The thermochromic properties of thin films of a mixture of a fluoran dye (FD: 2'-(o-chloroanilino)-6' -dibutyl-aminospiro[isobenzofuran-1(3H), 9'-[9H] xanthene]-3-one) and long-chain alkyl-phosphonic acid (Pn:n=carbon number of alkyl chain) were reviewed. This mixture has attracted considerable attention in recent days, for the mixure gives rise to the functions of coloring rewritable recording devices, which are sensitively dependent on the crystallization behavior of the mixture systems. Coloring states are attained in two phases; mixture liquid as well as mixture solid. The solidified coloring state was revealed at room tempemture by quenching the mixture liquid. The slow cooling, however, resulted in the discolored state. Upon heating from the solidified colored state, the discoloring initiates in a temperature range about 20℃ below the melting temperature of the phosphonic acid component. By repeating the processes of quenching, heating and slow crystallization, the rewritable function can be realized. To clarify the mechanism involved in this process, we precisely examined structural properties of the mixture solids colored and discolored states, and crystallization behavior of the discolored solid mixtures. It was confirmed that the colored sold mixture was in a lamella-type "supramolecules" conisting of FD and Pn, in which the carbon number n was larger than 14. By contrast, the discolored state is caused by fractional crystallization of phosphonic acids, decomposing the "sipramolecules" by either or slow crystalization.

View full abstract

The influence of the crucible rotation on the temperature distribution at the melt surface in Czochralski silicon growth has been studied by new CCD camera observation system. The thermal radiation energy from the melt surface is converted to temperature by blackbody calibration method and is recorded on video tapes as two-dimensional color images. The experimental results without a crystal showed the temperature distribution changes at the melt surface in three patterns: axisymmetric, n-folded (n≧2), and cellular patterns in order when the crucible rotation rate is increased. From three- dimensional time dependent calculations of the silicon melt flow in the down-scaled Czochralski configuration, the n-folded temperature distribution is qualitatively explained to be caused by the change of the melt flow due to baroclinic instability. The cellular temperature pattern at relative high crucible rotation rates is considered originating from Benard-Rayleigh and/or Marangoni-Benard instability.

View full abstract

Temperature fluctuations of silicon melt in the crucible were measured to evaluate the flow structure during the Czochralski silicon growth process. The impurities which influence the density anomaly of the silicon melt were doped as experimental parameters. The frequency components of these fluctuations were investigated. The flow structure under the growth interface region is found to be basically a random one that is supposed to represent the "soft urbulence" induced by Rayleigh-Benard convection. The addition of gallium which is known to suppress the density anomaly to the melt changes the flow structure to a laminar-like one when the melt depth is shallow, while the flow structure is still similar to a soft turbulence in case with boron-doped at the same melt depth. The addition of gallium also makes the temperature wave patterns considered to be caused by baroclinic instability stable below the free surface and the periphery of crystal, resulting in sharp temperature fluctuations. In cases of undoped and boron-doped melt, temperature fluctuations in these regions become gentle and the frequency components of temperature waves described above weaken considerably. The results of wavelet analysis indicate the existence of a stable wave in a comparatively long period position in such a situation. The difference between these two conditions suggests that the flow regime changes from the azimuthal periodic structure to a complex flow. The density anomaly of silicon melt in a CZ silicon growth process is supposed to exist and have a strongly influence on the melt flow structure in the crucible.

View full abstract

Previous studies on the measurements of surface tension of molten silicon containing our recent results were critically reviewed as follows: 1) brief discussions based on the Keene's review on the reported literature values of surface tension σ and its temperature coefficient dσ/dT mainly in relation to the effect of impurities, especially oxygen, by the contamination during the measurements of σ and dσ/dT, 2) short introduction and comments on the various methods for the surface tension measurement reported after the publication of Keene's review, 3) thermodynamic feature of the molten silicon-gas system concerning the relation between partial pressure P_<o2> of oxygen and SiO, and also the relation between oxygen concentration in the silicon and P_<o2>, etc., based on our results, 4) treatment of oxygen in the system for the surface tension measurements, for examples, the method and apparatus for controlling P_<o2> of the system and the quantitative influence of oxygen on σ and dσ/dT, which were obtained by our study.

View full abstract

The dependence of the effective segregation coefficient (K_e) of phosphorous on the growth rate (R) has been investigated by the Necking Method in the silicon Cz process for <100> and <111> growth orientations. The k_e in the relatively lower R region is explained by the BPS theory, however, the k_e in the higher R region is larger than that expected by the BPS theory. Additionally, the critical growth rate (R*), where the value of the coefficient deviates from the theory, depends on the growing crystal orientation. The R* in the <111> crystal is lower (〜0.5 mm/min) than that in the <100> crystal (〜3.0 mm/min). The following two different segregation models were proposed in order to explain the above phenomena. One is based on the deviati on from local equilibrium at the interface. The other is due to the combination of temperature dependence of physical properties and kinetic undercooling Especially, the latter model can quantitatively explain the experimental results.

View full abstract