Supersolidity, disorder and grain boundaries S. Sasaki, R. Ishiguro, F. Caupin, H.J. Maris and S. Balibar Laboratoire de Physique Statistique (ENS-Paris) now at North Western Univ. (USA) now at North Western Univ. (USA) now at Tokyo University (Japan) Brown University, Providence (RI, USA) de la Physique, Saclay, 23 oct S. Sasaki et al., Science 313, 1098 (2006) - S. Sasaki, F. Caupin and S. Balibar, Phys. Rev. Lett. 99, (2007) - S. Balibar and F. Caupin « topical review » J. Phys. Cond. Mat. 20, (2008) - S. Sasaki, F. Caupin and S. Balibar, to appear in J. Low Temp. Phys. (Nov. 2008)
a « supersolid » is a solid which is also superfluid solid : transverse elasticity, i.e. non-zero shear modulus a consequence of the localization of atoms crystals - glasses superfluid : a quantum fluid with zero viscosity a property of interacting Bose particles, which are indistinguishable and delocalized a paradoxical idea
could solid helium 4 flow like a superfluid ? E. Kim and M. Chan (Penn. State U. 2004): a « torsional oscillator » (~1 kHz) a change in the resonance period below ~100 mK 1 % of the solid mass decouples from the oscillating walls ? no effect in helium 3 (fermions) rigid axis ( Be-Cu) solid He in a box excitation detection Temperature (K) superfluid fraction (NCRIF)
early theoretical ideas Andreev and Lifshitz 1969: delocalized vacancies could exist at T = 0 ( the crystal would be « incommensurate ») BEC => superplasticity at low velocity and long times coexistence of non-zero shear modulus and mass superflow ! E0E0E0E0 zh zh Among others, off diagonal order (Penrose et Onsager 1956): no supersolidity symmetrization and overlap of wave functions (Reatto 1969, Chester 1969, Leggett 1970, Imry et Schwartz 1975), the rotation of a quantum solid (Leggett 1970)...
more recent ideas (a selection...) Prokofev, Svistunov, Boninsegni, Pollet, Troyer (MC calc.): no supersolidity without free vacancies ; the probability for a real quantum crystal to be commensurate and supersolid is zero 4 He crystals are commensurate (E vac = 13K) => not supersolid BUT BEC is possible in a 4 He glass (Boninsegni et al. PRL 2006, see also AF Andreev JETP Lett. 2007) Galli and Reatto 2006 : variational calculation with particular trial wave functions (« SWF ») which describe the properties of solid 4 He => supersolidity in a commensurate crystal ! Clark and Ceperley (2006) : supersolidity depends on trial wave functions no supersolidity in « exact » calculations ( quantum path Integral Monte Carlo); crystals are commensurate, no vacancies at T = 0, no supersolidity in perfect crystals other models of quantum crystals : Cazorla and Boronat (PRB 2006) Josserand, Rica and Pomeau (PRL 2007) supersolidity but are these models realistic ?
other theoretical ideas Anderson Brinkman and Huse (Science 2005): new analysis of the lattice parameter a/a (T) and the specific heat C v (T): a small density of zero-point vacancies (< ?); T BEC ~ a few mK ; s ? not confirmed by neutron scattering (Blackburn et al. PRB 2007) critics by H.J. Maris and S. Balibar (J. Low Temp. Phys. 147, 539, 2007) two transitions ? a vortex liquid ? could supersolidity be due to the presence of defects in crystals ? Dash and Wetlaufer 2005: the He-wall interface could be superfluid PG de Gennes (Comptes Rendus - Physique 7, 561, 2006): the mobility of dislocations could depend on T; frequency dependence ? (see Kojima et al. 2007) Pollet et al. (Monte Carlo): grain boundaries should be supersolid Boninsegni et al. 2007: dislocation cores should be supersolid Biroli and Bouchaud (2008): dislocations fluctuate and favor atom exchange
the role of disorder : annealing, quench-cooling, grain boundaries Rittner and Reppy (Cornell, ): supersolidity disappears after annealing s up to 20 % if the samples are grown by quench-cooling the liquid from its normal state Sasaki, Ishiguro, Caupin, Maris et Balibar (ENS, Science 2006) : a dc-flow experiment with solid helium 4 superfluid mass transport in the presence of grain boundaries
the role of disorder S. Sasaki et al. Science 313, 1098, 2006 a glass tube (1 cm ) cristallization from superfluid liquid at 1.3 K cool down to 50 mK a height difference between inside and outside any relaxation ? hélium solide hélium liquide fenêtre any variation of the inside level requires a mass current through the solid because C = 1.1 L
