HAL Id: jpa-00210619
https://hal.archives-ouvertes.fr/jpa-00210619
Submitted on 1 Jan 1987
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Ion channeling in the quasi-one dimensional blue bronzes A0.30MoO 3 (A = K, Rb)
B. Daudin, M. Dubus, J. Dumas, J. Marcus
To cite this version:
B. Daudin, M. Dubus, J. Dumas, J. Marcus. Ion channeling in the quasi-one dimensional blue bronzes A0.30MoO 3 (A = K, Rb). Journal de Physique, 1987, 48 (10), pp.1779-1786.
�10.1051/jphys:0198700480100177900�. �jpa-00210619�
Ion channeling in the quasi-one dimensional blue bronzes A0.30MoO3 (A = K, Rb)
B. Daudin, M. Dubus, J. Dumas (1) and J. Marcus (1)
Service des Basses Températures, C.E.N.G., avenue des Martyrs, BP 85 X, 38041 Grenoble Cedex, France
(Reçu le 27 avril 1987, revis6 le 22 juin 1987, accepté le 25 juin 1987)
Résumé.
2014Nous avons mesuré les rendements de rétrodiffusion de protons de 1 MeV pour les bronzes bleus
quasi unidimensionnels K0,3MoO3 et Rb0,3MoO3 dans la gamme de température 50-350 K. Nous avons mis en
évidence d’importants phénomènes d’hystérésis et des effets dépendant du temps sur de longues périodes.
Nous discutons nos résultats en termes de domaines d’onde de densité de charge et de parois de domaines
couplées aux défauts du réseau.
Abstract.
2014We have measured backscattering yields of 1 MeV protons in the quasi-one dimensional blue bronzes Ka0.30MoO3 and Rb0.30MoO3 at temperatures between 50 and 350 K. Large hysteresis and time dependent effects involving long time scales are found. The measurements are discussed in terms of charge density wave domains and domain walls coupled to lattice defects.
Classification
Physics Abstracts
71.45
-61.80
1. Introduction.
The molybdenum blue bronzes Ao.30MOO3 (A
=K, Rb) undergo a metal-to-semiconductor transition at
180 K [1]. Optical reflectivity measurements by Travaglini and Wachter [2] and resistivity measure-
ments by Perloff et al. [3] showed that these com-
pounds are quasi-one dimensional conductors at room temperature. X-ray diffuse scattering studies by Pouget et al. [4] have established that the transi- tion at 180 K is a Peierls transition towards an
incommensurate charge density wave state which
becomes quasi-commensurate below 100 K. The charge density wave (CDW) state results from a
nesting of the Fermi surface and from an electron-
phonon coupling ; it consists of a periodic modu-
lation of the electron density coupled to a periodic
lattice distortion. Sato et al. [5] and Pouget et al. [6]
have shown evidence of the formation of a Kohn
anomaly responsible for the softening of a phonon
mode at 180 K. In the semiconducting phase, Dumas
et al. [7] have shown that the blue bronze exhibits
non linear voltage-current characteristics beyond a
small threshold electric field. They have attributed
this non linearity to the sliding of the CDW. At low electric field, the CDW is pinned by its interactions with impurities but above a threshold field the CDW
is depinned and contributes to the conductivity [8].
Compelling evidence for the motion of the CDW has been given recently by Segransan et al. [9] by NMR studies of Rbo,3oMo03. A wealth of metastable
phenomena with long time scales both in the pinned
and current carrying state have been reported [10].
The observation of electrical and thermal hys-
teresis [10] dielectric polarization [11], various mem-
ory effects [12] as well as of time dependent 87 Rb NMR [13] and M05
+EPR [14] lineshapes and
metastable loss of transverse order under application
of an electric field [15] indicate that many metastable
states are involved in the pinned state of the CDW
even at temperatures as low as = 4.2 K.
To obtain a more detailed understanding of the
Peierls distorted state we have investigated the blue
bronzes Ko.3oMo03 and Rbo,3oMo03 by proton chan- neling technique in the temperature range 50 K- 350 K. This technique is very sensitive to detect small atomic displacements, of the order of 0.1 A.
