- Email: [email protected]
Influence of impurities on two-level systems in amorphous ice
Physica B 316–317 (2002) 513–516
Influence of impurities on two-level systems in amorphous ice$ N.I. Agladze*, A.J. Sievers Laboratory of Atomic and Solid State Physics, Center for Radiophysics and Space Research, Cornell University, C17 Clark Hall, Ithaca, NY 14853-2501, USA
Abstract High-density amorphous (HDA) ice has a two-level system (TLS) optical density of states with a strength comparable to that found in many conventional glasses, whereas for the low-density amorphous (LDA) phase the TLS strength is at least 30 times smaller. Isotopic substitution resulting in 50% of the molecules having lower symmetry fails to change this strength ratio suggesting that the observed difference is not due to a breaking of the tetrahedral selection rule in the HDA phase. By doping the samples with different concentrations of LiCl, methanol, HF, NaCl, LiOH, and NaOH we have explored both the defect-induced TLS optical density of states and the glass transition characteristics in the LDA ice phase. r 2002 Elsevier Science B.V. All rights reserved. Keywords: Two-level systems; Amorphous ice; Glass transition
The discovery of the irreversible production of an amorphous phase of ice by pressure at low temperature makes possible the measurement of bulk amorphous samples [1,2]. The high-density amorphous (HDA) form is produced by the compression of regular ice at 77 K. Upon heating to B120 K it transforms to the low-density amorphous (LDA) form. At still higher temperatures B150 K it transforms to the cubic crystalline phase Ic and then at 225 K to Ih the hexagonal crystalline phase [3]. So it is possible to make optical and thermal measurements on a variety of phases with a single sample [4]. In this paper, low-temperature far infrared absorption by two-level systems (TLS) is studied in both amorphous ice phases. We find that HDA $
Work supported in part by NSF-DMR and by NASA. *Corresponding author. Fax: +16-072-55-6428. E-mail address: [email protected] (N.I. Agladze).
ice has a TLS optical density of states with strength comparable to that found in many conventional glasses whereas for the LDA phase the TLS strength is at least 30 times smaller. Upon isotopic substitution resulting in 50% of molecules having lower symmetry (HOD) this strength ratio was unchanged suggesting that the observed difference is not due to a breaking of the tetrahedral selection rule in the HDA phase. The lack of an isotope effect also severely limits what exactly can be tunneling to produce the TLS spectrum. By doping the samples with different concentrations of LiCl, methanol, HF, NaCl, LiOH, and NaOH we have explored both the TLS optical density of states and the calorimetric characteristics of the glass transition in the LDA ice phase. For LiCl and methanol we find that the doping significantly changes the TLS optical density of states and the glass transition temperature Tg while for other impurities studied the
0921-4526/02/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 4 5 2 6 ( 0 2 ) 0 0 5 5 8 - 6
N.I. Agladze, A.J. Sievers / Physica B 316–317 (2002) 513–516
514
effects on both quantities are small. Two types of correlations have been observed between the highand low-temperature properties of the glass [5]. One correlation is represented by an activation form, relating the TLS optical density with Tg for both impurities. The other, only observed with LiCl doping, is a qualitative correspondence between fragility of the liquid phase and the TLS optical density of states. The spectroscopic samples of the HDA ice were produced by compressing regular ice I h at liquid nitrogen temperature to 1.570.2 GPa. The regular ice was made from HPLC grade water, degassed by triple freezing–thawing procedure and contained in indium open-ended cups, which had been placed in a 10 mm die set. The entire assembly was then immersed in liquid nitrogen. A schematic diagram of the sample production technique is shown in Fig. 1. To prevent the ice from sticking to the anvils both were covered with a 6 mm Teflon film. The thickness of the resulting amorphous ice cylinders used in these experiments varied from 0.4
b
a
c
Fig. 1. High-pressure apparatus for the low-temperature production of high-density amorphous phase of ice. (a) Hardened die set; (b) liquid nitrogen in container; (c) ice sample in indium open-ended cup.
