HENRIA. LEVYAND S. W. PETERSON
1536
[CONTRIBUTION FROM
THE
Vol. 7?5
CHEMISTRY DIVISIONO F OAK RIDGENATIONAL LABORATORY]
Neutron Diffraction Determination of the Crystal Structure of Ammonium Bromide in Four Phases1 BY HENRIA. LEVYAND S . W. PETERSON RECEIVED MAY17, 1952 The crystal structures of IiDIBr in four phases have been studied by neutron diffraction. Low temperature cubic ND,Br contains ordered, parallel ammonium ians. In tetragonal NDIBr, adjacent strings of ammonium ions are anti-parallel, and bromide ions are displaced toward a square array of near hydrogen atoms. Room temperature NDdBr contains ammonium ions a t random in two orientations. High temperature cubic iYD4Br appears to contain ammonium ions in which one, two or three hydrogen atoms make minimal approaches to bromide neighbors; further specification cannot be given from present data. In all structures there appears to be a significant attraction between D and Br. Information about thermal motions in the crvstal is derived. includina- evidence for rotatorv oscillation of the ammonium ions. The length of the N-D link is 1.03 =t6.0.2 A .
Introduction Polymorphism in the ammonium halides has attracted much experimental and theoretical interest since the discovery by Simon2aof a sharp anomaly in the specific heat of NH4Cl a t about -30", and by Simon, uon Simson and RuhemannZh of similar anomalies in NH4Br and N H d . Through the efforts of P number of workers2-6 it has been established that the following transformations occur 184.3' NH4Cl: phase I (SaCl) t----f -30.5' phase I1 (CsCl) + (2nd order)
phase I11 (CsCl)
175'
N D4C1: phase I (NaCl) +--+ phase I1 (CsC1)
-23.8"
a + phase I11 (CsCl) (2nd order)
137.8'
NH4Br:? phase I (SaC1)
+--+
phase I1 (CsC1)
-38.1"
+-----+ - 58.4 phase I1 (CsCl) +--+
ND4Br: phase I (SaC1)
O
- 104" NHJ:
phase 111 (tetragonal)
-125O
- 17.6" phase I (SaC1) +--+ --41.Fi0 phase I1 (CsCI) -+---+-
phase 111 (tetragonal) phase I V (CsC1)
phase I11 (tetragonal)
X-Ray diffraction methods have confirmed the existence of the above crystallographic phases and established the arrangements of nitrogen and halide ions to be as indicated in parentheses In the tetragonal phases this arrangement is a slight distortion from that of CsC1. Complete structure determinations require also the placement of hydrogen atoms, and this information has not been given by the X(1) This work was performed under t h e auspices of t h e Atomic Energy Commission and was presented in part before the International Congress of Pure a n d Applied Chemistry, New York, 1951. (2) (a) F. Simon, A n n . Physik, [4] 68, 241 (1922); (b) F. Simon, CI. von Simson and M . Ruhemann, Z . physik. Chew., 129, 339 (1927) ( 3 ) F. Simon a n d R. Bergmann, ibid., B8, 255 (1930). (4) J. A. A. Ketalaar, Nutare, 134, 250 (1934). ( 5 ) J. Weigle a n d H. Saini, H e h . Phys. A d a , 9,51.5 (1930). ( 6 ) A. Smits, J. A . A . Ketelaar a n d G. J. Muller, Z. p h y s i k . ( ' h e m , 8176, 359 (1936). (7) Recently C. C. Stephenson a n d H. E. Adams, J . Chein. Phys.. '20, 1658 (19521, reported a n e w transition in NHIBr at 78OK. o n cooling, 108.6'K. o n warming. This may correspond t o conversion between phase 111 and phase IV.