The channeled particles are dechanneled due to the effects of thermal vibrations, impurities, lattice im-
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:0198700480100177900
1780
perfections. The study of a CDW transition by this technique has been reported for the first time by Haga et al. [16] then by Nunez-Regueiro et al. [17]
in the case of TiSe2 and 1 T-TaS2. The first attempt
of ion channeling in Kfl.3oMo03 with He+ ions in the
temperature range 100 K-300 K has been reported by Abe et al. [18]. These authors have observed an
enhancement of the minimum backscattering yield,
X min’ in the vicinity of 180 K and have attributed this
anomaly to the softening of a phonon mode.
We report in this paper a more detailed study of
X min and show that this quantity depends strongly on
the thermal history of the sample and is a function of
time. Some preliminary results have been published
elsewhere [19]. Anomalies in the vicinity of 180 K
and 100 K are found on Ko.30MOO3 and Rbo.30Mo03.
Complementary resistivity measurements show that the resistance of the blue bronzes is also strongly hysteretic and time dependent. This non equilibrium
behaviour is discussed in relation with other metasta-
bility phenomena in the blue bronzes. We propose that the observed phenomena result from the pre-
sence of CDW domains and domain walls coupled to
lattice defects.
The blue bronze shows a layered type crystal
structure. The structure is monoclinic, built with clusters of ten distorted Mo06 octahedra. The clusters share corners along the monoclinic b-axis and the [102] direction and form infinite two-dimen- sional sheets. The alkali ions A+ lie between the sheets [20]. The structure, illustrated in figure 1,
contains three Mo sites. The 4d electrons are mainly
located on the Mo(2) and Mo (3 ) sites which are
Fig. 1.
-Crystal structure of the blue bronze showing the
infinite sheets of Mo06 octahedra 11 (201), separated by
the alkaline ions (8). The sheets contain the infinite chains 11 b.
involved in infinite chains of Mo06 octahedra along
the b-axis [21]. There are two alkali sites with large
and anisotropic Debye-Waller factors.
The structure can also be viewed as made of infinite chains of clusters of ten Mo06 octahedra running along the b-axis. X-ray diffuse scattering
studies [4] have shown that precursor diffuse
platelets above 180 K condensed into weak satellite spots. The superstructure reflections are charac- terized by the wave vector q
=10 a *, (1 - qb) b *, 0.5 c with qb = 0.26 at 110 K.
(1- qb) increases strongly with decreasing tempera- ture, is temperature independent below 100 K and extremely close to the commensurate value 0.75 [22].
2. Experimental techniques.
The single crystals used in these investigations were prepared at L.E.P.E.S. by electrolytic reduction of a
A2Mo04-Mo03 (A
=K, Rb) melt at 550 °C. Details of the preparation are given in reference [23]. The crystals appear as platelets with typical size 5 x
2 x 1 mm3 parallel to the (201) cleavage plane with
the b-axis as the long direction as illustrated in
figure 2a.
Fig. 2a.
-Schematic view of the samples in the two configurations used : i) beam perpendicular to the cleav-
age plane ; ii) beam parallel to the cleavage plane.
Fig. 2b.
-Typical energy spectrum of backscattered
protons. r : random ; c : channeled. The oxygen front is
shown and the energy window used is visualized : (1).
The ion channeling measurements were carried out using the 2 MeV Van de Graaf ion accelerator of the SBT/L ACC.
As the blue bronzes are highly sensitive to irradia-
tion effects [24], considerable care was taken to minimize defect production by the ion beam. In particular, preliminary experiments with He+ ions
indicated noticeable permanent radiation damage of
the samples.
All the experiments reported in this paper were therefore performed using 1 MeV protons as it was checked that they produced no significant damage.
On the other hand, the penetration depth of 1 MeV
protons in these material is as large as 10.5 um
whereas the energy window shown in figure 2b corresponds to an analysed depth of about 0.8 lim.
As a consequence, the number ,of defects produced by irradiation in the analysed region of the sample
was negligible. The beam was approximately 1 mm
in diameter and the intensity never exceeded 3 x 10- 9 A. The samples were cleaved prior to experiments and fixed to the sample holder with
silver paint. A permanent helium flow was used to cool the sample holder so that the measurement
temperature range was 50 K-350 K. The heating and colling rate were approximately 1 K/minute.