to 15 mm. The samples were removed at atmospheric pressure while still at 77 K, mounted in the sample holder while still under liquid nitrogen and the arrangement was then inserted into a precooled transmission light pipe-detector cryostat. Temperature dependent transmission spectra in the spectral region 2–30 cm1 were measured with a lamellar Fourier transform spectrometer together with a germanium bolometer which operates at 0.3 K [6]. At the conclusion of these measurements the LDA phase was prepared by first heating the samples to 145 K and cooling back down to low temperatures. The concentrations of the dopants were chosen high enough to have a measurable change in the far infrared spectra but less than a segregation limit or a glass forming limit as tested by calorimetry. The amorphous structure of the samples was tested using X-ray diffraction. A thin (B1 mm thick) sample of HDA ice was prepared in the same way as the samples for the far infrared measurements. In order to produce LDA and I c phases, the sample was subsequently warmed in situ to 140 and 190 K, respectively, and cooled back to 77 K. The X-ray diffraction spectra of HDA ice did not reveal any additional reflections except those produced by the indium ring and the copper holder and the results correspond well to the literature data both for vapor deposited [7] and pressure-induced amorphous phases [8]. The optical density of states of the TLS OðoÞ for a particular amorphous sample is obtained from the spectroscopic study of thermal bleaching of the absorption coefficient where the temperatures are varied between 1.2 and 11 K. Since the population factor of the TLS at these temperatures is controlled by a simple function tanh(_o=2kT) with both the frequency and temperature measured, the optical density of states prefactor to this population function is determined experimentally. The measured OðoÞ is shown in Fig. 2 for the isotopically mixed (H2O)0.5(D2O)0.5 ices with no impurities and for LDA H2O ice doped with a few concentrations of LiCl. The dash–dot curve is for the HDA phase and the broken curve is for the LDA phase. For comparison purposes the data (dotted curve) for the HDA phase of pure H2O are also shown. (The LDA results are the same as for
N.I. Agladze, A.J. Sievers / Physica B 316–317 (2002) 513–516
8
LiCl⋅7H2O
_
TLS optical density (cm 2) (x 1016)
6
4
3% LiCl 2 1% LiCl
0
0
5
10
15
20
25
_
Frequency (cm 1) Fig. 2. The TLS optical density of states for different amorphous ice systems derived from temperature-induced absorption spectra. The temperature has been varied between 1.2 and 11 K. Dotted curve: HDA phase of H2O; dash–dotted curve: HDA phase of (H2O)0.5(D2O)0.5 ; dashed curve: LDA phase of (H2O)0.5(D2O)0.5 ; the solid curves are labeled in the figure. The 1% and 3% LiCl solid traces identify defect-induced TLS spectra for the LDA phase.
the isotopic mixture.) These data can be compared with the results for LiCl 7H2O (solid curve), a standard electrolite glass [9]. The magnitude for the optical density of states of TLS for the HDA phase is comparable to the electrolite glass while the value for the LDA phase remains at least 30 times smaller than in the HDA phase, independent of the symmetry of the molecule itself. By extrapolating OðoÞ for the electrolyte glass to low frequencies the result can be combined with the published acoustic data for P% [9] to obtain the value of the transition dipole moment mb ¼ 1:4770:01D: The absence of a pronounced isotopic effect both for the range and for the strength of the TLS spectrum indicates that one TLS model
515
involving coupled librations of oxygen-centered tetrahedra is not valid for amorphous ice. A quantitative correlation between the low temperature characteristics of the doped sample and its glass transition properties can be established if there is no severe phase separation during cooling and the sample can be characterized by reasonable average values of the glass transition temperature and the fragility. Signs of phase separation were checked by methods of differential scanning calorimetry. For LiCl and methanol impurities in the LDA ice the phase separation was not detected (except for 7 mol% of methanol, where the melting of segregated pure methanol was observed). For NaCl a defect-induced TLS spectrum was not observed and the effect on the glass transition was negligible. For NaOH and LiOH calorimetry showed that phase separation occurred and the effect on TLS was small. Doping with HF produced even smaller effect on the TLS and the glass transition for HF and for LiOH was not observable. In summary, isotopic substitution and defect doping have been used to alter the optical properties of TLS in the two amorphous phases of a-H2O ice. A large TLS optical density of states is found for a-H2O and a-HOD in the HDA phase comparable with that observed for a standard electrolite glass. The LDA phase of both isotopic systems displays a surprisingly small number (30 times smaller) of TLS despite its known amorphous properties. Our detailed spectroscopic and thermal study of defect doping brings out a number of new features, among them that LiCl doping of LDA ice increases its TLS optical density of states, its glass transition temperature and its fragility. The absence of a TLS optical signature in the annealed LDA phase may signify that TLS are absent in the ideal amorphous material.
