Vibrational and relaxational properties of crystalline and amorphous ices

  • Published on

  • View

  • Download


Thermochimica Acta 461 (2007) 1443ReviewVibrational and relaxational properties of crystalline and amorphous icesG.P. Johari a,, Ove Andersson bAbstractPure watetallographica(Tg = 136 K)1 GPa to ahexagonal icwaters amoon heating thices that occThis processpressures. Tkinetically u1 GPa pressure, the pressure-amorphized solid relaxes to a lower energy state, becoming ultraviscous water at 140 K. But on heating at ambientpressure, it irreversibly transforms slowly to a low-density amorph that differs from glassy water and vapour-deposited amorphous solid. 2007 ElseKeywords: IcContents1. Introd2. Expe3. Vibra3. Relax4. Chara6. Time7. Mech8. Therm9. SummAcknRefer CorresponE-mail ad0040-6031/$doi:10.1016/jvier B.V. All rights; Amorphous solid and glassy water; Thermal conductivity; Heat capacity; Dielectric relaxation; Amorphization mechanismuction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15rimental methods for measurements at high pressures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16tional properties of the crystalline and amorphous ice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17The thermal conductivity of crystalline and amorphous solid water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Heat capacity of waters high-density amorph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20ation properties of the amorph and conversion to ultraviscous water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22Preparation of waters amorphous solid and glassy states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22Calorimetric behaviour, relaxation and glass-softening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23Dielectric relaxation of amorphous solid and glassy states of water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25cteristic changes during pressure-amorphization of ice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-, pressure- and temperature-dependence of the extent of amorphization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32anism of pressure-amorphization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35odynamics and kinetics of pressure collapsed amorph and of ultraviscous water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37ary and concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39owledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41ences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41ding author.dress: (G.P. Johari). see front matter 2007 Elsevier B.V. All rights reserved..tca.2007.03.011a Department of Materials Science and Engineering, McMaster University, Hamilton, Ontario L8S 4L7, Canadab Department of Physics, Umea University, 901 87 Umea, SwedenAvailable online 24 March 2007r forms 15 crystalline ices at different temperatures and pressures, and its solutions containing small molecules form three crys-lly distinct clathrates. Its vapours deposited on a substrate at T < 100 K produce a porous amorphous solid and pure water vitrifieswhen hyperquenched in micron-size droplets. At a temperature below 140 K, hexagonal and cubic ice collapse when pressure exceeds30% denser amorphous solid, which on heating at ambient pressure transforms to an amorphous solid with density similar to that ofe. In this essay, we describe (i) the thermal conductivity of the ices and clathrates and the thermal conductivity and heat capacity ofrphous solids, their thermodynamic paths and their transformations, and (ii) the dielectric relaxation time of ultraviscous water formede amorphous solids. We also describe the characteristics of pressure collapse and subsequent amorphization of hexagonal and cubicurs over a period of several days according to a stretched exponential kinetics and a pressure-, and temperature-dependent rate attributed to the production of lattice faults during deformation of the ice and the consequent distribution of the Born instabilityhis ultimately produces a kinetically unstable high-energy amorphs in the same manner as random deformation of crystals producesnstable high-energy amorphs, with density and properties depending upon their temperaturepressuretime history. On heating atG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 151. IntroductionPure water is known to form 15 crystalline phases. It isbelieved that it forms also at least three amorphous solids,although like any amorphous solid state of a material, the numberof its amorphous solids may be virtually infinite. The crystallinephases have been named hexagonal ice (ice Ih) and cubic ice(ice Ic), both of which are bulkier than water, and ices IIXII,as reviewed in Ref. [1], and two recently found phases ices XIII[2] and XIV [3] all of which are denser than water. Regions oftheir thermodynamic stability and the metastable conditions inwhich ices Ic, IV and XII have been formed are shown in thetemperaturepressure phase diagram in Fig. 1. Ice X, which isnot shown, has a centrosymmetric structure of hydrogen bonds.It forms at pressures higher than 44 GPa [4]. At ambient pres-sure, bulk water freezes to ice Ih, but sub-micron size dropletsof pure water freeze to ice Ic, as described in Ref. [5]. Onlyices IIIVII and XII have been made by compressing bulk waterand therefore these are shown to have a phase boundary withliquid water, except ice XII. Ice IV is formed by nucleation withorganic molecules [6] and it is a metastable phase in the temper-ature and pressure range of ice V. Ice XII is also a metastablephase formed directly from water in the presence of silica fibreswhich presumably act as a nucleating surface [7]. Many of thehigh-pressure ices have recently been produced by crystalliza-tion of a high-density apparently amorphous solid water or byheating the metastable ice XII [8].Fig. 1. Phasecrystalline phbeen made byIh and heatintionally disordof all those stwith liquid wawater containordered phaseto become orimetry. The shawhich high-dare stable forIn the crystal structures of the ices, each H2O moleculeforms four hydrogen bonds with its four neighbours. Two hydro-gen bondseach to twoing protonmolecule [a tetrahedrstructure ooxygen atonearly tetramoleculesand the hydlinear. Thiof their strform direcsites are rabe orientatno long-rastructures.abundant oand in itstals of icevapour.Becausemolecules,value of ththis entropis the gasing the temremaining cmade by ortively, iceslingtionnnealowlrizins arenallylingrmauratter cos ofen ball mown,] ande antiesntsen mres [].ter alamotratediagram of water and ice showing the regions of stability of 15ases and three amorphous phases. The amorphous phases havehyperquenching of water, pressure-amorphization of ices Ic andg of the pressure-amorphized solids. H2O molecules are orienta-ered in the structure of metastable ice Ic, and also in the structureable or metastable ices that have an equilibrium phase boundaryter, even when they form in the domain of another ice by freezinging nucleating agents. All of these hydrogen-atoms or protons dis-s have a finite configurational entropy. All these ices are expectedentationally ordered on cooling with a change in the crystal sym-ded region of the diagram is for the temperaturepressure range inensity and low-density amorphs and hyperquenched glassy watera long enough period to be studied.on cooorientatime awhat spressuproton[1]. Fiby cootransfoshort dWacrystalhydrogthe smare kn[10,11methanquantisedimehydrogstructu[19,20Waknowna subsare formed by a molecule donating protons, oneneighbouring H2O molecules, and two by accept-s, one each from the other two neighbouring H2O9]. Thus, an H2O molecule is hydrogen-bonded inal arrangement in the ice structures, resembling thef covalent bonds in silica. The angle between thems of the H2O molecules in ices Ih, Ic and VIII ishedral and the hydrogen bonds between two H2Oare linear. This angle varies greatly in other icesrogen bonds between two H2O molecules are non-s variation allows for the increase in the densityuctures. In the crystal structure of all the ices thattly by cooling water, H2O molecules at the latticendomly oriented. Therefore, these ices are said toionally disordered, or proton-disordered. There isnge order for the hydrogen atom positions in theirIt is probably correct to say that ice Ih is the mostrientationally disordered crystal phase on the