the PDF

| top | pdf |

The data incorporated into the Powder Diffraction File are acquired through contributions from individual scientists, corporate laboratories, literature surveys and a Grant-in-Aid programme. Approximately 200 leading scientific journals are searched manually for powder-diffraction data. Additional literature surveys covering patents, dissertations and the remaining open literature are performed using various online resources and search techniques.

Release 2019 (the current release as of this writing) contains more than 893 400 unique material data sets. The large size and comprehensive coverage of the PDF is achieved through the ICDD's historical sources of powder data (searches of the original literature, contributions and the Grant-in-Aid programme) as well as current and historic collaborations with crystallographic database organizations. Each PDF entry is assigned a unique identifying number of the format ss-mmm-nnnn. The integer ss indicates the source of the data: 00, ICDD location/generation of powder data; 01, Inorganic Crystal Structure Database; 02, Cambridge Structural Database; 03, NIST (a short-term collaboration focused on metals and alloys); 04, Pearson's Crystal Data; 05, ICDD extraction of atomic coordinates from published sources (including incommensurate/modulated structures). Powder-diffraction data for sources 01 through 05 are computed from the crystal structures provided by these sources.

The Powder Diffraction File is designed and produced in several different formats in order to serve different groups of users. The PDF-2 database is designed for phase identification of inorganic materials; many common organic materials have also been added to this database. The PDF-4+ database is the most advanced database and is designed for both phase identification and quantitative analysis. This database has comprehensive coverage of inorganic materials and contains numerous additional features such as digitized (raw) patterns, molecular graphics and atomic coordinates to facilitate Rietveld refinements. The PDF-4+ database is also available as a portable full-function WEBPDF-4+ version. The PDF-4/Minerals database is a subset of the PDF-4+ database, and is the most comprehensive collection of mineral diffraction data. The PDF-4/Organics database is designed for phase identification of organic and coordination compounds. It contains data from ICDD sources (both experimental powder patterns and extraction of coordinates) as well as patterns calculated from CSD entries.

Advances in hardware, software and computing power have led to the collection of higher-quality powder data, and thus have necessitated higher-quality reference data to perform more complex multiphase analyses and total-pattern analyses. The PDF now includes tools that permit users to evaluate different types of data collected using different types of detectors and different sources, including X-rays, neutrons and electrons. The goal is to include ideal specimen patterns in the PDF, patterns that can be modified by the user to correspond to the current experiment. The user can select the wavelength type and various instrumental parameters to simulate the whole diffraction pattern. A crystallite size calculation was added in 2007 and an orientation function in 2011.

Since 2006, the ICDD has begun to include several types of less-crystalline materials in the database, materials for which too much information is lost when reducing the raw data to a list of d-­spacings and intensities. These materials include clays and other layered materials, mixed-crystallinity polymers, amorphous materials and nanomaterials.

Nanomaterials often contain crystalline and amorphous fractions, and their powder patterns are difficult to generate from an ideal crystal structure. The ICDD has developed quality-evaluation methods for noncrystalline materials, and has established two additional quality marks: `good' (G) and `minimal acceptable' (M). These marks reflect the quality of the supporting data used to characterize the material. An amorphous material with a G quality mark has been characterized by independent analyses verifying the stated composition or thermogravimetric/differential scanning calorimetry analyses confirming the physical stability or the presence of a glass transition. A G quality mark indicates that the editors are satisfied that the pattern is representative of both the diffraction conditions and the stated chemistry and have confidence that the user can reproduce the pattern using similar conditions. The quality mark M indicates that the ICDD received some supporting documentation but it was insufficient for structural interpretation and classification of the material.

Great care needs to be taken in interpreting the patterns of mixtures of crystalline and amorphous phases, particularly in the definition and subtraction of the background. Significant work is under way to develop and adapt numerical techniques for processing full patterns of low-crystallinity materials.

3.7.2.2. Quality marks in the PDF

| top | pdf |

All data are critically reviewed and evaluated by the PDF editorial staff. Each pattern must pass through a four-tiered editorial review process before it can be included in the PDF. As technology evolves, the quality requirements for reference data also evolve. As a result, the information in the PDF is continuously reviewed and upgraded for accuracy and quality.

For many years, a quality mark has been assigned to each experimental PDF entry. A Star (*/S) pattern represents high-quality diffractometer or Guinier data. Several criteria must be satisfied for a pattern to be assigned a Star quality mark:

  • (i) The chemical composition must be well characterized.