the physics of grain boundaries solid liquid liquid solid liquid crystal 1 crystal 2 grain boundary liquid phase GB LS mechanical equilibrium of the liquid-solid interface : GB LS cos each groove signals the existence of an emerging grain boundary dynamics of grain boundaries : large mobility + pinning on walls
without grain boundaries, no flow with grain boundaries, superfluid flow time accelerated x 250 supersolidity is not an intrinsic property of the crystalline state of helium 4, a property associated with its defects annealing - quench (Rittner and Reppy )
cristal 2: relaxation at50 mK relaxation is linear, not exponential a superfluid flow at its critical velocity 2 successive regimes: 6 m/s for 0 < t < 500 s 11 m/s for 500 < t < 1000 s more defects in the lower part of crystal 2
crystal 1 : only one grain boundary relaxation at relaxation at V = 0.6 m/s stops when the grain boundary unpins and suddenly disappears If 1 grain boundary only with a superfluid thickness e ~ 0.3 nm, width w ~ 1cm the critical velocity inside should be: v c GB = ( D 2 /4ew s )( C - L )V = 1.5 (a/e)(D/w)( C / s ) m/s comparable with 2 m/s measured by Telschow et al. (1974) for liquid films of atomic thickness
Pollet et al. PRL 98, , 2007 grain boundaries : ~ 3 atomic layers superfluid except in special directions T c ~ 0.2 à 1 K depending on orientation critical velocity ?
a 1% superfluid density is large! (Rittner and Reppy 2007: 20% in quenched samples !) s = 1% => grain size ~ 100 nm s = 0.03% => 3 m is it possible ? may be how does the grain size depend on the growth method ? constant P or contant V from the normal liquid from the superfluid slow or fast cell geometry (T gradients)
the ENS high pressure cell 2 cells : 11 x 11 x 10 mm 3 or 11 x 11 x 3 mm 3 thermal contact via copper walls thickness 10 mm 2 glass windows (4 mm thickness) indium rings => leak tight stands 65 bar at 300K pressure gauge (0 to 37 bar)
at T < 100 mK from the superfluid, fast growth and melting of a high quality single crystal real time, 60 mK 11 mm
fast growth from the normal liquid at T > 1.8 K dendrites fast pressurization of the normal liquid => dendritic growth similar to Rittner et Reppy’s quenched samples ? T = 1.87 K 11 mm
helium snow flakes T = 2.58 K is the roughening transition reentrant ? facets at T < 1.3 K and at T > 2 K ? Balibar, Alles and Parshin, Rev. Mod. Phys remember Burton, Cabrera and Frank
slow growth from the normal liquid at constant V growth in ~ 3 hours a temperature gradient : T walls < T center the solid is transparent but polycrystalline
T = 0.04 K melting a crystal after growth at constant V T = 0.04 K liquid channels appear where each grain boundary meets the glass windows grains < 10 m ripening
T < 0.1 K ripening at the liquid-solid equilibrium (another example) T < 0.1 K minimization of surface energy mass transport through the superfluid latent heat L = 0
2 crystals + 1 grain boundary the groove angle is non-zero => the grain boundary energy GB is strictly the grain boundary energy GB is strictly < 2 LS => microscopic thickness, in agreement with Pollet et al. (2007). a complete wetting would imply GB LS (2 liq-sol interfaces with liquid in between) thickness 10 mm thickness 3 mm angle 2 the contact line of grain boundaries with each window is a liquid channel
angle measurement => grain boundary energy align the optical axis fit with Laplace’ equation near the cusp = 14.5 ± 4 ° = 14.5 ± 4 ° GB = (1.93 ± 0.04) LS other crystals : = 11 ± 3 ° = 11 ± 3 ° = 16 ± 3 ° angle 2
near a wall : wetting of grain boundaries The wall is favorable to the liquid. If GB is large enough, more precisely if + c < /2 the liquid phase wets the contact line. an important problem in materials science (see JG Dash Rep. Prog. Phys. 58, 115, 1995) prediction: liquid channels also where grain boundaries meet each other => between 25 and 35 bars a polycrystal contains many liquid channels grain 1 grain 2 liquid wall S. Sasaki, F. Caupin, and S. Balibar, PRL 99, (2007) wall grain 1 grain 2 GB largeur w ; épaisseur e invers t prop. à la profondeur z (w ~ 20 m à z = 1 cm) l c : longueur capillaire the width w and the thickness e are inversely proportional to the depth z l c : capillary length
between 3 grains stable channels if Miller and Chadwick Acta metall. ’67 Raj Acta metall. mater. ’90
grain boundaries in classical crystals Besold and Mouritsen ’94 Monte-Carlo simulation colloidal crystals Alsayed et al. Science ’05 T m =28.3 C Alsayed et al. Science ’05 T m =28.3 C 5 µm wetting of boundaries by the liquid phase ?