Earthatmosphere, and clouds occasionally contain crys-s Ih and Ic along with water droplets and waterof the orientational or proton-disorder of H2Oices Ih, Ic, IIIVII and possibly XII have a finitee configurational entropy. The maximum value ofy is equal to R ln(3/2) (=3.27 J (mol K)1), where Rconstant [1]. Except for ice II that forms by vary-perature and/or pressure of ices Ih, Ic, III and V, therystalline phases, ices VIII, IX, XI, XIII and XIV areientationally ordering, or proton-ordering, of respec-VII, III, Ih, V and XII. Ice VIII forms graduallyof ices VII and IX on cooling of ice III, whereasal ordering of ice Ih is achieved by isothermal, longling of the pure or doped ice Ih to produce, some-y, significant amounts of ice XI. Ice X is produced byg ice VII at 100 K to above 62 GPa. In its structure,symmetrically placed between two oxygen atoms, ice XIII [2] and ice XIV [3] have been producedof HCl-doped ices V and XII, respectively, and thetion has been found to be thermally reversible in aion.ntaining certain small molecules freezes to produceice clathrates. In their structure, H2O molecules formonds to produce cage-like structures which confineolecules. Three such crystalline forms of clathratestwo have cubic structures in different space groupsone hexagonal [12]. Some of the clathrates containd other hydrocarbons, and are found to occur in vastin the cold regions of the Earth and in submarine[1315]. Recently, it has been found that clusteredolecules can be stored in the cages of the clathrate1618]. These clathrates are of practical importanceso forms at least three amorphous solids. The oldestngst these was made by depositing water vapours onheld at a temperature below 100 K [21]. As formed,16 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443this solid is relatively porous and it exothermally anneals onheating [22]. Water droplets of less than 3m diameter dis-persed as a77 K to proglassy watedepositedshow identfor the heacous waterwater havetemperaturthe pressurthe (bulkieent pressurknown as tyears haveis generic.sity productemperatursolid HDAambient prto another77 K [28],Because oflow-densityis pressuridynamic stthe HDA fLDA is heato ice Ic wsolid wateand it is esamorphouscomets and[3843].Crystallable exampinterest. Thof cryobiolplanets andtional propsolids, partWe also dehave beensolids andmechanismthermodynemphasis hon watersscatteringwe includetal methodcapacity anpressures aphous solidhave beenonly briefly2. Experimental methods for measurements at highpressuresur studies of water and other materials, a hot-wire methoden used to measure both the thermal conductivity, ,e product of the density and heat capacity [34,4550].ethod is based on a mathematical solution of the time-ent equation for heat conduction and has been describedil earlier [51]. In this solution, the temperature rise Tnfinitely long, infinitely conducting wire immersed in anly large specimen is given by [52]2q2301 exp(u2)u3{(uJ0(u) J1(u))2+(uY0(u) Y1(u))2}du,(1)q is the constant heating power input per unit length,Cp/(wCw), = t/(Cpr2), t the time, r the radius of there, and Cp the density and heat capacity of the specimen,Cw the density and heat capacity of the hot-wire, J0 andsel functions of the first kind of zero and first order, andY1 are Bessel functions of the second kind of zero sample cell, illustrated in Fig. 2(A), is made of. The hot-wire itself is a 0.1 mm diameter 40 mm long Ni-he Teflon container is 30 mm deep and 37 mm internalA) Thd heaent os andy.n aerosol in N2 gas have been rapidly quenched toduce 23 mm-thick opaque layer of hyperquenchedr [2325]. In their annealed states both the vapour-amorphous solid and hyperquenched glassy waterical thermodynamic behaviours, the same Tg of 136 Kting rate of 30 K min1 [26] and become ultravis-at 140 K. The other two forms of amorphous solidbeen produced by pressurizing ices Ih and Ic at ae below 140 K to 1 GPa [2733]. The density ofe-amorphized state thus obtained by the collapse ofr) ices Ih and Ic structures is 1.17 g ml1 at ambi-e and 77 K [27]. This solid has therefore becomehe high-density amorph or HDA. Studies in recentshown, and it is now agreed, that the term HDAIt refers to all amorphous solids of unknown den-ed by pressure collapse of ices Ih and Ic at differente, pressure, and time conditions [3436]. When thisis recovered at ambient pressure and is then heated atessure, it transforms irreversibly (and exothermally)amorphous solid whose density is 0.93 g ml1 atwhich is similar to the density of ices Ih and Ic.its low-density, this third form has been called theamorph (LDA). When LDA at a low temperaturezed to 0.4 GPa, it converts to HDA. A thermo-ate analysis has shown that this HDA differs fromormed by pressurizing ice Ih to 1 GPa [37]. Whented at ambient pressure, it transforms exothermallyith negligible change in the density. Amorphousr has been found occasionally in the atmosphere,timated that large quantities of both crystalline andsolid forms of water are present in the nuclei ofinterstellar dust, and in satellites and giant planetsine and amorphous solid states of water are remark-les of hydrogen bonding and are therefore of generaley are also of practical importance in the disciplineogy, food sciences, astrophysics and geophysics ofsatellites. In this essay we describe some of the vibra-erties of the ices, as well as of waters amorphousicularly their thermal conductivity and heat capacity.scribe how calorimetry and dielectric spectroscopyused for characterizing crystalline and amorphousdiscovering new forms. Finally, we describe theof pressure-amorphization of ices Ih and Ic, and theamics and relaxation of the amorphous forms. As theere is on thermal properties, most of the informationamorphous solids obtained by X-ray and neutronstudies is excluded. For the sake of completeness,a brief description of the rather unusual experimen-s used for measuring the thermal conductivity, heatd dielectric relaxation time of these solids at highnd low temperatures. Studies of the waters amor-s by diffraction methods and by computer simulationreviewed recently [44], and therefore are mentionedhere.In ohas beand thThe mdependin detaof an iinfiniteT =where = 2hot-wiw andJ1 BesY0 andfirst orTheTeflonwire. TFig. 2. (tivity anarrangemial stresassemble high-pressure cell assembly for measuring the thermal conduc-t capacity (left) and the dielectric relaxation time (right). (B) Thef the high-pressure vessel showing the application of the uniax-arrangement that allows the use of helium cryostat in a massiveG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 17diameter with a tightly sealing Teflon cover. It closely fits insidethe piston-cylinder assembly of internal diameter 45 mm of ahigh-pressuuse the limwire is plathe Teflon cis surround1.4 s long pwire resistaature rise obetween itsthe heaterthe temperthe hot-wirinaccuracyrespectivel and Cpin good thphous solidensured onthe low-preperature codecreases wmeasuremedeviationsorder of maDielectrusing two tyof nominalcentric elec[53,54]. Thstructed frothe other bybut only witric electroused in theabove, filleinto the higted closelycapacitor ima Solartron100 Hz to 110 mHz togenerator isthe capacitplaced in ssimultaneoing at leascapacitanceeach frequeused by FoelectronicsThe limdielectric loof similar pJohari andwith the ceelectrode acting also as a pressure vessel. In their study the inter-electrode distance increased with increase in pressure, whichd thin thwhegeomraisr reaficanrmitter ththe gnd dl ofigmure Wthicked inl diusidergasressutionresisestied fs meelalateds hie diere coandthermery son cCped tasseate osolidre-amragethe rer grolingfromratihouhe thatere a nas nre assembly illustrated in Fig. 2(B). In order to bestited space available in high-pressure equipment, theced horizontally in a ring of constant radius withinell. At each heating event, the hot-wire probe, whiched by the material under investigation, is heated by aulse of nominally constant power during which thence is measured as a function of time. The temper-f the wire is then calculated by using the relationresistance and temperature, i.e. the wire acts as bothand the sensor for the temperature rise. Eq. (1) forature rise with time is fitted to the data points fore temperature rise, thereby yielding and Cp. Thein and Cp thus measured are 2% and 5%,y, at 298 K. It should be stressed that for measuringby this method the wire