  • (ii) The intensities must have been measured objectively; no visual estimation is allowed.

  • (iii) The pattern has a good range and an even spread of intensities.

  • (iv) The completeness of the pattern is sensible.

  • (v) The d-spacing of each reflection with d ≤ 2.500 Å is given to at least three decimal places. The d-spacings of reflections with d ≤ 1.2000 Å are given to at least four decimal places.

  • (vi) No serious systematic errors exist.

  • (vii) The |Δ2θ| value (i.e. the difference between the observed peak position and the position calculated from the unit cell) of a qualifying reflection is ≤0.05°. In the case of multiply-indexed reflections, only the minimum absolute Δ2θ is considered.

  • (viii) The average |Δ2θ| ≤ 0.03° for qualifying reflections.

  • (ix) No unindexed, space-group-extinct or impurity reflections are present.

An Indexed (I) quality mark indicates that the pattern has been indexed; therefore, the material is almost certainly single-phase. There is a reasonable range and spread of intensities, and the completeness of the pattern is sensible. The d-spacings of reflections with d ≤ 2.000 Å have at least three significant figures after the decimal point. No serious systematic errors exist. No qualifying reflection has |Δ2θ| ≥ 0.20° and the average |Δ2θ| is ≤0.06°. The maximum number of unindexed, space-group-extinct or impurity reflections is two, but none of these reflections are among the eight strongest lines.

A Blank (B) quality mark represents a mid-range quality. An O quality mark means that the data have been obtained from a poorly characterized material or that the data are known (or are suspected) to be of low precision and accuracy. Such patterns include those from multiphase mixtures or from a phase that is poorly characterized chemically. The O mark is commonly assigned to patterns for which no unit cell is reported, unless qualifying information indicates a single-phase material. Usually, the editor will have inserted a comment to explain why the O mark was assigned. For patterns with a unit cell, the following criteria are used to suggest the presence of two or more phases: the number of unindexed, space-group-extinct or impurity reflections is ≥3, or one of the three strongest peaks is unindexed.

Beginning with Release 2006, the quality-mark system was extended to patterns calculated from structural data supplied by ICDD partners. The focus of the quality mark is to determine the confidence level of the structural model used and its impact on the calculated pattern (especially for the purpose of phase identification). The major step involves several crystallographic and editorial checks by the ICDD, followed by extraction and flagging of the warnings/comments in the structural databases. The resulting calculated patterns are classified based on the significance and nature of the warnings. Any possible corrections that can be applied to resolve the errors are performed before publishing the calculated pattern.

The crystallographic checking rules are designed based on the expected quality of a contemporary crystal structure. An estimate of the missing electron density is made based on the difference between the reported composition and the structural composition. Transformations of nonstandard space groups are checked; the reported site multiplicities must match those generated by the symmetry operators. All of the eigenvalues of the anisotropic tensor matrix for each atomic displacement must be positive. All anisotropic tensor coefficients must be permitted by the site symmetry. Displacement coefficients should fall in the range 0.001 < U < 0.1 Å2. Isotropic displacement coefficients must be positive. Mixed displacement coefficients are converted to a standard type. The reported value of Z must be consistent with the sum of the site multiplicities. Lattice parameters are checked for missing decimal points, missing standard uncertainties and the magnitudes of the uncertainties. R factors close to the theoretical limits (0.83 for centrosymmetric structures and 0.59 for non­centro­symmetric structures) are signs of potential errors in the conversion to/from absolute/percentage values. Site occupancies cannot be greater than 1. Refining part of the structure as a group without locating the positions of the constituent atoms (for example, in C60) will generate a warning. Possible typographical errors in element symbols are checked by comparing the chemical formula, atomic coordinate list and chemical name. When a measured density is available, the percentage difference between the measured and calculated density is determined.

Many warnings/comments from the collaborating databases are used in assignment of the quality mark. Editorial comments on unusually short or long bond lengths or questionable bond angles are considered; the comment needs to be very specific for structures exhibiting disorder or partial/mixed occupancies. A listing of other types of comments considered is contained in the PDF-4+ database help documentation. Entries are assigned a quality mark of * (no warning found during data evaluation), I (minor warning), B (significant warning found), O (major warning), P (the structure was assigned by the editor based on a prototype) or H (hypothetical) according to the criteria in Table 3.7.1[link].

Table 3.7.1| top | pdf |
Criteria for the assignment of quality marks to calculated patterns in the Powder Diffraction File

A Star (*/S) pattern has no warnings.

