VOLUME )5, NUMBER 21 PHYSICAL REVIEW LETTERS 24 NovEMBER 1975
Two-, Three-, and Four-Atom Exchange Effects in bcc 3 He
J. H. Hetherington and F. D. C. Willard
Physics DePa&ment, Michigan State University, East Lansing, Michigan 48824
(Received 22 September 1975)
We have made mean-field calculations with a Hamiltonian obtained from two-, three-,
and four-atom exchange in bcc solid He. We are able to fit the high-temperature experi-
ments as well as the phase diagram of Kummer et al. at low temperatures. We find two
kinds of antiferromagnetic phases as suggested by Kummer's experiments.
We have been able both to fit the high-tempera, - ters R with an overall positive sign.
ture solid-'He magnetic data' and, through Without the o 4 terms in R only these two struc-
like that of Kummer et al. "
mean-field theory, to obtain a phase diagram
"
We consider only
three mechanisms of exchange, ' two-atom,
tures are possible in mean-field theory when on-
ly J, and J, are nonzero. Without proof, we
have assumed that consideration of these two
three-atom rings, and four-atom rings, and ad- phases is adequate even when the o' terms are
just their strengths to fit the data. present.
The Hamiltonian we take is of the form" The computation proceeded by examining at
each value of T and B all solutions of the mean-
K=-gpss J', o» o;,„—p. 5 Po; field equations for the scaf, naf, and paramag-
i v
- 4 Z J»» ia» ' (o» ' o» ) ( 0 ' +» ), netic phases and choosing the solution with low-
est free energy.
i/01
The high-temperature experiments can be sum-
where the sum over v indicates a sum over the
marized by noting that the partition function Z
neighbors of site i. The 0's are Pauli spin ma-
can be expressed in general in the form'3
trices. Ordinary two-atom exchange contributes
negatively to J, . Three-atom-ring exchange con- X ' lnZ = ln2+-', e2P2+. ..
tributes positively to J, and J, . Four-atom-ring +-', y'(1+-', n, p+. . .)+. . . , (2)
exchanges are of two kinds because there are two
kinds of closed four-atom circuits with nearest- where y= pBP. We note that specific heat at high
neighbor steps. One is folded (F) so that both T determines e, and Curie-Weiss 8 determines
pairs of opposite corners of the quadralateral are a, . Pressure experiments at zero field deter-
second neighbors. The second kind is planar (P)
diamond shaped so that ends of one diagonal are by integration. "
mine de, /dV and have been used to determine e,
Pressure experiments in mag-
second neighbors and ends of the other diagonal
are third neighbors. We have made calculations what self-contradictory, "
netic field" measure dn, /dV and, although some-
probably imply on inte-
using both kinds of four-atom rings but we are
able to fit the data using only the F-ring exchang- accepted. "
gration an n, somewhat smaller than is generally
We can express the results of all
es. The F-ring exchanges contribute negatively high-T experiments (adjusted to 24. 2 cm'/mole)
to Z, and J', and to two kinds of (o', o'2)(o', o'~)
~
by the statements that e, =6.94+0.3 (mK)' mainly
terms ("o"'), one where the dot products are be- on the basis of the results of Panczyk and Adams'
tween first neighbors and one where they are be- and that n, = —6.25+ 0.8 mK mainly from Kirk,
tween second neighbors. Osgood, and Garber. '
We solved the mean-field equations on the as- The theoretical expansions are n, = 8 J, + 6 J,
sumption that one of two antiferromagnetic states and e, =12J,'+9J,'+31.5K4', where J, and are J,
could exist: (i) the normal (spin-flop) antiferro- usual first- and second-neighbor exchange coef-
magnetic (naf) phase, where the two simple cubic ficients and K, is the strength of the o' term in
sublattices are the magnetic sublattices; and the F-ring exchange. K4 is defined so that the
(ii) the simple cubic antiferromagnetic (scaf) first- and second-neighbor exchange parts of the
phase, where each of the simple cubic sublattie- Hamiltonian due to F-ring exchange alone are J,
es is itself antiferromagnetic. The spins on the =3%4 and J2=2K4.
simple cubic sublattices are rotated 90' about B Our fit is obtained with J, = —0. 56 mK, J,
with respect to each other because this mini- =0.1'75 mK, and K, = —0.32 mK which lead to 0.,
mizes the nearest-neighbor o4 term, which en- = —5.7 mK and e, = 7. 2 (mK)' and to a phase dia-
1442
VoLUME 35& NUMBER 21 PHYSICAL REVIEW LETTERS 24 NOVEMBER 1975
I, O I I I I I I I I I I I I
J =-6 I I
J~ =-.I5
J I
= —.56
p. B/k = 0.
p. B/k = .2
K4 = —.