and the sample must remainermal contact. In the studies of the ices and amor-water, good thermal contact with the hot wire isly at pressures higher than 0.05 GPa, and thereforessure limit of the study is 0.05 GPa. Also, as the tem-efficient of electrical resistance of the nickel hot-wireith decrease in the temperature, the inaccuracy of nts increases to about 4% at 40 K. The standardof the data obtained in these measurements are angnitude smaller than the inaccuracy.ic measurements at high pressures are performed bypes of dielectric cells, one is a parallel plate capacitorly 125150 pF air capacitance, and the second a con-trode capacitor of nominally 19 pF air capacitancee parallel plate capacitor consists of six plates con-m either stainless steel or brass, each separated frompoly(ether-etherketone) spacers. A similar capacitorth four plates is illustrated in Fig. 2(A). The concen-de cell is constructed from a Cu alloy. The capacitorstudy is placed inside the Teflon container describedd with water, sealed with the Teflon cover and insertedh-pressure piston-cylinder apparatus in which it fit-. The capacitance and conductance of the dielectricmersed in the ice sample are measured by means of1260 impedance analyser in the frequency range ofMHz. For measurements at frequencies in the range100 Hz, a Hewlett-Packard model 33120A functionused to provide a sinusoidally varying signal to bothor containing the sample and a reference capacitoreries. The voltages over the capacitors are measuredusly by two Hewlett-Packard 3457A voltmeters dur-t one period by collecting 100 data points, and theand conductance of the sample are determined forncy. The measurement assembly is based on the onersman [55], but with a new function generator and.iting high-frequency dielectric permittivity andss during the pressure collapse of ice Ih and the effectressure increase on an ice clathrate were measured byJones [56] who used a concentric electrode capacitorll constructed from a 2% BeCu alloy with the outerchangeearlierice V,in thealso onies. Foinsignitric peto bettInCp a20 mfrom SUltrap5 mmmountinternaappliedkept unhelium[59]. Pfor fricof thesure isincreasature ichromis estimThiing thmeasusampleentialThus vphizatithe ,be notin thisslow rphouspressuan avevidingby othand coraised3. Vibamorp3.1. Tsolid wLiktures he cell geometry by an amount that had been estimatede study of dielectric properties of ice VI and later ofre details may be found [57,58]. A similar changeetry of the concentric electrode cell had occurreding the pressure from ambient to 1 GPa in our stud-sons given earlier [5658], this change is regarded ast. After including the measurement errors, the dielec-ivity and loss measured at high pressures are accuratean 3%.eneral experimental procedures for measuring ,ielectric properties, the Teflon cell is filled withpure water (tissue-culture grade water purchasedaAldrich, or the water purified by using Milli-QaterSystems) and then sealed with a tightly fitting,, Teflon cover. This hermetically sealed assembly isside a piston-cylinder type pressure vessel of 45 mmameter, as illustrated in Fig. 2(B) and the load isng a hydraulic press. The whole pressure vessel isvacuum and cooled by the refrigerator using a closedcycle, as illustrated in Fig. 2(B) and described earlierre is determined from the load/area with a correctionwhich is established using the pressure dependencetance of a manganin wire. The inaccuracy in pres-mated as 40 MPa at 1 GPa and 298 K and, due toriction, 60 MPa at 40 K and 1 GPa. The temper-asured inside the Teflon cell by using an internalumel thermocouple. The inaccuracy in temperatureas 0.5 assembly and equipment for measur-lectric properties, and Cp, also allows one tontinuously the temperature difference between thethe Teflon cell wall. This is equivalent to differ-al analysis performed at a very slow heating rate.low crystalcrystal transformation and crystal amor-an also be studied with ease simultaneously withand dielectric measurements. For caution, it shouldhat experiments on pressure-amorphization of icembly are particularly prone to failure because thef pressure and temperature change causes the amor-to frequently crystallize, particularly in the broadorphization range of 0.81.1 GPa. In our experienceof only one in five experiments is successful in pro-equired data. This has not been the case in the studiesoups who have used 10 times higher pressurizationrates, and in which the pressure in most studies was0.1 MPa to 1.5 GPa in 5 min.onal properties of the crystalline ands iceermal conductivity of crystalline and amorphousormal crystal, an amorphous solid at low tempera-o configurations available to its structure. Therefore,18 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443its properties are seen to be entirely vibrational in origin. Theseare determined mainly by the intermolecular distances, whichusually decrease with increase in density as the pressure isincreased or the temperature is decreased. In this respect, thephonon heat capacity, Cp, of an amorph changes in a mannersimilar to that of a crystal, i.e. decreases with decrease in tem-perature T and increase in pressure p. But as structural disorderchanges the mean free path for propagation of phonon withtemperature differently from that in a crystal and other phononscattering modes become prominent, thermal conductivity, ,of amorphous solids changes quite differently from those ofthe crystalline state. It seems that of all the non-configurationalthermodynamic properties of solid states of water, only of itscrystalline and amorphous solids has been studied in detail athigh pressures. It was done by the Umea University group aspart of their comprehensive study of the solid forms of water, ofthe equilibrium crystalcrystal, irreversible crystalnoncrystal,and noncrystalcrystal transitions, as well as crystal melting.They have also studied the effect of temperature on of waterscrystalline and amorphous solid phases. Since the magnitude of depends upon both the frequency of phonons and the distanceof their propagation through the structure of a material, first wedescribe its characteristic features.For convenience of discussion, the measured data for thesolid formsperature inby Ross etindicated nof water, ied. It nowrecently [4ture by dehave referrA cursois determinFig. 3. Thermtaining tetrahystates, and twture. The cryswere made arwhich in turn determine the phonon frequencies, but also by theUmklapp scattering and other anharmonic effects that decrease. For exam 3a lower tFig. 3, andhigher vasame densiice Ic is remof an ice clain the protothe proton-the ice clathand appearat higher teweakly onon the typeit seems thamong thesAs is secharacteristalline phasand its temues are welx is a constrangecrysonclnd ae reperCvC isphon. At hourcpp sto ain zo a pcantlf thees nen sucwsses ptheas Tper1.eveonlyhe dcrytheonsof water and liquid water are plotted against the tem-Fig. 3, together with the results previously reportedal. [4547]. The various solids and liquid water areext to the plots. Of the total of 15 crystalline phasesof all except ices IV, X, XIII and XIV has been stud-seems that of ice XII had probably been studied9], but as it still needs a confirmation of the ice struc-termining the diffraction studies of the sample, weed to it as metastable ice/ice XII in Fig. 3.ry examination of Fig. 3 shows that of the icesed not only entirely by their density and structure,al conductivity of most of the crystalline ices, the ice clathrate con-drofuran as guest molecules in the (a) ordered and (b) disorderedo amorphous solid phases of water is plotted against the tempera-talline and amorphous ices and the pressure at which measurementse as indicated. Data are taken from Refs. [4650,67].in thecal forhave cnear awith ththis tem = 13wherev theeventsmain sUmklabine inBrilloualent tsignifisum oenergiipate iIt folloincreaing, orvariesthe temTHowcommcases tfor thefollowdeviatiple, the higher density ice V (1.23 g cm [60]) hashan the lower density ice I (0.93 g cm3), as seen inice VIII with a higher density (1.46 g cm3 [60]) has alue than ice V. More clearly, ice Ih and ice Ic have thety, phonon frequency and heat capacity, and yet ofarkably less than that of ice Ih. Thermal conductivitythrate containing tetrahydrofuran as guest moleculesn-ordered state is plotted as curve (a) in Fig. 3, and indisordered state in curve (b). The