Minor warning (I)Significant warning (B)Major warning (O)
Density calculated from reported and calculated compositions differ (1% < x < 3%) Density calculated from reported and calculated compositions differ (1% < x < 15%) Density calculated from reported and calculated compositions differ (15% < x)
     
No e.s.d. reported/abstracted on cell dimensions Lattice parameters taken from figure (approximated) Incorrect lattice parameters
     
Magnitude of e.s.d.s on lattice parameters > 1000 p.p.m. Missing decimal point in lattice parameter Incorrect space group
     
0.07 < R < 0.12 (single crystal), 0.10 < R < 0.15 (powder) 0.12 < R (single crystal), 0.15 < R (powder) Incommensurate/modulated structure. Only average structure of the subcell is given.
     
No R reported/abstracted Anisotropic displacement tensor is non-positive definite Published atomic coordinates are wrong
     
Reported Z is inconsistent with the sum of the site multiplicities Anisotropic tensor coefficient not permitted by site symmetry Structural database removed the entry corresponding to a published calculated pattern
     
Type of experiment (single crystal/powder) is not mentioned Magnitude of displacement coefficients outside the range 0.001 < U < 0.1 Å2  
     
Structure corrected by the editor Uiso < 0.0  
     
Difference between measured and calculated density > 2% Source-database warning on bond length/angle  
     
Misprint in original paper corrected in database Average structure of a modulated structure  
     
Site occupation factor > 1.0 Probable site-occupation factor deduced from the nominal composition  
     
  Part of the structure was refined as a group without locating the constituent atoms  
     
  Comments containing a reference to a contradicting structure exist  
     
  Structure determined from projections  
     
  Structure determined using electron diffraction  

3.7.2.3. Features of the PDF

| top | pdf |

Most users access the PDF through the software provided by their instrument manufacturer, but it is a powerful standalone database. The PDF is a large relational database consisting of many linked tables. The complete set of features can be accessed through the PDF front end supplied by the ICDD. It is possible to directly access a PDF entry by entering its PDF number. However, one can search for an entry or a class of entries through a series of search tabs. Queries from multiple tabs can be combined in a single search, or individual searches can be saved in a history and combined using Boolean operations. The results of such searches can be analysed as a group or can be used as subfiles for SIeve, the search/index phase-identification add-on for the PDF.

Selections on the main search screen permit selection by the source of data, quality mark, primary/alternate, ambient/non-ambient and subfile or subclass. The comprehensive nature of the PDF means that there are often many entries for an individual material. The ICDD editorial staff and volunteer task groups assign one experimental and one calculated entry (if present) as primary entries for each phase so that the user can avoid the duplication if desired. The other entries are designated as alternates. The subfiles and subclasses provide a convenient means for the user to limit the size of the search universe based on prior knowledge and result in faster searches and fewer false-positive matches.

Perhaps the most commonly used screen is the Periodic Table tab for chemistry searches. Individual elements, groups, periods and pre-defined selections (nonmetals, semimetals etc.) can be selected and combined in various ways. The `and' operation requires that all selected elements be present in the entries in the selection set, but other elements can also be present. The `or' operation requires at least one of the selected elements to be present. The `only' operation requires that all of the selected elements, and only those elements, be present in the hit. The `just' operation results in a hit list of entries that contain the selected elements in all combinations: elements, binaries, ternaries etc. The results of these four types of element searches can also be combined using Boolean operations. An alternative way of using periodic-table screening is through the labelling of each element with `yes', `no' or `maybe' to indicate elements that are known to be present, absent or unsure in the specimen.

The Formula/Name tab facilitates searches on formula, empirical formula, structural formula and formula type ANX [as in the Inorganic Crystal Structure Database (ICSD)]. The formulae may be exact or contain individual elements or strings. Searches on the number of elements present in the compound, as well as composition searches (by weight or atom per cent, i.e. wt% or at.%), are also possible.

The Formula/Name tab also permits searches on compound name, common name, mineral name and all names. It is also the screen from which searches on zeolite structure-type code (the International Zeolite Association codes are used) and mineral classification (according to the International Mineralogical Association) are performed.

Under the Reference tab, searches on author, journal name, CODEN, year, volume and title of the paper are possible. The titles were not originally included in PDF entries, but have been added to all entries in recent years. Also possible from this tab are searches on the Chemical Abstracts Service Registry Number (CASRN). Such searches are very powerful for organic compounds, with their complicated nomenclature. CASRNs are present for many, but not all, PDF entries.