References [1] O. Mishima, L.D. Calvert, E. Whalley, Nature 310 (1984) 393. [2] E.G. Ponyatovsky, O.I. Barkalov, Mater. Sci. Rep. 8 (1992) 147.
516
N.I. Agladze, A.J. Sievers / Physica B 316–317 (2002) 513–516
[3] Y.P. Handa, O. Mishima, E. Whalley, J. Chem. Phys. 84 (1986) 2766. [4] N.I. Agladze, A.J. Sievers, Phys. Rev. Lett. 80 (1998) 4209. [5] N.I. Agladze, A.J. Sievers, Europhys. Lett. 53 (2001) 40. [6] N.I. Agladze, A.J. Sievers, S.A. Jones, J.M. Burlitch, V.W. Beckwith, Astrophys. J. 462 (1996) 1026.
[7] P. Jenniskens, D.F. Blake, Science 265 (1994) 753. [8] L. Bosio, G.P. Johari, J. Teixeira, Phys. Rev. Lett. 56 (1986) 460. [9] J.F. Berret, M. Meissner, Z. Phys. B 70 (1988) 65.
Influence of impurities on two-level systems in amorphous ice$ N.I. Agladze*, A.J. Sievers Laboratory of Atomic and Solid State Physics, Center for Radiophysics and Space Research, Cornell University, C17 Clark Hall, Ithaca, NY 14853-2501, USA
Abstract High-density amorphous (HDA) ice has a two-level system (TLS) optical density of states with a strength comparable to that found in many conventional glasses, whereas for the low-density amorphous (LDA) phase the TLS strength is at least 30 times smaller. Isotopic substitution resulting in 50% of the molecules having lower symmetry fails to change this strength ratio suggesting that the observed difference is not due to a breaking of the tetrahedral selection rule in the HDA phase. By doping the samples with different concentrations of LiCl, methanol, HF, NaCl, LiOH, and NaOH we have explored both the defect-induced TLS optical density of states and the glass transition characteristics in the LDA ice phase. r 2002 Elsevier Science B.V. All rights reserved. Keywords: Two-level systems; Amorphous ice; Glass transition
The discovery of the irreversible production of an amorphous phase of ice by pressure at low temperature makes possible the measurement of bulk amorphous samples [1,2]. The high-density amorphous (HDA) form is produced by the compression of regular ice at 77 K. Upon heating to B120 K it transforms to the low-density amorphous (LDA) form. At still higher temperatures B150 K it transforms to the cubic crystalline phase Ic and then at 225 K to Ih the hexagonal crystalline phase [3]. So it is possible to make optical and thermal measurements on a variety of phases with a single sample [4]. In this paper, low-temperature far infrared absorption by two-level systems (TLS) is studied in both amorphous ice phases. We find that HDA $
Work supported in part by NSF-DMR and by NASA. *Corresponding author. Fax: +16-072-55-6428. E-mail address: [email protected] (N.I. Agladze).