plots show that ofrate is the lowest amongst all the solid forms of waters to be somewhat continuous with of liquid watermperatures. Since has been found to depend onlythe crystal structure of a clathrate and also weaklyof guest molecules in its symmetrical cages [61,62],at of the ice clathrates would be generally loweste solids.en in Fig. 3, and its temperature dependence istic of the ice crystalline phases and that different crys-es can be distinguished by both the magnitude of perature dependence. The thermal conductivity val-l described by an empirical relation, Tx, whereant specific to the ice and its value for different ices is0.61.4 [33]. Such temperature dependence is typi-talline phases [62]. Moreover, theoretical discussionsuded that of mono-atomic crystals at temperaturesbove their Debye temperature would vary linearlyciprocal temperature. In terms of the Debye theory,ature dependence is written as [63]2s (2)the specific heat capacity contribution from phonons,on velocity and s is the time between scatteringigh temperatures, above the Debye temperature, thee for phonon scattering is known to be three-phononcattering [64]. In such a process, two phonons com-third, which has a wave vector that is outside the firstone. Consequently, this phonon is physically equiv-honon inside the Brillouin zone, which moves in ay different direction than that indicated by the vectororiginal two phonons. The number of phonons withar the Brillouin zone boundaries, which can partic-h an event, increases proportionally to temperature.that the probability for Umklapp scattering eventsroportionally with T and the time between scatter-Umklapp scattering time, which is now equal to s1. As the quantities , C and v in Eq. (2) vary withature much less than s, this leads to the relation,r, deviations from the T1 relation have beenobserved in crystalline solids [62,63], and in someeviations are relatively strong. This is the case alsostalline ices, as is evident in Fig. 3, which seem tovariation, Tx, with x in the range 0.61.4. Theof course are related to the structure of the ices and theG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 19hydrogen bonding in them, and it is not expected that they wouldfollow the Debye theory for thermal conductivity. Nevertheless,for ices Ih,they seemof ice VIIIindicated band IX, x iture depenT1. Fture stablewhich indithe anticipaices, the strx > 1, mayand/or to aanharmoniis attributedorder. Thebut one soucrystallinitsome of thexcluded bdisorderedordered iceThermainvariablytalline phalow-densitybeen foundperature, mof behavioudensity, higher thanof the two1.8 times thThe structucubic latticinterpenetrMoreover,XI and Ic hpressure (hpressure icbonds withgen atomsof phonons2.4 GPa prfor .In the crhydrogen bbouring oxmolecules [formed bysmall inorgneither lieorientationin the claththeir molectemperatures, long range orientational order of tetrahydrofuran(THF) molecules inside the cages of the ice clathrate has beend byon thanndetamped are. Bcreain Fimiclinete, c-TH. 3, anallytionandtiveltribuing rity [6ed aes blso boriongslesmalrystafor aes alhaving aintertrononoy Relectechnowhizeded bingA ma inressuuickressuultime sty ofg icrelatn ule beII and XI, the value of x is close to 1, and thereforeto follow the Debye equation. Thermal conductivityhas an unusually strong temperature dependence asy the high value of 1.4 for its x. For ices Ic, III, VII,s in the range 0.80.9, which is a weaker tempera-dence than that expected from the Debye equation,or the metastable phase ice XII, and high tempera-phases of ices V and VI, x is in the range 0.60.8cates the weakest temperature dependence. Despiteted less validity of the Debye equation for most of theonger temperature dependence of on T, i.e., whenbe attributed to the higher-order phonon processesfinite thermal expansion which is determined by thec forces, whereas a weaker temperature dependencecommonly to scattering arising from structural dis-reason for these differences is not well establishedrce for a low value of x can also be a somewhat poory and/or the orientational-disorder in the structure ofe ices. However, the latter possibility seems to bey the finding that the large x for the orientationallyice Ih is almost the same as for the orientationallyXI [48].l conductivity of a liquid and solid is known to almostincrease with its density, and a high-density crys-se has often been found to have a higher than acrystalline phase [65]. But of ices Ih and Ic hasto decrease with increase in pressure at a fixed tem-aking these two as exceptions to this general patternr. In relation to the crystalline ices, with increasingof ices VII and VIII, the highest density phases, isof all other phases, and despite the same density, of the orientationally ordered ice VIII is aboute value for the (orientationally disordered) ice of these two ices consists of two interpenetratinges and it is not certain how s in Eq. (2) is changed byation of two independently hydrogen-bonded lattices.the lowest density, ambient pressure phases: ices Ih,ave a higher value of than most of the other high-igh-density) ices. It is apparent that these ambiente phases, whose structure contains linear hydrogena tetrahedral angle between the neighbouring oxy-have a highest mean free path or lowest scattering, and hence the highest . Only, ices VII and VIII atessure, which are much denser have a higher valueystal structure of the ambient pressure ice clathrates,onds are non-linear and the angle between the neigh-ygen atoms varies considerably for different water11]. In addition, the symmetrical cage-like structuresH2O molecules in the clathrate structures containanic and organic molecules. These guest moleculesexactly at the centre of the cage nor have the samefrom one cage to the other. Thus the guest moleculesrate structures have a long-range order neither forular positions nor for their orientations. But at lowdetectemisedsame mied inhigh teices anperatuwith inis seenthey mcrystalclathra17 H2Oin Figentatioorienta0.60respecwas atscattermonicobservincreasmay a[48] onAmmolecutively ssame cthat, asbecomlike beby usiwhichcause slike ph1981 bthermouse inWeamorpproducand takthe HD1.5 GPboth pmore qslow pbe theis to bidentitsurizinover aover, ato havdielectric measurements by Davidson [11], who sur-is clathrates potential for polarization-cooling in theer as magnetic cooling. This ordering has been stud-il by Suga et al., by using calorimetry [66]. Even atratures these clathrates have a much lower than theremarkably different variation of its value with tem-riefly, in contrast with crystals, whose decreasesse in the temperature, of ice clathrates increases, asig. 3. Thus, in their against temperature behaviour,glasses and other amorphous solids and not theices. The plot of against temperature for the iceontaining tetrahydrofuran (THF) as guest molecules,F and 1.8 104:1 molar ratio of KOH:H2O [67,68]re shown in curve (a) for the low temperature ori-ordered and in curve (b) for the high temperatureally disordered phases that x is negative, equal to0.24 for the low and high temperature phases,y. The 15% difference between of the two phasested to proton-ordering, which seems to decrease theate slightly, possibly through a decrease in the anhar-7]. A similar change in the magnitude of has beent the temperatures of ices Ih to XI transition where y 20%, but without a significant change in x. Thise explained in terms of decrease in anharmonicityentational ordering of ice Ih.t the ice clathrates containing different guestin their cage-like structures, changes by a rela-l amount when the guest molecules are changed in thel structure of the clathrate [61,62,68]. This indicatesn amorphous structure, phonon propagation distanceso strongly limited in the ice clathrates. This glass-our of the (crystalline) clathrates has been explainedresonance scattering model [69,70], according toactions between the encaged molecules and phononsg phonon scattering. It is worth noting that the glass-n scattering of a crystal was originally observed inoss et al. [71]. It is now being used [72] to obtaintric materials [64] with improved characteristics fornology.discuss the magnitude of of the pressure-solid water, HDA. In our study, this solid wasy very slowly pressurizing ices Ih and Ic at 130 Kseveral hours to reach 1.2 GPa. This is in contrast toade by rapidly pressurizing ices Ih and Ic at 