The backscattered particles were counted using a
multichannel analyser and a typical spectrum is shown in figure 2b where the energy window used for the experiments is visualized. The minimum
backscattering yield, X miD’ is defined as the ratio
NCIN, where Nc and Nr are the number of particle
backscattered in, respectively, the channeled and
random geometry.
As far as Ko,3Mo03 is concerned, two geometries
were used (see Fig. 2a) :
i) In most experiments, the sample was oriented
so that the beam direction was along a channel perpendicular to the cleavage plane and to the b-
axis.
ii) Additional experiments were also performed
with the beam axis parallel to the cleavage plane of
the platelets and perpendicular to the b-axis. In this
last case, the samples were about 1.5 mm thick and
the beam diameter was reduced to 0.5 mm.
The experiments on Rbo.3Mo03 were performed using only the geometry defined in i).
The usual experimental procedure was to orientate
the sample at room temperature and to measure
X min as a function of temperature down to 50 K.
Then, the alignement was controlled in order to make sure that the sample had not moved during the cooling process. X min was subsequently measured
from 50 K to 350 K and a new alignement control
was done at high temperature.
Resistivity measurements were made on freshly
cleaved samples with the standard four probes
configuration. Gold wires were attached with silver
paint on indium evaporated areas.
3. Results.
3.1 THERMAL CYCLING EXPERIMENTS.
-In a first set of experiments, X min was measured as a function of temperature in the 50 K-350 K range and several
cycles were completed. The results for two of them
are plotted in figure 3a and 3b and correspond to
Fig. 3a.
-Minimum backscattering yield as a function of temperature.
Fig. 3b.
-Minimum backscattering yield as a function of temperature,. The dotted line is only an eye guide (see
text). The transition temperature, Tc, is indicated.
two different impact point on the same sample. The
characteristic features exhibited are :
a) A smooth variation of X min for decreasing
temperature and no sharp anomaly around Tc = 180 K.
b) A large effect for increasing temperature
which corresponds to an increase of X min around
100 K and a slow recovering achieved for tempera-
ture between 300 and 350 K, well above the transi- tion temperature. It is worth noting that the highest
value of X min obtained during cycle b (see Fig. 3b) is
close to the random value and therefore indicates that a very strong disorder is present in the sample.
For the experiments reported in figure 3a, the
sample was aligned at 50 K and a new impact point
1782
was chosen before increasing the temperature. As a consequence, the large increase in X min which was
observed could definitely not be attributed to poss- ible irradiation defects.
c) For the decreasing temperature part of the cycle shown in figure 3b, the experimental points are
fitted by a straight line between 200 and 300 K but a
noticeable deviation from this line is observed for T 220 K, in a range where premonitory effects of
the transition are known to take place [4].
This deviation which was not always present was nevertheless observed during various experiments,
for decreasing temperature, and we attribute it ot the establishment of long-range order in the sample.
3.2 HYSTERESIS EFFECTS.
-The results plotted in figure 4 correspond to various thermal cycles below
and above Tc. The most striking result is that the change of slope observed on the heating curve
around 100 K when the minimum temperature
reached was 50 K (see Fig. 3a and 3b) can be shifted
to higher temperature. It is actually clear in figure 4
Fig. 4.
-Hysteresis in the minimum backscattering yield
versus temperature curves for Ko.3Mo03 after thermal
cycling. Closed symbols correspond to increasing tempera-
ture (1, 3, 5). Open symbols correspond to decreasing temperature (2, 4).
that the temperature associated to this change of slope is strongly correlated to the maximum tempera-
ture reached during the cycle just before and that the
system has a memory of the previous stages. This
memory effect has some analogy with that found by
Mutka et al. (12) in their resistivity measurements.
It is also clear that this effect is present for temperatures larger than Tc and the evolution of
X min plotted in figure 5 as a function of time is an
evidence for metastability.
Fig. 5.
-Time-dependance of the minimum backscatter-
ing yield at T
=200 K (open lozenges), taken on a heating cycle. The closed triangle is the value measured at room
temperature on the virgin sample (i.e. before any thermal
cycle).
3.3 IRREVERSIBILITY EFFECTS.
-In figure 6 the
result for thermal cycling between 350 K and 190 K is shown. An hysteresis is present but, in contrast
with figure 5, the stabilization which was done at 190 K before heating the sample again was uneffec- tive, as shown in the inset of figure 6.