The Classification and Crystallography tabs contain searches on Pearson symbol code, space group and space-group number, prototype structure, centrosymmetric/noncentrosymmetric and whether the entry contains atomic coordinates. Searches on the authors' cell, the Pearson's Crystal Data cell or the reduced cell are also possible. I find it useful to use fairly large tolerances (say 0.3–0.5 Å on edges) in such searches. I prefer to examine a longer list of potential matches which contains the correct phase, rather than risk missing an identification.

The Diffraction tab includes searches on the longest (lowest-angle or highest d-spacing) and strongest lines in the pattern. A line can be specified to be one of the three longest/strongest, or the first, second or third. This screen also includes searches on density, I/Ic (which is I/Icorundum, a measure of the inherent scattering power of a phase and useful in quantitative phase analysis), melting point, R value, colour and Smith–Snyder figure of merit. There are check boxes to select whether the entries in the hit list include `PD3' patterns (raw data) and property sheets. These property sheets are PDF documents embedded in an entry. These sheets are starting to be included for materials in subfiles that are defined by a particular property, such as superconductivity or thermoelectricity. These sheets are generated by groups of ICDD consulting editors.

Once a hit list has been generated, an individual entry can be selected (double clicked) to bring up the complete PDF entry. The results display can be customized using the Preferences menu (or by right clicking in the entry). The powder pattern can be plotted and additional PDF entries and/or raw data can be overlaid and scaled. In the Plot window, a PDF entry can be exported to several formats. The most useful is a CIF; the crystal structure described by the CIF can then be imported into to the user's graphics or Rietveld package.

Free-text searches of the comments are also included on this screen. These are particularly useful, as ICSD collection codes and CSD refcodes are included in the comments. If the user has the CSD installed on the same machine as the PDF, the PDF entry links live to the coordinates in the CSD entry.

By using the Results menu option when a hit list is displayed, ranges of cells in the spreadsheet can be selected and simple descriptive statistics (mean, median and estimated standard deviation) can be generated. Also under Results is a Graph Fields option. The variables used for the x and y axes of the plot can be selected and both scatter plots and histograms can be generated. Each of the points in such a plot is `live' and can be clicked to display the full PDF entry. Fig. 3.7.1[link] shows a plot of the cubic lattice parameter with respect to at.% Fe in FeO (Fe and O only, space group No. 225) under ambient conditions. From such data it is easy to generate a correlation between the Fe stoichiometry and the lattice parameter.

[Figure 3.7.1]

Figure 3.7.1 | top | pdf |

A plot generated by the Results/Graph Fields function in the Powder Diffraction File. The search was for entries containing only Fe and O and with space group No. 225 (resulting in FeO entries) measured under ambient conditions. One outlier was removed from the hit list manually. The trend in the cubic a lattice parameter with Fe content is apparent, as well as the large number of apparently stoichiometric FeO entries, some of which may not be correctly characterized.

An optional add-on module to the PDF is SIeve (Search Index). This is a peak-based search/match program which enables the use of a manually entered (or imported) peak list or derives a peak list from imported ASCII raw powder-diffraction data in several formats. It also has a flexible ASCII data-import module. Hanawalt, Fink, Long8 (the eight lowest-angle peaks in the pattern) or electron-diffraction searches can be carried out. Again, there is a Preferences option to customize the searches. A particularly useful (and easy-to-use) feature is the ability to apply a filter to the search/match. This filter can be selected from several pre-defined filters and/or any previous search in the session (stored in a history list). The combination of conventional search/match and Boolean searches can be very powerful, as illustrated in the next section.

3.7.2.4. Boolean logic in phase identification

| top | pdf |

Most phase identifications are carried out using the peak-based or full-pattern algorithms supplied by the instrument vendor. These often work well for major phases and can be customized to improve their success in identifying minor/trace phases. The native capabilities of the PDF (not all of which are accessible through some vendors' software) can be very powerful in identifying those extra peaks that result from a Rietveld difference plot (or any difference plot from pattern-fitting software) using the major phases. Below we use examples to illustrate several strategies.