77 K toa few minutes. Since the pressure-amorphization isre and time-dependent [73], and reaches completionly at higher temperatures, the amorph produced byrization of ice Ih at 130 K in our study is likely toate state of the high-density amorph. This aspectressed because there is a confusion regarding thevarious HDAs that had been produced by pres-e Ih at different temperatures to different pressuresively short and often unspecified time period. More-timate, presumably highest density HDA seemsen produced when the amorphous solid formed by20 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443pressurizing ice Ih at 77 K was heated to 160 K while undera pressure greater than 0.8 GPa [74]. The plots in Fig. 3 showthat amonghighest den value ofof long-ranexcitationsincreases wphous or stpronouncedx0.03,phous SiO2a high Debin a discusperature.)Althougsolid, we uas the strucphous, as oTo do so, wwhich is wror the lineteristic timbe indepenobtained bfor the Umtively. Oneformed byingly smalldecreases,independenin comparideterminedvary with tat temperatwith increadecreases,the relativeindicates this only sligvelocity.Thermawhich has bat p < 0.07It shows this inconsiststant, or evis a characing with iIt is also reof is conLDA behavdeposited aLeitner [76[77] by usia pre-facto values sldistinctively, the calculations have shown that increases withincreasing temperature, opposite of that found for LDA.rmamoring to itsen oands thelati, paree sooundprotom tin icperaen-aIh ogent there doalso[80,8tes penerer hend tt forisordcontThisolidre.alrefromwatent wadiffenal eer [4uid ceat ccomeases. ThGPatingK ws ames athehangre cage wst the solid forms of pure water, is lowest for thesity HDA produced in our experiments. The lowera disordered solid is generally attributed to the lackge propagating phonons, at least for high-frequency, and this is likely to be the case for this HDA. Its ith the temperature, as for most glasses and amor-ructurally disordered solids, but the increase is less, as is indicated by its lower value of the quantitythan is normally found for glasses, of which amor-is a well-known example [75]. (Note that SiO2 hasye temperature, which should be taken into accountsion of weak or strong dependence of on tem-h Eq. (2) had been derived for an ideal crystallinese it here to interpret qualitatively the change of ture of a material changes from crystalline to amor-ccurs on pressure-amorphization of ices Ih and Ic.e consider the grain boundary scattering time bsitten as, bs = /v, where is the inter-grain distancear dimension of the crystal [63]. Since the charac-es for various scattering processes are assumed todent of each other, the resulting scattering time isy the sum: 1s = 1bs + 1Ums for the boundary andklapp scattering denoted by bs and Ums, respec-reasonable approach thus is to consider that the solidpressure-amorphization of a crystal has a vanish-crystal size. In this approach, as the size of crystalss becomes smaller and finally reaches a temperature-t, limiting value. In that case, 1Ums can be neglectedson with the large 1bs value and hence would beby , C and v in Eq. (2). The quantities , C and vhe temperature but much less than s. To elaborate,ures near the Debye temperature, C increases slowlyse in the temperature, whereas the phonon velocityand thus the two partially compensate. In this respect,ly small increase in with increase in the temperatureat increase in C of HDA with increasing temperaturehtly more than the concurrent decrease in the phononl conductivity of the low-density amorph, LDA,een produced by heating HDA at ambient pressure orGPa, is also plotted against the temperature in Fig. decreases with increase in the temperature. Thisent with the generally held view that a roughly con-en slightly increasing with increasing temperatureteristic of a disordered solid and a strongly decreas-ncreasing temperature is a characteristic of a crystal.markable that as far as the temperature dependencecerned, HDA behaves like an amorphous solid andes like a crystal [50]. For comparison, of vapour-morphous solid water has been calculated by Yu and], who used the term glassy water, and by Klingerng molecular dynamics simulations and Eq. (2) andr of 1/4 instead of 1/3. Both approaches have yieldedightly higher than 0.2 W m1 K1 at 100 K. MoreTheother aaccordbut alshas beices Ihit seemclose rand Icthe thrbeen fof thedent fratomsthe temhydrogto icesthe oxy67, bupressuIc hasteringindicaas is gconsidLDA athe plofor a dis notslope.not a spressuIt isdifferglassyambienamicadditioof watis a liq3.2. HToIt incrdensityp > 0.5on hea160waterdistanctortingThis cpressuto chanl conductivity of LDA is distinguished from that ofphous solids in one more aspect. Not only it varieso, Tx with x being close to 0.6, as for a crystal, decreases with increase in pressure. This featurebserved for a very few crystalline phases and forIc, but not for an amorphous solid. At first sight,at the mechanism that determines of LDA has aon with the mechanism that determines of ices Ihticularly in view of the fact that the density [28] oflids is the same, and their specific heat [78,79] hasto differ only by a small amount, if any. The effecton- or hydrogen-atom disorder on has been evi-he observed increase by 20% in when hydrogene Ih become partially ordered to produce ice XI, butture dependence remains approximately the same ontom ordering, as seen in Fig. 3. In going from LDAr Ic, the hydrogen-atom disorder is maintained whileatoms become ordered and increases by a factor ofqualitative dependence of upon the temperature ores not change. Such a similarity between LDA and icebeen observed from inelastic incoherent neutron scat-1] and inelastic X-ray scattering [82] studies, whichhonon propagation up to unusually high frequencies,ally found for crystals. It also seems significant tore whether or not there is continuity between ofhat of liquid water in a temperature plane. In Fig. 3, of liquid water against temperature shows that, asered solid, it has a positive slope, and that this plotinuous with the plot for LDA, which has a negativeseems to be a further demonstration that LDA isthat would be obtained by cooling water at ambientady known that thermodynamic properties of LDAthose of vapour-deposited amorphous solid andr [83]. The lack of continuity between of LDA andter in a temperature plane and the known thermody-rences between LDA and glassy water [83] providevidence against the validity of the two-liquid model4], which had assumed that ambient pressure waterounterpart of solid LDA.apacity of waters high-density amorphplete this section we now discuss the Cp of HDA.with increase in the pressure, which increases theis is particularly significant because HDA at a fixedhas also been found to become gradually denserfrom 77 K, and then to explosively crystallize athen p is 1 GPa [74,84]. Increase in pressure onorphous solids may cause both the oxygenoxygennd the OHO angles in its structure to change by dis-structure differently in different parts of bulk phase.e would differ from the manner in which increase inuses the distances and angles in the crystalline icesithout distorting the structure. It is generally knownG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 21that Cp of a solid decreases with increase in the density and,although it has not been measured for the high pressure formsof the ices, experiments have shown that the frequency of trans-lational lattice vibrations of all high pressure ices, including iceVI (at pressures of the high-density amorphous ice formation)increases as the density increases with increase in the pressure[85]. The origin for the observed decrease in Cp with increasein pressure and density has consequences for the recent findingthat some of the vibrational features of HDA, as studied by neu-tron scattering, are crystal-like [86]. In relevance to the structureof HDA, these findings need to be interpreted together with theabove-mentioned changes observed in Cp of the ices.The measured of the HDA at 130 K is plotted againstthe pressure in Fig. 4(A). The plot shows an asymptoticincrease with pressure, which is well described by the equa-tion: = 0.5856 + 0.1708p 0.0781p2. The anharmonic part ofCp that is associated with the thermal expansion is givenby the relation, Cp Cv = TV2/, where V is the molarvolume, the volume thermal expansion coefficient and is the isothermal compressibility. The quantity Cp Cv hasbeen estimated as 0.025 J mol1 K1 at 100 K [87]. By usingV = 15.4 106 m3 mol1 [27], = 0.10 GPa1 [32], and ofFig. 