Sources of loss processes in phonon generation and detection experiments with superconducting tunneling junctions

Abstract

Observing the phonon yield, i.e. the ratio of the experimental phonon signal amplitude and the corresponding calculated value, phonon losses within the generation-detection system can be localized and determined quantitatively. With tin junctions on pure silicon substrates immersed in liquid helium the phonon yield is 3-5%. Under vacuum conditions the yield rises to 10-12 % indicating strong phonon transmission to the helium bath. The experimental lifetime for 280 GHz phonons in the silicon substrate is longer than 65 las indicating negligible volume losses and losses at the free substrate surface. It is further shown, that volume losses inside the phonon generator and detector are small compared to the total loss of about 90 %. By phonon reverberation measurements we find evidence that the main sources for phonon losses are localized at the boundaries of the tunneling junctions to the substrate. This is supported by an increase of the phonon yield with improved polishing from about 9 % (mechanical), 10% (chemical) to 12 % (sputter etching). A SIMS analysis indicates the presence of carbonhydrates and probably of water in the boundaries. This layer of extraneous molecules together with the nonideal surface structure of the substrate and the evaporated films weakens the mechanical bonding between the tunnel junctions and the substrate and is possibly causing strong phonon splitting by anharmonic forces.

Analysis of the phonon spectrum emitted by superconducting Al-tunneling junctions at high quasiparticle injection rates

“…This last assumption (7"~tle___0.2) is confirmed by the good agreement between experimental and theoretical values (diodes la, lb, 2a, 3a). The deviations in the case of diodes 8a, 8b, 9b can be explained by the possibility of an enhanced surface transmission from the metal film into the He-bath due to adsorbed impurities [28]. On the other hand, only in the i-U characteristics of diodes with pure Al-films a clear gap reduction could be observed at eU=2A (Fig.…”

“…In applying this to a superconducting film having a thickness d>>Ao (~=l,t), the well-known result is yielded [6], that the relative values of the phonon fluxes of different polarizations against the boundaries of the generator are independent from the interaction parameters gig, or At/A t : FGl/Fat=gl~a/(gt~a), ~F~= ~F~Gs+ T~GH~ (3) i.e., the phonon fluxes are determined by the internal interaction strength as well as be the phonon transmissions 5F~o s and ~['aGHe of the two boundaries of the generator to the substrate and to the helium bath, which are averaged over all angles of incidence [18,19]. In time-resolved measurements, the intensity of longitudinal and transverse phonon signals may be determined by time of flight separation of the polarizations [28]. In working with Al-detectors at low temperatures, this method becomes difficult, as a consequence of the relatively long detector time constants (5-30bts).…”

“…If a tunneling detector is irradiated by the recombination phonons of a tunneling generator (A o < A~) and the frequency of the recombination phonons does not exceed 4Ao, then the theory predicts a signal amplitude of [28]:…”

“…In working with Al-detectors at low temperatures, this method becomes difficult, as a consequence of the relatively long detector time constants (5-30bts). Nevertheless, the detector signal should markedly depend on the phonon polarization, since the phonon rate ri D entering the detector is related to FGo by [28] …”

“…In addition, it is known, that phonons decay at the metal -sub- strate boundary to a significant percentage up to 65%. The decay is different for longitudinal and transverse modes and depends on the method of substrate surface preparation [28,34]. …”