a continuity for the T 2 TK behaviour from the dilute limit through
concentrated alloys to periodic heavy fermion systems [316]. Does this mean that the argument yielding (11.46) was flimsy, and CK is a false concept? That would be unjust. Careful investigation confirms that the large Kondo cloud is a reality [372]. However, the situation is quite complicated. At finite T,the behaviour has to be described by the simultaneous use of two length scales: &, and the thermal length scale &. Furthermore, one has to be careful in interpreting measurements by local probes like NMR, because they are sensitive to the details of the internal structure of the Kondo singlet. Part of the story is that the Kondo exchange polarizes the electron sea, and the resulting f-c spin correlations have an oscillating cx cos k F r / r 2 character. A little later, we will see that this gives rise to the RKKY interaction. The formation of the Kondo cloud imposes a T-dependent envelope upon the oscillating correlations; at low T, the remaining length scale is (K. We refer to [372] and [45] for details. The genuinity of the large Kondo cloud, juxtaposed with the survival of Kondo features in concentrated alloys, is paradoxical. It suggests that there should exist a theoretical framework in which typical large-T manifestations of the Kondo effect follow from single-impurity Kondo compensation for cimp< (a/<K)3, and they are not only not destroyed, but conceivably even amplified by impurity-impurity interactions for Cimp > ( a / < ~ )The ~ . behaviour is governed by a single energy scale TK (which should not be much different from (11.35)), and scales with Cjmp. The results of resistivity measurements on CezLal-zCu6, for which a 'T of 3.5-5K was suggested [316], may serve as illustration (Fig. 11.14). The behaviour is unified and smooth37 for T > 10K. However, substan- tial differences are seen at low T. Low-s samples follow by and large the independent-impurity Kondo behaviour: the resistivity increases mono- tonically with lowering T, saturating at T = 0 as described by (11.45). On the other hand, concentrated samples, and most clearly pure CeCu6, show a resistivity maximum, and after a steep decrease of p, follow the + Landau regime law p M po A*T2. In these systems, we may speak about the emergence of low temperature phase coherence between the 37Thehumpy character of some Iowa curves is ascribed to crystal field excitations which we cannot discuss here.