I(s) =
2
N
1
g exp[2vri (rz. s)]
lowing the averaging procedure for I(s) described in the 0 0. 2 0.4 0. 6 0. 8 1.0 1.2 1.4 1.6
previous paragraph. In the vicinity of the Bragg maxima, wave vector 2 sinH/A (A']
the interval As between neighboring s values was 0.0061.
In the region between the peaks, As was 0.0175. The s
was varied between s =0.1 and 1.7. The one-dimensional FIG. 4. Comparison of the measured ( + --) and computed
plot I(s) was compared with the experimental interfer- ( ) interference functions for nanometer-sized crystalline
ence function I(s). All computed interference functions Fe. The model system assumed for the computations is a boun-
were convoluted by a Gaussian instrument function in or- dary structure consisting of four atomic layers in which atoms
der to simulate the effect of instrumental broadening of are randomly displaced. The displacements of the atoms corre-
the diffractometer used for the experiments. sponding to the two layers in the boundary core are assumed to
The effect of interparticle interference is not included be 15% of the NND, in the two adjacent layers the displace-
in I(s). However, interparticle interference has been ments are 7% of the NND. The displacement of 15% of the
shown to yield only nonvanishing scattering in the small- NND corresponds to the average nearest-neighbor displacement
angle region which is not within the scope of this paper. of the atoms in a Feg082O glass (Ref. 10).
10
I I I I I I
A comparison of the experimental and the computed data, if the nanometer-sized crystalline Fe is assumed to
interference functions is shown in Fig. 4 assuming a consist of 75 vol. % of 6-nm crystallites and 25 vol. oo f %%u
short-range-ordered grain boundary structure. The boun- 4-nm crystallites; the size distribution assumed was based
daries were assumed to have a thickness of four atomic on TEM observations. "
layers, which agrees with the present knowledge of the The computer simulations failed to reproduce the ex-
atomic structure of grain boundaries in metal. perimental I(s) for the models of multilayered boundary
Although the peak positions and the widths of the com- component (eight layers), smaller crystallite size (4 nm),
puted I(s) are consistent with the experimental observa- and large atomic displacements (60%%uo of the NND in the
tion, the calculated peak heights are excelled and the cal- first outer layer and 30% of the NND in the second outer
culated attenuation is slower than the experimentally ob- layer of the one-crystal system).
served one. Furthermore, the calculated background is Furthermore, it was not possible to reproduce the ex-
too low in the range between s=0. 1 and 0.45 (below the perimental data with an enhanced thermal diffuse scatter-
position of the first peak). The discrepancy increases at ing by assuming a model of an assemblage of small crys-
large s. The poor agreement between the computation
and experiment suggests that the model based on a short-
range-ordered grain boundary component cannot satisfac-
torily explain the experimental observation.
In a second set of computations, an attempt was made
to match the experimental data by assuming a non-short- -15
range-ordered grain boundary component. A structure of
this type was generated by displacing the boundary atoms
as described in the caption of Fig. 5. It may be seen that
the computed I(s) curve reproduces not only the heights
-10
and widths of all the peaks, but also matches approxi-
mately the background intensity, especially in the regime
between s=0. 1 and 0.45 where the fit with the model as-
suming a short-range-ordered interfacial component (Fig.
4) was particularly poor. It may be mentioned here that
J'I,', 3
the growing background intensity in this regime in Fig. 3
is consistent with the observation of Sass et al.
Nevertheless, some discrepancy between the computed .
and observed I(s) curves remains at large values of s. In I I
-15
(bj
-10
crystals in which the boundary atoms are randomly displaced in 6% of the ND ( ~- . ). No grain-boundary structure was in-
the same way as in the case of Fig. 5. cluded in the computation.
35 X-RAY DIFFRACTION STUD IES OF THF STRUCTURE Q 9089
p (
'
taie
atoms were rando ml y dis p laced by o an
At s=1.85, where the + -10
in the compute d in t ensiit t h [5% of
f NND, Fi . 7(b)] h of
the experimen a .
and 0.45, none off the computed p d curves cacan reproduce the
data. This discrepancy resuultss from neglect-
rain-boundary
ing grain- scattering an d dooes
curac of
ginate from the accuracy o the computations.
In fact, the agreem
reement between t h e experx erimental and
computed I(s) curves rves mmayy be seen for 6--nm Au crystals
iso ate from one another soo that
which are isolated a no grain
boundaries can be formed F&g. F . 8). In the computation, 0. 4 0. 8 1.0 1.2 1.4 1.6
'""""- """'" 6-- .ld-, "11---
'
e arameter of gold an d a displacement
isp parame-
ere used. The latter matches the am-
plitude of thermal vibration of 6-nm go cry
Q Q. 2
wave
O. g
vec tor 2 sine tA')
f 13.