ice has a TLS optical density of states with strength comparable to that found in many conventional glasses whereas for the LDA phase the TLS strength is at least 30 times smaller. Upon isotopic substitution resulting in 50% of molecules having lower symmetry (HOD) this strength ratio was unchanged suggesting that the observed difference is not due to a breaking of the tetrahedral selection rule in the HDA phase. The lack of an isotope effect also severely limits what exactly can be tunneling to produce the TLS spectrum. By doping the samples with different concentrations of LiCl, methanol, HF, NaCl, LiOH, and NaOH we have explored both the TLS optical density of states and the calorimetric characteristics of the glass transition in the LDA ice phase. For LiCl and methanol we find that the doping significantly changes the TLS optical density of states and the glass transition temperature Tg while for other impurities studied the
0921-4526/02/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 4 5 2 6 ( 0 2 ) 0 0 5 5 8 - 6
N.I. Agladze, A.J. Sievers / Physica B 316–317 (2002) 513–516
514
effects on both quantities are small. Two types of correlations have been observed between the highand low-temperature properties of the glass [5]. One correlation is represented by an activation form, relating the TLS optical density with Tg for both impurities. The other, only observed with LiCl doping, is a qualitative correspondence between fragility of the liquid phase and the TLS optical density of states. The spectroscopic samples of the HDA ice were produced by compressing regular ice I h at liquid nitrogen temperature to 1.570.2 GPa. The regular ice was made from HPLC grade water, degassed by triple freezing–thawing procedure and contained in indium open-ended cups, which had been placed in a 10 mm die set. The entire assembly was then immersed in liquid nitrogen. A schematic diagram of the sample production technique is shown in Fig. 1. To prevent the ice from sticking to the anvils both were covered with a 6 mm Teflon film. The thickness of the resulting amorphous ice cylinders used in these experiments varied from 0.4
b
a
c
Fig. 1. High-pressure apparatus for the low-temperature production of high-density amorphous phase of ice. (a) Hardened die set; (b) liquid nitrogen in container; (c) ice sample in indium open-ended cup.
to 15 mm. The samples were removed at atmospheric pressure while still at 77 K, mounted in the sample holder while still under liquid nitrogen and the arrangement was then inserted into a precooled transmission light pipe-detector cryostat. Temperature dependent transmission spectra in the spectral region 2–30 cm1 were measured with a lamellar Fourier transform spectrometer together with a germanium bolometer which operates at 0.3 K [6]. At the conclusion of these measurements the LDA phase was prepared by first heating the samples to 145 K and cooling back down to low temperatures. The concentrations of the dopants were chosen high enough to have a measurable change in the far infrared spectra but less than a segregation limit or a glass forming limit as tested by calorimetry. The amorphous structure of the samples was tested using X-ray diffraction. A thin (B1 mm thick) sample of HDA ice was prepared in the same way as the samples for the far infrared measurements. In order to produce LDA and I c phases, the sample was subsequently warmed in situ to 140 and 190 K, respectively, and cooled back to 77 K. The X-ray diffraction spectra of HDA ice did not reveal any additional reflections except those produced by the indium ring and the copper holder and the results correspond well to the literature data both for vapor deposited [7] and pressure-induced amorphous phases [8]. The optical density of states of the TLS OðoÞ for a particular amorphous sample is obtained from the spectroscopic study of thermal bleaching of the absorption coefficient where the temperatures are varied between 1.2 and 11 K. Since the population factor of the TLS at these temperatures is controlled by a simple function tanh(_o=2kT) with both the frequency and temperature measured, the optical density of states prefactor to this population function is determined experimentally. The measured OðoÞ is shown in Fig. 2 for the isotopically mixed (H2O)0.5(D2O)0.5 ices with no impurities and for LDA H2O ice doped with a few concentrations of LiCl. The dash–dot curve is for the HDA phase and the broken curve is for the LDA phase. For comparison purposes the data (dotted curve) for the HDA phase of pure H2O are also shown. (The LDA results are the same as for
N.I. Agladze, A.J. Sievers / Physica B 316–317 (2002) 513–516
8
LiCl⋅7H2O
_
TLS optical density (cm 2) (x 1016)
6
4
3% LiCl 2 1% LiCl
0
0
5
10
15
20
25
_
Frequency (cm 1) Fig. 2. The TLS optical density of states for different amorphous ice systems derived from temperature-induced absorption spectra. The temperature has been varied between 1.2 and 11 K. Dotted curve: HDA phase of H2O; dash–dotted curve: HDA phase of (H2O)0.5(D2O)0.5 ; dashed curve: LDA phase of (H2O)0.5(D2O)0.5 ; the solid curves are labeled in the figure. The 1% and 3% LiCl solid traces identify defect-induced TLS spectra for the LDA phase.