3.7.2.4.1. Water-still deposit

| top | pdf |

A water still in my home eventually generates scale, much of which flakes off the walls, permitting easy analysis in the powder diffractometer. Any commercial search/match program will easily identify magnesian calcite (Fig. 3.7.2[link]; files kadu1389.gsas, kadu1389.raw and iitd26_0510.prm, available in the supporting information). There are, however, three additional weak peaks at d/I = 4.788/38, 3.3089/51 and 2.3697/56. In Naperville, Illinois, the tap water comes from Lake Michigan. The bedrock underlying the Chicago region is the Racine Dolomite. Given the identity of the major phase in the scale and the source of the water, it seems likely that any minor phases will be mineral-related and contain some combination of the elements Ca, Mg, C, O and H (to include the possibility of hydrates and hydroxides). Accordingly, a search of mineral-related entries containing `just' the elements Ca, Mg, C, O and H was performed and used as a filter in a Hanawalt search using these three peaks. This limits the search universe to 692 of the 328 660 entries in the PDF-4+ in 2012. The seven highest goodness-of-match entries in the hit list were brucite, Mg(OH)2. This phase was added to the Rietveld refinement. Analysis of the difference plot indicated an unaccounted-for peak at a d-spacing of 3.3089 Å. A search for mineral-related entries with the same chemistry and having one of their three strongest peaks in the range 3.309 (30) Å yielded the vaterite polymorph of CaCO3 as the hit with the highest goodness of match. This phase was added to the Rietveld refinement. The final quantitative phase analysis was: 94.7 (1) wt% Ca0.84Mg0.16(CO3), 5.2 (4) wt% Mg(OH)2 and 0.2 (1) wt% vaterite.

[Figure 3.7.2]

Figure 3.7.2 | top | pdf |

The result of applying a commercial search/match program (Jade 9.5; Materials Data, 2012[link]) to the powder pattern of a water-still scale. Weak peaks not accounted for by the major magnesian calcite phase are apparent and additional tools in the Powder Diffraction File were needed to identify the additional phases.

The composition of the major phase was refined, constraining the sum of the Ca and Mg site occupancies to equal 1.0. To understand how this fitted with previous magnesian calcites, a search for compounds containing only Ca, Mg, C and O, ambient conditions and space group No. 167 was carried out. Some manual editing of the hit list was required. Adjusting the preferences to include the display of composition in at.% and the unit-cell volume made it convenient to plot the variation in unit-cell volume as a function of Mg content x in Ca1−xMgx(CO3) (Fig. 3.7.3[link]). This magnesian calcite in the water-still scale has a higher Mg concentration than most, but falls close to the trend line. The flexibility and content of the Powder Diffraction File makes such data mining relatively straightforward.

[Figure 3.7.3]

Figure 3.7.3 | top | pdf |

Variation of the unit-cell volume with the magnesium content in magnesian calcites in the Powder Diffraction File.

3.7.2.4.2. Vanadium phosphate butane-oxidation catalyst

| top | pdf |

Vanadyl pyrophosphate [(VO)2P2O7] catalysts are used commercially for the selective oxidation of butane to maleic anhydride. Modern third-generation search/match programs [using the background-subtracted, Kα2-stripped data (files goed80.gsas, GOED80.raw and d8v3.prm); Fig. 3.7.4[link]] had no trouble in identifying the desired major phase (VO)2P2O7, but had difficulty with the minor phases that were clearly present. Unless the display of duplicate entries is turned off, most programs will yield several duplicate hits at the top of the list. Both 00-050-0380 and 04-009-2740 are Star quality, but only the Linus Pauling File (LPF) entry 04-009-2740 contains atom coordinates for a Rietveld refinement. Entry 01-070-8726 has the lower-quality B mark.

[Figure 3.7.4]

Figure 3.7.4 | top | pdf |

The results of applying a commercial search/match program (Jade 9.5; Materials Data, 2012[link]) to the (background-subtracted, Kα2-stripped) powder pattern of a butane-oxidation catalyst. The first three patterns in the hit list had equivalent figures of merit. The PDF entries 00-050-0380 and 04-009-2740 had Star quality marks and 04-009-2740 contained the atomic coordinates necessary for a Rietveld refinement. Additional peaks are apparent. The phases that give rise to them were identified using the native capabilities of the Powder Diffraction File.

The native capabilities of the PDF proved helpful in identifying the minor phases. The lowest-angle peak not accounted for by the major phase is at a d-spacing of 7.2107 Å. A search for phases containing just the elements V, P, O and H (known from the synthesis procedure) and having one of their three strongest peaks in the range 7.21 ± 0.05 Å (an estimated range) yielded only the single hit 00-047-0967: H4V3P3O16.5(H2O)2. This is a low-precision (O quality mark) pattern from a US Patent (Harju & Pasek, 1983[link]), and the pattern contains only four lines. The comments in the PDF entry indicate that this hydrated phase was formed by exposing a catalyst to ambient conditions, so it seems chemically reasonable but poorly defined.