short and long range L S S t Schick and Shih PRB ’87 short range forces: complete wetting the thickness t diverges at the liquid-solid equilibrium long range forces : incomplete wetting, microscopic thickness consequences : impurity diffusion (ice), mechanical rigidity
the channel width w ~ (P-P m ) -1 the channel width w decreases as 1/ z (the inverse of the departure from the liquid-solid equilibrium pressure P m ) agreement with new measurements of the contact angle c liquid channels should disappear around P m + 10 bar (where 2w ~1 nm)
hysteresis of the contact angle advancing angle : 22 ± 6 ° (copper) 26 ± 7 ° (glass) 37 ± 6 ° (graphite) receeding angle : 55 ± 6 ° (copper) 51 ± 5 ° (glass) 53 ± 9° (graphite) perhaps more hysteresis on copper rough walls than on glass or graphite walls whose roughness is smaller. as expected from E. Rolley and C. Guthmann (ENS-Paris) PRL 98, (2007) meltinggrowing copper copper
a stacking fault : low energy, no liquid channel growth shape between 2 crystals with same orientation GB smaller larger + c > /2 no liquid channel equilibrium shape see also H. Junes et al. (Helsinki) JLTP 2008 : 2 = 155 ± 5°
2 possible interpretations of the experiment by Sasaki et al. (Science 2006) mass transport either - along the grain boundaries (then v c ~ 1 m/s) - or along the liquid channels (then v c ~ 3 mm/s). This would explain why flow was observed also at 1.13 K - future experiments : change the cell geometry, reduce the width of liquid channels with an electric field study h(t) with more accuracy with 1 fixed grain boundary liquid liquid solid solid
are grain boundaries (together with their liquid channels) responsible for everything ? Chan et al. (sept. 2007): anomalies in single crystals grown at constant T and P NO!
a superfluid network of dislocations ? along one dislocation, a coherence length l ~ a (T*/T) percolation of quantum coherence in a 3D network ? May be, but T c << 100 mK... furthermore, helium 3 impurities increase T c ! Boninsegni et al. PRL 2007
shear modulus oscillator period Day and Beamish (Nature 2007): the shear modulus increases ! the shear modulus increases by ~ 15 % below 100 mK (depending on He3 content!) pinning of dislocations by 3He impurities ? and grain boundaries (samples grown at constant V) ? another explanation for torsional oscialltor measurements : K increases ? similar anomalies
my opinion in June experiments: 2 main interpretations are possible 1 - superfluidity of some fraction of the mass, associated with defects in a mysterious way (I/K) 1/2 decreases because the inertia I decreases (I/K) 1/2 decreases because the inertia I decreases 2 - no superfluidity a change in the dynamics of defects (dislocations ? grain boundaries ?) the elastic constant K increases theory: a quasi-general consensus no supersolidity in a perfect crystal (without defects) dislocations, grain boundaries and glassy helium (if it exists...) should be superfluid at low T
new experiments by Rittner and Reppy 2008 no period change in a blocked annulus there is a macroscopic mass current in a free annulus
Kim and Chan 2008: influence of 3 He impurities the dependence on 3He concentration is consistent with a simple model of adsorption on dislocations
new experiments by Beamish et al He3 hcp and He4 hcp : same elastic anomalies He3 bcc: no elastic anomalies He4 bcc: supersolidity ? supersolidity is clearly linked to quantum statistics
... and my opinion at the end of August 2008 Supersolidity appears when 3He adsorbs on defects ! WHY ?? in liquid 4 He, 3 He acts against superfluidity the existence of supersolidity in solid helium 4 is confirmed it is associated with the existence of defects but also to the presence of 3He impurities it is not a purely elastic effect An idea by B. Svistunov (Trieste, 22 August 08): by condensing at the nodes of a dislocation network, the 3He could provide a connection between the dislocations of the 4He crystal... provide a connection between the dislocations of the 4He crystal...
perspectives Measure the rotational inertia and the shear modulus in samples with well characterized disorder : single crystals (oriented ? measure the dislocation density ?), polycrystals, glassy samples... optics, X - rays, neutrons, thermal conductivity... search for superfluidity inside grain boundaries : flow experiments in a cell with 1 stable grain boundary and reduce the effect of liquid channels (electrostriction) measure the quantum dynamics of defects theory of quantum defects (quantum metallurgy)