the isotopic mixture.) These data can be compared with the results for LiCl 7H2O (solid curve), a standard electrolite glass [9]. The magnitude for the optical density of states of TLS for the HDA phase is comparable to the electrolite glass while the value for the LDA phase remains at least 30 times smaller than in the HDA phase, independent of the symmetry of the molecule itself. By extrapolating OðoÞ for the electrolyte glass to low frequencies the result can be combined with the published acoustic data for P% [9] to obtain the value of the transition dipole moment mb ¼ 1:4770:01D: The absence of a pronounced isotopic effect both for the range and for the strength of the TLS spectrum indicates that one TLS model
515
involving coupled librations of oxygen-centered tetrahedra is not valid for amorphous ice. A quantitative correlation between the low temperature characteristics of the doped sample and its glass transition properties can be established if there is no severe phase separation during cooling and the sample can be characterized by reasonable average values of the glass transition temperature and the fragility. Signs of phase separation were checked by methods of differential scanning calorimetry. For LiCl and methanol impurities in the LDA ice the phase separation was not detected (except for 7 mol% of methanol, where the melting of segregated pure methanol was observed). For NaCl a defect-induced TLS spectrum was not observed and the effect on the glass transition was negligible. For NaOH and LiOH calorimetry showed that phase separation occurred and the effect on TLS was small. Doping with HF produced even smaller effect on the TLS and the glass transition for HF and for LiOH was not observable. In summary, isotopic substitution and defect doping have been used to alter the optical properties of TLS in the two amorphous phases of a-H2O ice. A large TLS optical density of states is found for a-H2O and a-HOD in the HDA phase comparable with that observed for a standard electrolite glass. The LDA phase of both isotopic systems displays a surprisingly small number (30 times smaller) of TLS despite its known amorphous properties. Our detailed spectroscopic and thermal study of defect doping brings out a number of new features, among them that LiCl doping of LDA ice increases its TLS optical density of states, its glass transition temperature and its fragility. The absence of a TLS optical signature in the annealed LDA phase may signify that TLS are absent in the ideal amorphous material.
References [1] O. Mishima, L.D. Calvert, E. Whalley, Nature 310 (1984) 393. [2] E.G. Ponyatovsky, O.I. Barkalov, Mater. Sci. Rep. 8 (1992) 147.
516
N.I. Agladze, A.J. Sievers / Physica B 316–317 (2002) 513–516
[3] Y.P. Handa, O. Mishima, E. Whalley, J. Chem. Phys. 84 (1986) 2766. [4] N.I. Agladze, A.J. Sievers, Phys. Rev. Lett. 80 (1998) 4209. [5] N.I. Agladze, A.J. Sievers, Europhys. Lett. 53 (2001) 40. [6] N.I. Agladze, A.J. Sievers, S.A. Jones, J.M. Burlitch, V.W. Beckwith, Astrophys. J. 462 (1996) 1026.
[7] P. Jenniskens, D.F. Blake, Science 265 (1994) 753. [8] L. Bosio, G.P. Johari, J. Teixeira, Phys. Rev. Lett. 56 (1986) 460. [9] J.F. Berret, M. Meissner, Z. Phys. B 70 (1988) 65.