To see whether this phase had been better characterized by a crystal structure, the four peaks were entered into SIeve+ and a Hanawalt search using a wider than default tolerance of 0.3° on the peak positions and the `just' chemistry filter V, P, O and H was carried out. As expected, PDF entry 00-047-0967 was at the top of the hit list, but close to the top was entry 04-017-1008 (Shpeizer et al., 2001[link]): [H0.6(VO)3(PO4)3(H2O)3](H2O)4. The article by Shpeizer et al. (2001[link]) indicates that this phase was formed from an anhydrous precursor by exposing it to ambient conditions. The single-crystal structure was obtained at 173 K. The similarity of the two PDF entries (Fig. 3.7.5[link]) and the difference in data-collection temperatures makes it clear that these correspond to the same phase, and that the structure of [H0.6(VO)3(PO4)3(H2O)3](H2O)4 could be used in a Rietveld refinement.

[Figure 3.7.5]

Figure 3.7.5 | top | pdf |

Comparison of the low-quality experimental PDF entry 00-047-0967 with the high-quality calculated pattern 01-074-2749 located by searching the experimental pattern against the rest of the PDF. The similarity in patterns and chemistry demonstrated that the two phases were the same and that the coordinates used to calculate entry 01-074-2749 could be used in a Rietveld refinement of a butane-oxidation catalyst.

There were still unaccounted-for peaks at 3.5823 and 3.0760 Å. Under the assumption that these came from a single phase, two separate searches for phases containing just V, P, O and H and with one of their three strongest lines in the ranges 3.58 ± 0.03 and 3.08 ± 0.03 Å were carried out and then combined (using the History option) with a Boolean `and' operation. All five of the entries on the hit list corresponded to α-VOPO4. This yellow V5+ compound was consistent with the altered colour of the V4+-based catalyst, and is a common impurity.

Close examination of the Rietveld difference plot from a refinement including these three phases indicated that there was a weak shoulder at a d-spacing of 3.985 Å. A search for phases containing just V, P, O and H and having a strong peak near this d-spacing yielded β-(VO)(PO3)2, another common catalyst impurity (Fig. 3.7.6[link]). Including this compound as a fourth phase yielded a satisfactory Rietveld refinement and a quantitative analysis of 84.8 (1) wt% (VO)2P2O7, 5.9 (1) wt% [H0.6(VO)3­(PO4)3(H2O)3]­(H2O)4, 5.6 (1) wt% α-VOPO4 and 3.7 (1) wt% β-VO(PO3)2.

[Figure 3.7.6]

Figure 3.7.6 | top | pdf |

The four crystalline phases identified in a butane-oxidation catalyst.

3.7.2.4.3. Valve deposit from a piston aviation engine

| top | pdf |

Applying a commercial search/match program to the diffraction pattern of a deposit from a valve in a gasoline-powered aircraft engine easily identified quartz and corundum. The specimen was scraped from the valve seat and micronized. The corundum represents abrasion from the elements of the micronizing mill, as it was not present in the pattern of the as-scraped sample. Metal particles were visibly present in the deposit, so one could reasonably guess the presence of both ferrite and austenite (Fig. 3.7.7[link]; files maso04.gsas, maso04.rd and padv.prm). A Rietveld refinement using these four phases was carried out.

[Figure 3.7.7]

Figure 3.7.7 | top | pdf |

The four phases identified in a valve deposit from an aircraft engine by automated search/match methods and guessing based on the appearance of the sample. The pattern has had the background and Kα2 peaks removed.

Six peaks picked from the difference plot were entered into SIeve+ and a Hanawalt search was carried out. No chemically reasonable simple compounds were near the top of the hit list, so extra information was sought. An XPS analysis indicated the presence of Pb, Br, Fe, P, O and C (and H assumed). Aviation gasoline is still leaded, and ethylene dibromide is sometimes added as a lead scavenger. The result of a `just' chemistry search using these seven elements (6543/328 660 entries) was applied as a filter to the Hanawalt search. Near the top of the hit list was PbBr2. Although apparently surprising, this phase is reasonable given our chemical knowledge. Lead bromide was added to the Rietveld refinement. Further analysis of the difference pattern using the same techniques indicated the presence of cohenite, Fe3C, from the steel, and Fe3Fe4(PO4)6, the reaction product of the steel with a phosphate fuel additive. The final Rietveld refinement yielded a quantitative analysis of 26.5 (4) wt% austenite (γ-Fe, stainless steel), 47.9 (4) wt% ferrite (α-Fe, carbon steel), 17.7 (4) wt% quartz (sand/dirt), 2.9 (2) wt% PbBr2, 2.6 (2) wt% Fe3Fe4(PO4)6 and 2.2 (2) wt% cohenite (Fig. 3.7.8[link]).