Fig. 6.
-Hysteresis in the minimum backscattering yield
versus temperature curve above Tc. Inset : Time-depend-
ance of X min at 190 K, between cycle 1 and 2.
In a following step, the time dependance of
X min was investigated below Tc (see Fig. 7). As
shown in the inset of figure 7, X min was measured as a function of time at T
=164 K. It was found that the time scale was very large and the stabilization could not be achieved. The sample was then cooled
again prior to a further heating up to 350 K. It appears in figure 7 that the maximum of X min around T
=220 K was therefore larger than in most ex- periments, indicating that the memory of the stabili- zation at 164 K was kept by the sample. In addition,
a strong annealing of X min is observed between 250 and 350 K which is consistent with the increase of the disorder observed during the stabilization process at 164 K.
Fig. 7.
-Irreversibility in the minimum backscattering yield versus temperature curve. Inset : Time-dependance
of X min at 164 K corresponding to the part 2 of the curve.
During the heating (curve 4) steps of Xmin seem to
occur. This effect was observed on other samples but
the results are not reported here.
In order to clarify the time dependant effects,
another tefnperature cycle was done using different
Fig. 8.
-Minimum backscattering yield as a function of temperature. Cooling and heating rate were 20 K/hour.
experimental conditions : the temperature was var- ied step by step as usual but a 20 min stabilization
was achieved for each experimental point. The
results are plotted in figure 8 and an increase of
X min is clearly shown upon heating. This effect is not
so large as in figure 3 but is nevertheless more
important than for most experiments (not reported here) ; This is consistent with the time dependant
effects displayed in figure 7.
3.4 BEAM ALIGNED PARALLEL TO THE CLEAVAGE PLANE.
-An additional experiment was done using
the geometry described in section 2.ii) (see Fig. 2a)
as it was assumed that the channeling would be
a priori better in this direction due to the large interplanar spacing. Actually, Xmin was found to be
comparable to previous values and no significant improvement was found. The results are plotted in figure 9 and a significant increase of Xmin is observed but, surprisingly, for decreasing temperature. Fur- thermore, this effect was present only for the first
cycle and the following cycles have reavealed (see Fig. 9) a behaviour similar to most experiments performed whith the previous geometry.
Fig. 9.
-Minimum backscattering yield as a function of temperature. The channeling direction was parallel to the cleavage plane (see text). Curve 1 (closed dots) corres- ponds to the first cooling of the sample. Curve 2 and 3
were measured next.
3.5 SUMMARY OF EXPERIMENTAL DATA. - As far
as the channeling yield is concerned, the exper- imental features can be summarized in the following
way :
-
strong time dependant effects were observed
above and below Tc ;
-
hysteresis effects were observed above and below T, ;
-
all these effects were reproducible but strongly
sample-dependant. One should note that it is also
the case for transport properties [10] ;
1784
-
an evolution was observed towards a state
where, after several cyclings, no strong anomalies
were present. For most
«virgin
»samples, no large
anomalies were present and we stress again that the sample quality is very important as far as these experiments are concerned. (The « virgin » state
refers to a sample prior to any thermal cycling, with
no consideration, a priori, of the defects concen-
tration. )
-
similar experiments performed of Rbo.3MoO3
revealed no significant differences with respect to
Ko.3Mo03.
4. Resistivity measurements.
It has been previously reported that the transport
properties of Ko.3oMo03 and Rbo,3Mo03 are entirely
similar [10].
Figure 10 shows the thermal hysteresis of the
Ohmic resistance of a Rbo.3oMo03 sample in the temperature range 90 K-110 K. Hysteresis always
appears upon thermal cycling whatever the explored
temperature interval. These effects are more striking
in electron irradiated samples, as demonstrated by
Mutka et al. [12].
Fig. 10.
-Hysteresis in the resistance vs. temperature
curves for a Rbo.30MoO 3 sample after thermal cyclings.
Cooling or heating rate : dT/dt
=2 K/min.