4 .4
p. B/k =
K4= —.52 p. B/k = .6
p. B/k = I.O
I.O— (b)
it
0
2 3 / 5—
/
/
0
naf
/ (c)
/
/
5— /
0
I I I I I I I I
I.O 2.0 3.0
naf paf'a T(mK)
scaf
(a) FIG. 2. Entropy versus temperature at several mag-
netic fields. This figure should be compared to Fig. 2
I i i i i I
of Ref. 7.
I.O 2.0
T(mK)
FIG. I. (a)' Phase diagram of He below 3 mk as de- region shows that a critical point also occurs in
termined by experiment of Ref 7(op. en circles) and this case but it is "hidden" by the lower-free-en-
according to mean-field theory based on coefficeints ergy scaf phase.
shown. (b) Inset shows how first-order transition with
critical point (CP) can be obtained with s1ight adjust- Taking a hopeful stance we might identify the
ment of parameters. naf-paramagnetic transition with the bump in
specific heat near 2 mK noticed by Dundon and
Goodkind, ' Halperin et al. , ' and Kummer et al. '
gram as shown in Fig. 1(a). In Fig. 1(a) the ex- In the eventuality that such identification cannot
perimental phase boundary from Kummer et al. ' be made we know that the naf-paramagnetic tran-
is shown by the open circles. We find three dis- sition temperature is rather sensitive to im-
tinct phases: To the right (above 1.8 mK) is the provements in the calculation beyond mean field.
paramagnetic phase. Below about 1.8 mK is the We hypothesize, however, that the scaf-naf
naf phase. Below about 1 mK and for pB/k below boundary would be less affected because an ex-
0.45 mK is the scaf phase. The dashed line indi- amination of the variation of free energy with
cates a separation of the naf phase into high-en- temperature shows that a small relative energy
tropy and low-entropy regions (taken as the line shift would have a large effect on the transition
where S/k =0.4). Note that the ratios of the sizes temperature of the naf-paramagnetic transition
of exchange (two-atom:three-atom:P:F) used to while in the case of the other transitions in Fig.
calculations. ' "
obtain this fit conflict with present theoretical 1(a) the same relative free-energy shift would
have a much smaller effect. We would therefore
Figure 2 shows the entropy as a function of expect that the second-order paramagnetic-naf
temperature at various magnetic fields, which transition would move down to the region of the
should be compared to Fig. 2 of Kummer et al. other transitions in Fig. 1 so that the naf region
Note that along the upper "phase" line of Fig. 1 to the right of the dashed line and to the right of
we do not have a discontinuity in S as a function the seaf phase might disappear.
of T but, in exact accord with Kumrner et al. , In conclusion we have shown that there is no
we find a rapid entropy change which becomes problem fitting almost all available solid-3He
smoother at higher field. The physical reason magnetic data with two-, three-, and four-atom
for such a rapid entropy change without an actual exchanges. Also we have found an amazing rich-
phase transition became apparent when the pa- ness of possible behavior. The possibilities when
rameters were adjusted slightly away from a the full B-T-P space is explored are significant
good fit to the high-T data and a phase diagram because we find the phase diagram to be sensitive
like that in Fig. 1(b) was obtained. In this case to parameter shifts and we imagine that the ra-
we find an actual first-order transition which tios of the presumed physical exchanges should
ends in a critical point. Examination of the naf change under pressure.
phase in the case in Fig. 1(a) "under" the scaf Finally we would urge direct measurement of
1443
VOLUME ) 5, NUMBER 21 PHYSICAL REVIEW LETTERS 24 NovEMBER 1975
the spin structure of the various phases if at all S. B. Trickey, W. P. Kirk, and E. D. Adams, Rev.
possible. We note that if our phase diagram is Mod. Phys. 44, 668 (1972).
not correct even spiral" or more complicated R. B. Kummer, E. D. Adams, W. P. Kirk, A. S.
phases are not out of the question. We would al- Greenberg, R. M. Mueller, C. V. Britton, and D. M.