[Figure 3.7.8]

Figure 3.7.8 | top | pdf |

The seven phases identified in the valve deposit from an aircraft engine.

3.7.2.4.4. Isocracker sludge

| top | pdf |

An isocracker is a refinery unit which simultaneously carries out cracking and isomerization reactions to produce more high-octane gasoline. A black deposit isolated from such a unit was surprisingly crystalline (Fig. 3.7.9[link]; files NALK157.gsas, NALK157.raw and padv.prm). It was easy to identify small concentrations of elemental sulfur, pyrrhotite-4M (now called pyrrhotite-4C), haematite, lepidocrocite and dolomite, but the major peaks did not match well those of any entry in the PDF.

[Figure 3.7.9]

Figure 3.7.9 | top | pdf |

The phases identified in a deposit from a refinery isocracker. At the time, the (NH4)Fe2S3 was a new phase, identified by analogy to KFe2S3, rasvumite.

It seemed likely that a mineral-related phase would serve as a structural prototype for an apparently new phase, so two separate searches for mineral-related phases with one of their three strongest peaks in the d-spacing ranges 7.09 ± 0.03 and 5.57 ± 0.03 Å were combined. The two hits in the search list were both uranium minerals. These seemed un­likely in a refinery deposit(!). Widening the search ranges to 7.09 ± 0.10 and 5.57 ± 0.07 Å yielded rasvumite, KFe2S3 (PDF entry 00-033-1018), as the second entry in the hit list.

The fit to the major peaks in the deposit was reasonable, but there should not be any potassium in a refinery deposit and none was detected in a bulk chemical analysis. When the jar containing the deposit was opened, it smelled strongly of ammonia. Ammonium and potassium ions are about the same size and often form isostructural compounds. The infrared spectrum of the deposit was dominated by bands of ammonium ions.

The potassium in the structure of rasvumite (PDF entry 01-083-1322, used as a reference) was replaced by nitrogen. Analysis of potential hydrogen-bonding interactions yielded approximate hydrogen positions in the ammonium ion. These positions were refined using a density-functional geometry optimization. This model yielded a satisfactory Rietveld refinement (Fig. 3.7.10[link]) and the quantitative analysis 45.7 (2) wt% (NH4)Fe2S3, 12.8 (4) wt% S8, 22.0 (6) wt% lepidocrocite (γ-FeOOH), 5.5 (5) wt% haem­atite (α-Fe2O3), 6.6 (3) wt% pyrrhotite-4C (Fe7S8) and 6.6 (3) wt% dolomite [CaMg(CO3)2; limestone environmental dust]. The powder pattern and crystal structure of (NH4)Fe2S3 are now included in the PDF as entry 00-055-0533.

[Figure 3.7.10]

Figure 3.7.10 | top | pdf |

The final Rietveld plot from refinement of the isocracker deposit.

3.7.2.4.5. Amoxicillin

| top | pdf |

The amoxicillin powder from a commercial antibiotic capsule was highly crystalline. Its powder pattern (files kadu918.gsas, KADU918.raw, d8v3.prm and KADU921.rd) was matched well by the PDF entries 00-039-1832 and 00-033-1528 for amoxicillin trihydrate, but there was an additional peak at a d-spacing of 16.47 Å (5.37° 2θ). With such a low-angle peak, it seemed prudent to measure the pattern again starting at 3°, and another peak was observed at d = 24.80 Å (3.56° 2θ).

A search of the PDF-4/Organics 2013 for phases having two such peaks among their longest (lowest-angle) peaks yielded entry 00-005-0010 for calcium stearate at the top of the hit list, as well as two lead stearates. We can safely assume that lead stearate is not present in a pharmaceutical. Calcium stearate, however, has its strongest peak at 1.76°, so another pattern was measured starting at 1.5° 2θ. This peak is indeed present (Fig. 3.7.11[link]).