In the previous iron computations,
p onstant mass den-
a con FIG. 9. Comparison of o the measured and comp uted interfer-
in to the bulk crysta ll ine density was of nnanometer-sized cryst alhne Fe.. To simulate
ence functions o
ot ex lain the density d efiiciitob o di th he atoms in the irstt ou ter
er layer are radially
Furthermore, iff t e atomst s in the first outer moved outwards (expansion) and then ran om y i
layer are displace ced b y 50% of the same way as in the case of Fig. 5.
atoms overlapped.d. In order to matc h thee c computations to
all measured density of the nanometer-
~, df,
e s ecimens, t h e d ensiit y variation was
g inate totally from a ensi y
aries. A lower density o t e in er
grain bounndaries.
,h W(r
-10
0
1 2 3 4 5 6 7 8
suit is shown in Fig. 9. As may be noted from the com- ray of 6-nm bcc Fe crystals were computed by assuming
parison between Figs. 5 and 9, the density variation and the boundary regions between crystals to be short-range
hence the atomic overlapping have little effect on the in- ordered or to consist of randomly displaced atoms. The
terference function. experimentally observed interference function can only be
To characterize quantitatively the atomic order in matched by the computations if the interfacial component
nanometer-sized crystalline materials, the probability of a nanometer-sized crystalline material is assumed to
function for interatomic distances of the interfacial com- have no short-range order.
ponent was computed for the structural model assuming
the grain-boundary component to be short-range ordered
(Fig. 4) or to consist of randomly arranged atoms (Figs. 5 ACKNOWLEDGMENTS
and 6). The computed functions are shown in Fig. 10. If
the boundary atoms are assumed to exhibit short-range or- The authors are indebted to Professor Ruppersberg and
der (Fig. 4), the corresponding probability function shows his associates (Universitaet des Saarlandes) for valuable
similar features as a glassy structure. However, if the advice in performing the x-ray measurements as well as
boundary atoms are displaced by 50% of the NND (Figs. for several very helpful discussions. The previous
5 and 6), the corresponding probability function resembles cooperation with Dr. P. Marquardt (University Cologne)
a structure similar to Fig. 1(c), in other words, a structure and the discussion with Professor C. N. Wagner, UCLA,
without short-range order. are gratefully acknowledged. The work has been finan-
cially supported by the Alcoa Foundation, the Bun-
V. SUMMARY desministerium fiir Forschung und Technologie (BMFT,
Contract No. 523-4003-03M0023-4), the Deutsche
To test the structural model of nanometer-sized crystal- Forschungsgemeinschaft, and the Volkswagen Founda-
line materials, the interference functions of a random ar- tion.
R. Birringer, U. Herr, and H. Gleiter, Trans. Jpn. Inst. Met. S. L. Sass, J. Phys. (Paris) 46, C4-71 (1985).
Suppl. 27, 43 (1986). In this reference and others, 7V. Tiensuu, S. Ergun, and L. Alexander, J. Appl. Phys. 35,
nanometer-sized crystalline materials have been called nano- 1718 (1964).
crystalline materials. 8X. Zhu, Ph. D. thesis, Universitaet des Saarlandes, 1986.
2H. Gleiter, Mater. Sci. Eng. 52, 91 (1982). F. Betts and A. Bienenstock, J. Appl. Phys. 143, 4591 (1972)
~
3C. G. Granqvist and R. A. Buhrman, J. Appl. Phys. 47, 2200 ' E. Nold, P. Lamparter, G. Rainer-Harbach, and S. Steeb, Z.
(1976). Naturforsch. 36a, 1032 (1981).
4C. N. J. Wagner, in Liquid Metals: Chemistry and Physics, E. Hort, Diploma thesis, Universitaet des Saarlandes, 1986.
edited by S. Beer (Dekker, New York, 1972), p. 257. ' G. S. Cargill III, in Atomic Energy Review, edited by U.
5D. T. Cromer, Acta Cryst. 18, 17 (1965). D. T. Cromer and J. Gonser (IAEA, Vienna, 1981), Suppl. No. 1.
B. Mann, Acta Cryst. A 24, 321 (1968) ~ J. Harada and K. Ohshima, Surf. Sci. 106, 51 (1981).
K. R. Milkove, P. Lamarre, F. Schmueckle, M. D. Vaudin, and