ray studies. I t is required, of course, for an understanding of the A-transition in the chloride and the corresponding transitions in the other salts. PaulingSonce suggested that these transitions mark the onset of essentially free rotation of the ammonium ions in the crystal. Frenkele suggested an zilternative picture in which the transitions mark the onset of disorder; this view was given detailed description and a theoretical treatment of the Bragg-Williams type by Nagamiya.Io A great deal of evidence has now accumulated favoring the second of these hypotheses, particularly the fact, pointed out by Lawson," that in NH4C1 phase I1 the specific heat a t constant volume exceeds by 3 cal./mole-deg. the value expected for free rotation. The results of analysis of infrared and Raman data on the ordinary and deutero chlorides and bromides a t various temperatures b y Wagner and Hornig12also support the disorder 'hypothesis. Neutron diffraction, because of the possibility of locating hydrogen atoms, is a most suitable tool for establishing the complete structures of these materials. Such studies of NH4Cl and ND&l in phase 1113and ND4Cl in phase II114ahave already been reported and have confirmed the order-disorder hypothesis (a report that NlhC1 in phase I1 is ordered1@has proved i n c ~ r r e c t . ' ~ We ~ ! ~have extended these studies with determinations of the hydrogen positions in all four phases of ND4Br and report the results below. *4 preliminary report covering phases IV and I1 was made earlier. Experimental Deuteroammonium bromide was selected for investigation because of the existence of four phases, and because deuterium is better suited to powder neutron diffraction studies than is hydrogen.16 Reagent grade NH4Br was deuterated essentially t o completion by means of repeated solution in 99.8y0 D?O followed by evaporation of the solvent, with precautions to preveut exchange with atmospheric moisture. The chemical purity of the vacuum-dried sample was established by chemical analysis and by X-ray diffrac( 8 ) L. Pauling, Phys. Res.. 36, 480 (1930). (9) J. Frenkel, Acta Physicocheirz., 3, 23 (1935). (IO) T. Nagamiya, PYOC.Phys. .Math. SOC.J a p a n , 24, 137 (1942); 26, 540 (1943). ( 1 1 ) A . W. L a w s o n , Phys. Rev,67, 417 (1940). (12) E. L . Wagner a n d D. 1'. Hornig, J . Chem. P h y s . , 18, 296, 305 ( 1 950). (13) Henri A. Levy a n d S. W. Peterson, Phys. Rev., 86, 766 (1952). (14) (a) G. H. Goldschmidt a n d D . G . Hurst, ibid., 83, 88 (1951). (b) G . H. Goldschmidt and D. G. Hurst, ibid., 86, 797 (1952). (15) Iienri A. Levy a n d S. W. Peterson, ibid.. 83, 1270 (1951). ( I O ) C. G. Shull. E . 0. Wollan, G. A. Rlorton and a'.L. Davidson. i b i d . , 73, 842 (1948).
April 5 , 1953
CRYSTAL STRUCTURE OF AMMO"
BROMIDEIN FOUR PHASES
I537
tion. The isotopic purity was measured by neutron trans- characteristic temperature, 400'K. Observed data are mission and also by determination of the density of a re- listed in Tables I-V. Uncertainties appended correspond covered final portion of solvent; the two methods were in to estimated standard deviations arising from countitlg excellent agreement, giving 97.5% D. statistics, background estimation and systematic errm in Diffraction patterns were obtained from the finely pow- standardization. dered sample packed into a cylindrical, thin-walled (about Calculated values ol jFt were obtained from speciallza0.006") aluminum container about 1 cm. in diameter. The tions of the usual formulam neutron spectrometer is similar to one described by Wollan F h k t = Zf, exp [ - B , (sin 6/k)*]exp [2ai(hx, byl IzL)] and Shull,17 but differing from it in that the scattered neutron intensity is measured with a linear count-rate meter where the summation is over a unit cell, x,,yLand 2%are the and recorded continuously as a function of scattering angle position parameters of atom i in terms of fractions of the on a strip-chart recorder. A servo mechanism holds the unit cell dimensions, B , is a temperature factor coefficient, primary neutron beam to constant intensity by adjusting the and, in the neutron case, the coherent scattering amplitude thickxiess of an interposed absorber. The primary beam is f, is given by (uCOh/4?r)'/2. Values of the coherent scattering monochromatized bv reflection from a (111) face of a single cross sections Uooh for many elements and nuclides have been crystal of copper, Gielding a narrow band ofowave lengths centering a t 1.16 A . Soller slits can be inserted ahead of the monochromatpr and in the entrance of the detector to adjust the resolution of spectrometer to the optimum for the problem a t hand. For data a t temperatures other than ambient, the sample and cell were placed in an evacuated aluminum cryostat. The total number of neutrons recorded in each Bragg reflectioh