4. (A) Tyat 130 K agaidensity amorpnotations refehad been prepthe Debye moRef. [35].HDA equal to that for ice Ih at 100 K ( = 3 105 K1) [88],we determine the term TV2/ as 0.015 J mol1 K1, whichseems negl 1 1[35].The Cp oplotted in Fically as thBut its valpressures bume, doesin Cp withof (Cp/pwith increathermal expmal expans(Cp/p)Tcates that thpositive vain p increastudy, we hGruneisenIh [89].The DebCv vary wittemperatursound velopressure tomass of theThe Debye0.01 MPa tin Cv by Debye tem[90]. Sincean equallyterm is negwith the cand analytiphodece inhe thenc4BT3v2hre vd x =d antend. Its4(Aquaeasementh wipical plots of the thermal conductivity of the amorph measurednst the pressure. (B) The specific heat per mole of waters high-h (HDA) measured at 130 K is plotted against the pressure. Ther to different runs in which each time a new sample of the amorphared and studied. Smooth curves show the calculations based ondel for thermal conductivity and heat capacity. Data are taken fromof thefor theincreasof Cp.In tdepend = k2whetime anWe use[86] exEq. (3)in Fig.plot isof increxperistrengtigible in comparison with its Cp of 23 J mol Kf HDA at 130 K measured as a function of pressure isig. 4(B). The plot shows that Cp decreases asymptot-e volume decreases when the pressure is increased.ue is not expected to become constant at very highecause compressibility, a measure of decrease in vol-not become zero at high pressures. The decreaseincrease in pressure for solids, or a negative value)T , indicates that the phonon frequency increasesse in pressure. It is thermodynamically related to theansion by (Cp/p)T = T(2V/T2)p. Since the ther-ion coefficient, = V1(V/T)p, a negative value ofindicates that increases as T increases. This indi-e -determining Gruneisen parameter would have alue and, consequently, a decrease in V with increaseses the frequency of the phonon modes. In a recentave deduced that the average for the low-frequencyparameters is positive for HDA but negative for iceye model has also been used to estimate how andh the pressure. This is done by calculating the Debyee, D = vh/kB(62/M)1/3 where v is the averagecity, which increases from 2300 m s1 at atmospheric2600 m s1 at 1 GPa pressure [32,35], and M is thevibrating unit, taken as the molecular weight [35].temperature was found to increase from 230 K ato 271 K at 1 GPa, which corresponds to a decrease1.1 J mol1 K1. For comparison, we note that theperature of ice Ih is 220 K at atmospheric pressurethis decrease is 2% less than the decrease in Cp forlarge pressure increase [35], and since the TV2/ligibly small, the measured decrease in Cp agreesalculated decrease in Cv, within the measurementcal errors. This shows that increase in the frequencynon modes with increase in pressure can accountrease in Cp of HDA. Hence, we conclude that anpressure does not greatly affect the anharmonic parteoretical details of the Debye model for the frequencye of phonon scattering, Eq. (2) for is written as [63]3 D/T0(x) x4ex(ex 1)2 dx (3)is the phonon velocity, (x) the resultant relaxationh/kBT with being the phonon angular frequency.earlier finding of phonon-like excitations in the HDAing up to high frequencies and evaluated by usingvalue, which is plotted against p as a continuous line), is significantly less than the measured , but thelitatively similar to that of the measured . The rate(d/dp) is however much smaller than that observedally, which may indicate a change in the scatteringth increase in pressure.22 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443The results for HDA at 130 K, shown in Fig. 4(A) and (B),refer to the state of a solid produced by a much slower isothermalpressurizatHDA in alstate obtainpropertiesdifferent. Fwould be hpressurizata large effeinto accoundescribingFig. 4(A) awith pressuof HDA shof HDA prthe pressurless pronoudenser formform.On thewhen the ithe pressurDebye enerphonons caGPa, and anot quantitmation of adescriptionscattering sinduced depressure-de4. Relaxatto ultravisConsequor a liquidmined by cscanning cwhich is tathe heat florate, or ocshape rise.of the ultramaterials, wthe endothdures haveare truly amunable totalline solian earlier pmetal-alloymolecularwe first desduced, andtemperatur4.1. Preparation of waters amorphous solid and glassystateswelous sin diaterew m100th anter gglasquidinedis it ofufficd, thvacunitroplatfter aed toype-buied bosolube aryoster irsoniept aottoo theoplee hig3 mceduirreged inanded aparas man cylniaxto 130ed iperhe Htempbly bDA130Pa [ore mion of ice Ih at 130 K than has been used to producel other studies, and it is therefore denser than theed by pressurizing at 77 K. As a consequence, thesuch as , Cp, v and D would also be somewhator example, its Debye temperature and sound velocityigher than that of HDA produced by the usually rapidion at 77 K. The difference in density itself could havect on some properties, and must therefore be takent in a discussion of HDAs properties by appropriatelythe samples preparation and history. As shown innd (B), and Cp of the sample do not vary stronglyre (and density). However, the ultimate dense formould have a lower Cp and higher than the samplesoduced by rapid pressurization at 77 K. Moreover,e-induced changes in and Cp would be slightlynced as the compressibility would be smaller for thes of HDA and lowest for the ultimate or the densestbasis of the preceding analysis, we conclude thatnternal energy of HDA is increased by increasinge, the frequency of phonon modes increases and thegy and the anharmonicity decrease. While the Debyen explain the decrease in Cp of HDA by 5% perlso the increase in towards a plateau value, they doatively describe the increase in within the approxi-constant phonon scattering strength. Its quantitativewould require that a term for pressure-dependenttrength, which would be consistent with the pressure-crease of the anharmonicity, and/or a term for otherpendent processes be included.ion properties of the amorph and conversioncous waterences of molecular self-diffusion that allows a solidto explore different configurations are usually deter-alorimetry and dielectric spectroscopy. Differentialalorimetry yields a glass transition temperature, Tg,ken as either the onset of the sigmoid-shape rise inw signal (or equivalently Cp) for a given heatingcasionally as the mid-temperature of the sigmoid-Dielectric spectroscopy yields the relaxation timeviscous liquid formed on heating the glassy state. Forhose vitreous state rapidly crystallizes on heating,erm is not clearly observed and alternative proce-been used for determining whether or not theyorphous, particularly when diffraction methods aredistinguish an amorphous solid from a microcrys-d. Discussion of this subject has been provided inaper in relevance to the majority of hyperquencheds, glassy water, and amorphous solid state of lowweight hydrocarbons [91]. In the following sectionscribe how the various amorphous solid waters are pro-then how their relaxational properties change withe.It isa vitreparedsolid won a fbelowtact widiameof thewith limaintaa flaskamounAfter sreleaseof theliquidcopperthereainsulatbath. Ha homeproducan aerglass tuum cthat ena supeplate ka flat bopen tThe drat a ratlike, 2the proratherappearsampledescribPreHDA ia pistothen u77 Kbelowproducthe temtion. Ta lowassemgen. Lof 1150.07 Gtwo ml known that water does not supercool easily to formolid. Therefore its amorphous solids have been pre-fferent ways. Briefly, vapour-deposited amorphous(ASW) is made by slow deposition of water vapoursillimeter thick copper plate kept at a temperatureK. This copper plate is held by screws in close con-other copper plate that forms the bottom of a 68 cmlass tube containing liquid nitrogen. The other ends tube is kept open to atmosphere for replenishingnitrogen. In this home-made assembly, vacuum isand the path of vapours from water contained innterrupted by a baffle. Thus only a fraction of thevapours is allowed to deposit on the copper plate.ient deposition in a period of 34 h, the vacuum ise glass tube containing liquid nitrogen is taken outum assembly, and quickly immersed in a bath ofgen. The ASW sample is then dislodged from thee into a liquid nitrogen bath by using a scraper, andll handling of the sample is done by