[Figure 3.7.11]

Figure 3.7.11 | top | pdf |

Phases identified in amoxicillin powder from a commercial capsule.

The primary literature suggests that the compound in PDF entry 00-005-0010 is really calcium stearate monohydrate, and that its structure (like those of many other stearate salts) has not yet been determined. The CSD entry for amoxicillin trihydrate (AMOXCT10; Boles et al., 1978[link]) contained some incorrect H-atom positions and was missing an H atom, so these were corrected before a Rietveld refinement was carried out.

3.7.2.4.6. Pseudoephedrine

| top | pdf |

As P. W. Stephens was measuring the powder pattern of a commercial pseudo­ephedrine-based decongestant on beamline X16C at the National Synchrotron Light Source at Brookhaven National Laboratory, he noted that extra peaks were present. The lowest-angle peak was at a d-spacing of 12.73 Å, and other peaks occurred at 5.74, 4.62 (strongest) and 4.407 Å. A search in the PDF-4/Organics for compounds having the string `ephed' in the name, a long line at 12.73 ± 0.05 Å and a strong line at 4.62 ± 0.02 Å yielded the single hit 00-041-1946, pseudo­ephedrine hydrochloride, a reasonable im­purity in pseudoephedrine.

3.7.2.4.7. Commercial multivitamin: Centrum A to Zn

| top | pdf |

Commercial multivitamins are challenging phase-identification problems because they contain small concentrations of many different components. The application of a commercial search/match program to a pattern of Centrum A to Zn collected on beamline ID-32 at the Advanced Photon Source at Argonne National Laboratory using a wavelength of 0.495850 Å (files centrum.gsas and id320304.prm) easily identified brushite, CaHPO4(H2O)2, and sylvite, KCl (Fig. 3.7.12[link]).

[Figure 3.7.12]

Figure 3.7.12 | top | pdf |

Phases identified by automated search/match in a Centrum A to Zn multivitamin tablet. Additional phases were identified using the native capabilities of the Powder Diffraction File.

To identify additional phases, 64 peaks with d > 1.91 Å were picked from the plot and entered into SIeve+ in the PDF-4/Organics 2013 database. The PDF-4/Organics database was used to enhance the success in identifying organic compounds, and the relatively short d-spacing limit was used to ease the identification of the simple inorganic compounds which are often present in commercial vitamins.

A Hanawalt search using these peaks easily identified iron fumarate (00-062-1294), szmikite [MnSO4(H2O); 00-033-0906], L-­ascorbic acid (02-063-2295), monetite (CaHPO4; 01-070-0359) and calcite (CaCO3; 00-005-0586). Note that these hits come from four different data sources; searches based on just one source would not have identified all of these compounds.

There were strong high-angle peaks that had not yet been accounted for at d-spacings of 2.4762, 2.1068, 1.4900 and 1.4783 Å. These four peaks were entered into a new Hanawalt search, which identified periclase (MgO; 01-071-3631) and zincite (ZnO; 01-075-9742).

Superimposing the peaks for all of these compounds onto the raw data made it clear that there were broad peaks in the pattern at d-spacings of approximately 5.8750, 5.3273, 4.3277 and 3.9217 Å. Since the lowest and highest angles of these four were the best defined, separate searches for compounds having each of these peaks as one of their three strongest lines were combined using a Boolean `and'. Among the hit list was cellulose Iβ (00-060-1502), which is a common constituent of pharmaceuticals. The structure model from PDF entry 00-056-1718 was added to the Rietveld refinement as a ninth phase.

One last peak at 5.9915 Å was unaccounted for. A search for pharmaceutical-related compounds with this peak as one of the three strongest included nicotinamide (02-063-5340; niacin or vitamin B3). Ten phases were thus identified and these account for all of the peaks in the pattern.

References

Boles, M. O., Girven, R. J. & Gane, P. A. C. (1978). The structure of amoxycillin trihydrate and a comparison with the structures of ampicillin. Acta Cryst. B34, 461–466.Google Scholar
Harju, P. & Pasek, P. (1983). Vanadium–hydrogen–phosphorus–oxygen catalytic material. US Patent 4374756; PDF entry 00-047-0967.Google Scholar
Shpeizer, B., Ouyang, X., Heising, J. M. & Clearfield, A. (2001). Synthesis and crystal structure of a new vanadyl phosphate [H0.6(VO)3(PO4)3(H2O)3].4H2O and its conversion to porous products. Chem. Mater. 13, 2288–2296.Google Scholar








































to end of page
to top of page