Lee, Phys Rev. Lett. 34, 517 (1975).
so urge a more careful experimental examina-
W. P. Halperin, C. N. Archie, F. B. Rasmussen,
tion of the region near the bottom of the dashed 8, . A. Buhrman, and R. C. Richardson, Phys. Rev. Lett.
phase line to determine if an actual critical point 32, 927 (1974).
exists. A. K. McMahan, J. Low Temp. Phys. 8, 116 (1972}.
1, . Zane, J. Low Temp. Phys. 9, 219 (1972).
l.
A. K. McMahan and J.
W. Wilkins, Phys. Rev. Lett.
M. F. Panczyk and E. D. Adams, Phys. Rev. 187, 35, 376 (1975).
321 (1969}. ' A concise method of derivation is given by W. J. Mul-
M. F. Panczyk and E. D. Adams, Phys. Rev. A 1, lin, Phys. Rev. B. (to be published}.
1356 (1970). G. A. Baker, Jr. , H. E. Gilbert, J. Eve, and G. S.
J. M. Dundon and J. M. Goodkind, Phys. Rev. Lett. Rushbrooke, Phys. Rev. 164, 800 (1967).
32, 1343 (1974). W. P. Kirk and E. D. Adams, Phys. Rev. Lett. 27,
T. P. Bernat and H. D. Cohen, Phys. Rev. A 7, 1709 392 (1971).
(1973) . L. Goldstein, Phys. Rev. A 8, 2160 (1973).
W. P. Kirk, E. B. Osgood, and M. Garber, Phys. 'R. A. Guyer, Phys. Rev. A 9, 1452 (1974).
Rev. Lett. 23, 833 (1969). T. A. Kaplan, Phys. Rev. 116, 888 (1959).
Neutron-Inelastic-Scattering Measurements of Phonon Perturbations
by Defects in Irradiated Copper*
R. M. Nicklow, R. R. Coltman, F. W. Young, Jr.
, and R. F. Wood
Solid State Division, Oak Ridge Nationa/ Labowato~, Oak Ridge, Tennessee 37830
(Received 28 July 1975)
Neutron-scattering measurements have been made at 10 R of phonons in a copper crystal
neutron-irradiated at 4 E. The results, though clearly showing q-dependent defect-phonon
perturbation effects, could not be explained entirely by simplified calculations for the (100)
split interstitial. A 800-K anneal removed the irradiation-induced line broadening and fre-
quency shifts but only - 50"(() of the additional peak structure, which was completely re-
moved only after an 800-K anneal.
Et is well established that the irradiation of and Nicklow. 'The direct observation of such
crystals at very low temperatures produces va- resonant modes would provide new information
cancies and interstitials, and extensive research about the structure of the interstitial, its dynam-
has been focused on determining the stable struc- ical properties, and the interatomic forces which
ture of self-interstitials, particularly in fce couple it to the lattice.
metals. Recent theoretical studies have predict- Here we report the first results of neutron-co-
ed that the (100) dumbbell configuration is the herent-inelastic-scattering experiments on per-
stable form for the interstitial in Cu' and this turbed phonons in a neutron-irradiated crystal.
has been corroborated by diffuse x-ray' and elas- The experiment was carried out on a high-purity,
tic-constant measurements. ' The theoretical low-dislocation-density crystal of dimensions
work predicted that the dumbbell interstitial pos- 0.6xl. 8x3.8 cm', with the [110] direction paral-
sesses low-frequency librational modes. ' Wood lel to the longest dimension. The crystal was ir-
and Mostoller~ and Schober, Tewary, and Dede- radiated by therma/ neutrons at 4.2 K for 24
richs' have shown that these can undergo strong days in the low-temperature facility at Oak Ridge
resonant hybridization with the phonons of the National Laboratory. ' The concentration of
host crystal, which possibly could be detected by Frenkel pairs was =40 ppm randomly dispersed
neutron scattering even for very low concentra- through the crystal. After irradiation the sam-
tions («1/~) as in the somewhat analogous ease ple was transferred at 4.2 K to a liquid-helium
of KC1:CN recently studied by Walton, Mook, cryostat on the triple-axis spectrometer at the
1444