using thermallyols with the sample immersed in the liquid nitrogenrquenched glassy water (HGW) is also made by usinglt assembly. In its preparation, pure water droplets arey means of an ultrasonic nebulizer and dispersed asin nitrogen gas. The aerosol is confined to a largend the droplets are then allowed to enter a high vac-tat through a 200 or 300m aperture. The dropletsnto the vacuum assembly through this aperture gainc speed and hit a 35 mm diameter 4 mm thick coppert 77 K. This copper plate is mounted at the end ofm tube containing liquid nitrogen whose one end isatmosphere for replenishing it with liquid nitrogen.ts thus splat on the surface of the copper plate and coolher than 105 K s1. After several hours, a porcelain-m thick layer of HGW is thus obtained. Details ofre are given in Refs. [2325] and micrographs of theular shape of the splat-cooled droplets of HGW haveRefs. [92,93]. The procedures for recovering HGWits handling in a liquid nitrogen bath are the same asbove for ASW.tion and recovery of HDA and LDA are less by first freezing water at ambient pressure insideinder assembly kept immersed in liquid nitrogen andially compressing the ice Ih formed isothermally at1.5 GPa, or else (nonisothermally) at temperaturesK to 1.5 GPa. As mentioned earlier here the HDAsn this manner differ in properties depending uponature pressure and time profiles used in the prepara-DA sample is depressurized to ambient pressure aterature and then extracted from the piston-cylindery pushing it out of the cylinder into liquid nitro-is produced by heating HDA to a temperature rangeK at ambient pressure [28] or at a pressure below34,50]. Waters amorphous solids are produced byethods: one is by irradiation of ice Ih at low tem-G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 23peratures by high energy photons or other particles [9496] andsecond is by decompression of ice VII to ambient pressure at77 K and h4.2. CalorAmbienhave shownon isothermcussed earldecreases aand isolateout in thebecome enin the poresperatures bup to 273additional fified by thused as pureduction ofirst heatedface area bthe pores dAnnealingperatures cbecome moform hydroatoms. Thidensifies a vtive temperequilibriumties. Anneaalmost entiits Tf withUntil 1endothermthe sampleat a highzling how[99] couldby 35 J mstudies shoping exothheating scaexotherm whigh tempetallizationand HGW,on structurcrystallizeit was dedularge exothsoftening ean anneal-a[25,26] andprocedure,for a period long enough to allow its structural relaxation and tolose its relatively high enthalpy, then cooled to a lower temper-nd fiy remallTg eTgd ofstateings forhous101]etailHDaturclicsed. Atiesffectss, tcansin1ns fooadalpyral rW irmiclizatlizatr ASW. Ilousndehentheirmpen reheir9 1creaol1lly by lar[105eter.ransentradystates shoeir coa mirecaHGWd, theating to T > 120 K, which produces LDA [97].imetric behaviour, relaxation and glass-softeningt pressure differential scanning calorimetry studiesthat the enthalpy of both ASW and HGW decreasesal annealing in vacuum for several hours, as dis-ier [26]. In the case of ASW the surface area rapidlynd as sintering proceeds any pores in it become closedd from each other. When the initial heating is carriedpresence of N2 or other gases, the adsorbed gasesclosed in the pores during sintering. Once enclosedthe gases cannot be removed by pumping at low tem-ut are released when the sample is gradually warmedK. This release of gases causes the appearance ofeatures in the DSC scans which become further mod-e different thermal conductivities of the gas that isrge gas. To minimize the exothermic effects due tof surface area on initial heating, ASW samples arein vacuum up to 113 K, which decreases the sur-y several hundred m2 g1 and no gas is enclosed inuring the sintering of the micro-porous solid [26].at higher temperature or slow heating to higher tem-auses weakly hydrogen-bonded water molecules tobile and any dangling OH groups in the structure togen bonds with the neighbouring molecules oxygens occurs in addition to the structural relaxation thatapour-deposited amorphous solid and reduces its fic-ature, Tf, the temperature at which the glass and theliquid would have the same thermodynamic proper-ling or slow heating of HGW decreases its enthalpyrely as a result of structural relaxation and reducestime.987, all attempts to detect the glass-softeningfor ASW and HGW had been unsuccessful, becauses rapidly crystallized to ice Ic even when heatedrate in a DSC experiment [98], and it was puz-use of adiabatic calorimetry in an earlier studyhave shown an (endothermic) rise in Cp of ASWol1 K1 at a temperature of 135 K. Further DSCwed [26,98,100] that there were two partly overlap-erms in the differential scanning calorimetry (DSC)ns for both ASW and HGW. The low temperatureas due to structural relaxation of the solid and therature and deeper exotherm was due to the rapid crys-of the ultraviscous water, formed on heating ASWto ice Ic. It appeared that part of the heat releasedal relaxation could cause ASW and HGW to rapidlyin the glass-softening endotherm region. Therefore,ced that a procedure that eliminated the relativelyerm during the scanning could help reveal the glass-ndotherm in a DSC scan. This led to development ofnd-scan procedure for determiningTg of ASW, HGWhyperquenched metal alloy glasses [91,101]. In thisthe sample is first annealed at a certain temperatureature aenthalpgibly sof theminingmethoreousto showwell aamorpdo so [A dHGW,temperbut cydiscusproperheat eertheleDSC s30 K mthe scatial, brof enthstructufor HGexothecrystalcrystalthan foof HGanomais not uwere tbelowto a teand theshow tand 12The in2.0 J munusuationallearlierparamglass-trate ofis alreglassystanceand thshowsWeASW,tory annally heated to obtain its DSC scan. As a result, thelease on structural relaxation during heating is negli-, and therefore does not interfere with the appearancendotherm. The procedure has been useful for deter-of materials that cannot be vitrified by the normalcooling at a rate of up to 100 K s1, and whose vit-once produced rapidly crystallizes on heating priora Tg endotherm in a normal DSC scan [91,101], asdetermining whether a solid is microcrystalline or, particularly when diffraction methods are unable to.ed discussion of the entropy and enthalpy of ASW,A and LDA has appeared in Ref. [102], where thee dependence of their enthalpy and the irreversiblethermodynamic paths and crystallization have beens mentioned above, a quantitative comparison of theof ASW, HGW and LDA must take into account thes resulting from their spontaneous relaxation. Nev-o describe some of these features briefly here, theof ASW, HGW and LDA obtained by heating atrate are shown in Fig. 5(A). The top two plots showr the as-made samples of ASW and HGW. The ini-exotherm observed for ASW is due to the releaseon reduction in the surface area as well as due toelaxation. This exotherm is much less pronouncedn which only enthalpy relaxation occurs. The deepfeature with a minimum at T near 160 K is due toion of ASW and of HGW to ice Ic. After the onset ofion, the exotherm for HGW has a slightly lesser slopeW, as if there are two steps involved in crystallizationt has been speculated that this may be related to thebehaviour of supercooled liquid water, but its reasonrstood [26,103]. The samples discussed in Ref. [102]structurally relaxed by annealing at a temperaturecrystallization onset temperature or else by heatingrature just below their crystallization onset, coolingheating to obtain a DSC scan. The arrows in the plotsTg. Its value is 136 1 K for ASW and HGW [26]K for LDA for the heating rate of 30 K min1 [104].se in Cp is 1.6 J mol1 K1 for ASW and HGW andK1 for LDA [104]. This increase occurs over anroad temperature range, which indicates an excep-ge distribution of relaxation times and, as discussed,106], a low value for the relaxations nonlinearityMoreover, the relatively small increase in Cp in theition range has been interpreted in terms of a smallopy change of a state whose configurational entropysmall [107]. Further studies of the hyperquencheds of dilute solutions of inorganic and organic sub-wed that Tg varies with both the nature of impuritiesncentration [108,109] in the glassy state and that Tgnimum value for a certain concentration.ll that like all physical properties, the enthalpy of, HDA and also of LDA depends on the sample his-erefore, is