F (Ehk~in the equation below) was I determined directly from the chart recording of counting rate versus time, integrating by means of the trapezoid rule, and occasionally checking by a triangle approximation. For this purpose the diffuse background level must be estimated from values of the scattered intensity between resolved peaks. A considerable uncertainty sometimes is attached t o this estimation because of incomplete resolution of peaks and the considerable structure which appears in the background of some of the patterns. This uncertainty is reflected in the precision of the resulting structure factor values. Fig. 1.TTypical neutron diffraction patterns from four phases of KDdBr. The Suitable corrections, where necessary, were made for scattering patterns shown for phases I1 and I11 were made with greater collimination, produced from the aluminum container. by insertion of soller slits. The points were read from chart recordings of the counting Observed values from several pat- rate. The dashed lines represent estimated diffuse background counting rates. terns. suitably normalized. were averaged, and from them absolute values of jF were ob- measured by Shull and W01lan.l~ In this work the values tained by means of the following formula appropriate to the used werefN = 0 . 9 4 , z 1 f ~=, 0.67 and f~ = 0.62 (for 97.5% case of a cylindrical sample bathed in radiation D, 2.5% H)JS In the early stages of the study, an earlier , was used; this resulted in the shorter value of f ~ 0.85, N-D distance reported previously.I6
+
This expression is a minor adaptation of that commonly used for the X-ray case. The symbol IOrepresents the incident intensity, E the counting yield of the detector, V the sample volume, I , w , w and r the height, width, angular velocity and radius of travel of the sensitive aperture of the detector, p' and p , respectively, the packed powder and crystal densities, 0 the Bragg angle, N the number of unit cells per unit volume, j the multiplicity of the reflection and F i t s structure factor. The factor A corrects for absorption in the sample; it has been tabulated18 as a function of p R and 8, where p is the total linear attenuation coefficient of the sample, measured for 1.16 A. neutrons, and R-is the radius of the sample. The quantity in parentheses was numerically evaluated from similar neutron diffraction data, prepared under identical conditions, from the ( l l l ) , (200) and (220) reflections of pure powdered nickel. The coherent cross section1s upi, was taken as 13.4 barns, and the temperature factor for nickel was calculated from the Debye (17) E 0 Wollan and C. G . Shull, P h y s . Rea , 7 8 , 830 (1948)
Low Temperature Cubic ND4J3r (Phase IV) X-Ray studies of ND&r a t - 140' have shown" that nitrogen and bromine atoms have the CsCl arrangement,and that the unit cell dimension a0 is 3.981 A. A determination a t thispboratory by Mr. B. S. Borie yielded a0 = 4.010 & 0.001 A . (based on CuKa = 1.5418 A.) a t abaut -140'. A neutron pattern obtained a t liquid nitrogen temperature and illustrated in Fig. 1 was consistent with a0 = 4.00 A., showing that the arrangement of deuterium atoms in the crystal dues not call for a unit cell different from that given by X-ray data. Reasonable deuterium positions giving tetrahedral coordination around nitrogen are provided by positionszs (e) of space group Tdl: xxx, xx?, Xx?, x?x, with nitrogen in (a): 000 and Br in (b): fr3fr. Refinement of the model required the introduction of separate Debye-Waller temperature factors for N, D and Br, and yielded excellent agreement.
(18) "International Tables for the Determination of Crystal Structures," Val iI, Gebruder Borntraeger, Berlin, 1935, and Edwards Bros , Ann Arbor, Mich , 1944, p 584 (19) C. G Shull and E 0 Wollan, Phys R e v , is, 527 (1951)
(20) References 18,p. 567. (21) S. W. Peterson and Henri A. Levy, P h y s . Rev., 87, 462 (1952). (22) A. Smits and D . Tollenaer, Z . physik. Chem., B62,222 (1942). (23) T h e notation is that of reference 18.
HENRIA. LEVYAND S. W. PETERSON
1538
The effect of a rotatory oscillation of the ammonium ion, as found in phase I1 and described in the following section, was tested; an oscillation with half-angle a of 5' is consistent with but not necessarily required by the data. A comparison of jP is presented in Table I. The preferred parameter galues are: x = 0.148, a = 5', BN = 0.8 A.2, = 0.2 A.2, {D = 1.0 The S-D bond distance is 1.026 i. 0.02 A . TABLE I NEUTRONDIFFRACTION DATAFROM NDdBr, PHASE IV, A T - 195' jFz Obsd.
hkl
f0.4 'f1 . 4
i2:)
17.9 72.3 16.1 5.0 1.7 87.1 30.8
76,0
3,5
310 311 222
4.1 f2.1 13.3 f 3 . 5 41.5 i 8
100 110 111 200 210 211 220
a
f0.8
f 0.6 f 1.0 f 2.4 2.0
Calcd.a 18.0 71.8 15.6 5.1 0.6 84.6 31.9
%::)77,7 2.5 11.9 41.0
hkl
jF2 Obsd.
320 321 400
5.8 f 4 163.2 i 8