Figure 11 shows the variation of the resistance after a thermal cycling and relaxation at two different temperatures. The resistance is always larger on heating than on cooling. On the cooling curve, the resistance increases as a function of time at a given temperature while on the heating curve, it decreases with time. Similar results have been observed by
several authors in the blue bronzes and in TaS3 [25].
Non-Debye character of the relaxation was observed
by L. Mihaly et al. [11].
After a thermal cycling the resistance measured at
190 K is still weakly time dependent. The relative change OR/R is less than one per cent over a period
of one hour. More pronounced effects were observed
on different samples [26]. For T > 190 K we find no
time dependence of the resistance.
Fig. 11.
-Hysteresis in the resistance vs. temperature
curve for a Ko.30M003 sample after relaxation at 109 K (a)
for 30 min then at 117 K (b) for 30 min ; dT/dt
=2 K/min.
5. Discussion.
It is well accepted that the metastability phenomena
in incommensurate CDW systems are related to the random distribution of the pinning strengths of the
CDW by impurities or other point lattice defects.
The disorder due to random pinning leads to
«glass
like » characteristics of the CDW state which are
particularly striking in pure, W-doped or electron
irradiated blue bronzes [10, 12]. The CDW distorted
state can be understood only in taking into account
the large number of internal degrees of freedom of
the CDW condensate. Only deformable models can account for hysteresis and various time dependent
and memory effects. In this context, the existence of CDW domains and CDW domain walls has been
proposed by several authors [27]. Due to the pinning
forces the CDW condensate is always in a stressed
state. It can be viewed as an
«electronic solid » [28]
which possesses its own defects [29].
We discuss our results in terms of CDW defects
interacting with lattice defects, first the results in the
temperature range 180 K-350 K, then those in the
temperature range 50 K-180 K.
5.1 TEMPERATURE INTERVAL 180 K T 350 K.
-
In the explored temperature range and using
different cooling rates most surprising is the absence of marked anomaly in X min during cooling from a virgin state. The observation of an anomaly in the Young’s modulus [30] in the vicinity of the Peierls transition temperature in common with other CDW materials as well as anomalies in the lattice parame- ters [21] would suggest a possible change in Xmin-
However, when the sample is heated from 50 K,
X min becomes strongly temperature dependent and
its values are much larger than those observed on
cooling especially in the vicinity of 200 K. An increase of xm;n always corresponds to an increase of
disorder. The hysteresis disappears only at ~ 350 K
and the sample returns to its virgin state. The time dependence of X min is found only on the heating
curve. For T ~ 200 K X min decreases slowly with
time. This clearly indicates that the system has been driven far from equilibrium. If X min was only due to
the softening of a phonon mode one would not
expect hysteresis and time dependent effects. On the other hand the temperature dependence of X m;n does
not follow that of the order parameter [22] unlike in TiSe2 [17]. Therefore we rather believe that Xmin is dominated by the presence of mobile defects. Since the anomaly of x min is rather large in view of the small ion displacements associated with the Peierls
transition, the defects involved might be incommen-
surate CDW domains slowly rearranging and inside
which the wavevector of the modulation is pinned to
out of equilibrium values. This hypothesis is suppor- ted by the fact that reasonable point defects concen-
trations cannot account for such a large increase of X min’
No firm conclusion on the possible shape of CDW
domains can be given, even with the two configura-
tions employed. In particular, the possible existence
of domains elongated parallel or normal to be b-axis
cannot be deduced from the above data, although it
is desirable to elucidate this point for a better understanding of CDW transport.
This domain structure acquired at low tempera- ture, would persist above 180 K. The hysteresis
above - 180 K may have the same origin as the
remanent extra 87 Rb NMR line [13]. This latter effect has been discussed in the framework of the defect density wave concept (DDM) developed by
Lederer et al. [31]. In this model, a small fraction of
the lattice defects order periodically and are coupled
to the CDW modulation.
A weak electrical hysteresis in the Ohmic resist-
ance also persist above 180 K, AR/R decreases by
-
1 % over one hour at 190 K. These results seem to be also in agreement with the DDW model. In
addition, this picture seems consistent with the observations in 1 T-TaS2 and in (TaSe4)2I near the
Peierls transition of an acoustic noise emission [32]
which corresponds to a release of elastic energy.
5.2 TEMPERATURE INTERVAL 50 K T 180 K.
-