not likely to be the same for samples of24 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443Fig. 5. (A) Dcrystallizationhyperquencheendotherm is mdecrease in thThe bottom thby the arrow.of ASW, HG129 1 K forHGW and 2 Jtion (two arrotemperature ais described inmeasured magIc transformathe enthalpy idifferent thHGW, theto ices Ih orelaxes toslightly belformed onto a lower-1 GPa towhich the eon the heatperature foand Ic is analogous to increasing the cooling rate for producingssy state at a fixed pressure. Nevertheless, we have showniatiothetivebeenhs, aandAlsoAs gIc hapathrmathe glathe varIc overrespecTg hasamorpices Ihcated.the LDto icecyclictransfoSC scans showing the glassliquid transition endotherms andexotherms for vapour-deposited amorphous solid water (ASW),d glassy water (HGW) and low-density amorph (LDA). The Tgasked by the heat released on enthalpy relaxation and surface areae top two curves, which also show sharp crystallization exotherms.ree show the glassliquid transition endotherms with Tg marked(B) An illustration for the temperature dependence of the enthalpyW and LDA. The Tg s are 136 1 K for ASW and HGW andLDA, and the increase in Cp is 1.6 J mol1 K1 for ASW andmol1 K1 for LDA. Arrows show the direction of transforma-ws opposite in direction denote the reversibility of the path). Thend pressure conditions are marked, and the notation for the pathsthe text. The plots for the enthalpy are drawn such as to show thenitude of qirrev for HDA LDA, LDA ice Ic and HGW icetions. The vertical axis has no scale because the absolute value ofs not known. The data are replots from Figs. 2 and 3 of Ref. [102].ermal and pressure histories. In the cases of ASW andrapid cooling rate necessary to avoid crystallizationr Ic inevitably produces a high-enthalpy state, whicha lower-enthalpy disordered state on annealing at Tow Tg. In the case of HDA, a high-enthalpy state ispressurization of ices Ih or Ic at 77 K, and it relaxesenthalpy state on heating at a fixed pressure of e.g.T above 130 K. Consequently, the exact manner innthalpy varies with T on heating at 1 GPa dependsing rate. In terms of the enthalpy, lowering of the tem-r producing HDA by isothermally pressurizing ices Ihand (iii) froIt shoulHGW (andand these cnetwork stconvert onstructure oof ultraviscon coolinginfer that tnot necessaof LDA anthe thermadependencphous soliddielectric rStudies129 K to for differenhigher thanHGW [112fusion kineLDA havedissolve infurther by vrate of orieknown to ocontainingproton-defitures, havecontainingcontainingremains atstructure inincrease thif orientatito the increLDA struct129 K, thetion of NHhydrogen-bdielectric rTg. The TgThus not on of the enthalpy of HDA, LDA, ASW, HGW and icetemperature range of interest in Fig. 5(B), where theTgs have been indicated. Here, the change in slope atmade consistent with the relative change in Cp of thend the manners of irreversible transformation fromIc to HDA on compression to 1 GPa has been indi-, the irreversible path for HDAs conversion to LDA,lassliquid transition and finally the crystallizationve all been indicated by arrows. It is shown that thefor ice Ic via HDA and LDA has three irreversibletions, (i) from ice Ic to HDA, (ii) from HDA to LDAm the high temperature state of LDA to ice Ic.d be stressed that in Fig. 5B, the paths of LDA andASW) follow different enthalpytemperature curves,urves do not meet. This means that the hydrogen bondructure of LDA at T above its Tg of 129 K does notheating or cooling to the hydrogen bond networkf HGW. Also, the hydrogen bond network structureous water formed on heating HGW does not convertto the hydrogen bond network structure of LDA. Wehere is a difference between the short-range order,rily limited to first near neighbours in the structuresd HGW. Similar differences have been indicated byl conductivity data of LDA [50], whose temperaturee is more like that of a crystalline than of an amor-. More recently this difference is also found fromelaxation measurements of LDA [110].of deuterated LDA have shown that Tg increases from133 K on deuteration, an increase of 4 K, but 35 Ktly prepared samples [111], which is significantlythe increase of 1 K observed for the deuterated]. This also indicates the difference between the dif-tics of LDA from HGW. Further detailed studies ofbeen performed by adding different impurities thatice Ih and are incorporated in its lattice structure, andarying the impurity components such as to alter thentational diffusion of water molecules in LDA, as isccur in ice Ih [111]. Briefly, LDA made from ice IhHF, NH3 and NH4F, which are expected to producecient and proton-rich topologically disordered struc-shown that the glass-softening endotherm for LDAHF becomes too broad to yield its Tg. The Tg of LDANH3 decreases to 127 K and of that containing NH4F129 K. Since NH3 in the hydrogen-bonded networkcreases the number of protons, and is hence seen toe dielectric relaxation time [113], it is expected thatonal diffusion of H2O molecules were to contributease in the number of configurations accessible to theure, Tg of the NH3-doped LDA would be higher thanvalue for pure LDA. For the same reason, introduc-4F is seen not to introduce any extra protons in theonded structure and therefore should not effect theelaxation time or, for the reason given above, LDAsof NH4F-doped LDA does remains at 129 K [111].nly the thermodynamic vibrational and relaxationalG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 25properties of LDA differ from those of ASW and HGW, butalso the effects of deuteration and impurities on their dynamicsdiffer.4.3. Dielecstates of waInitiallyon ASW [relaxationin the samthat the resurementscrystallizattemperaturallow meascrystallizatbeen studietime [116,1the Avramilization is 5an uncertaithe exponea temperatuthe ASWsMore retangent, tabeen analyvalue of thbe estimatefrequenciesthe dielectr2, is giventan meas(tan meas(where tanerature-depto tan measparameterquencies is the diel, tan backues measucondition atan -relaxadiel by = (diwhere imaxdiel =maxdiel =The dielectEq. (6), as described in detail earlier [119]. The dielectric relax-ation time of ASW has thus been estimated as 23 s at 140 Kat ofstimperscourienlesthaour onal aedomaterion timeer bystimshedscouK montren dty aatureasus mre toorphtheandeasuofhizedpeaks incmpeeasee prodicacantlnotrmeonfie betinedagaslowum0 andslore-amf thato 13ionTheto beaftertric relaxation of amorphous solid and glassyterdielectric loss factor measurements were performed114] and HGW [115] in order to investigate if thetime of its ultraviscous state could be determinede manner as for other glasses. But it was observedlatively slow rate of heating needed for such mea-always crystallized the sample to ice Ic and thision also yielded a peak in the loss factor againste plots. Although dielectric measurements did noturement of their crystallization kinetics to ice Ic, theion kinetics of ASW made from H2O and D2O hasd by both calorimetry and FTIR spectroscopy in real17]. It has been found that crystallization followsequation and the activation energy for the crystal-1 kJ mol1 for H2O and 66 kJ mol1 for D2O withnty of 10 kJ mol1. It has also been suggested thatnt of the Avrami equation can be seen also to indicatere-dependent activation energy or barrier height tocrystallization to ice Ic [116,117].cently, the measured change in the dielectric lossn , of both ASW and HGW with temperature hassed in detail, and it has been shown that approximatee dielectric relaxation time of ultraviscous water cand from an analysis of the data available at only two[118,119]. In this analysis, the measured value ofic loss tangent, tan meas at two frequencies, 1 andby1) tan background2) tan background =(21)(4)background is the frequency-independent but temp-endent value of the dielectric loss which contributes, the asymmetric distribution of relaxation timeof the DavidsonCole equation [120] and the fre-1 and 2 are much greater than 1/diel, where dielectric relaxation time. For a reasonable choice ofground can be determined from the tan meas val-red at two frequencies. For the same 2 2 1t a fixed temperature, the dielectric loss, , andtion() = [tan meas() tan background] are related toel); diel = 1( tan -relaxation)1/(5)s the limiting high frequency permittivity. Since,tan[2( + 1)], (6)1, in this distribution of relaxation times [120,121].ric relaxation time, diel, therefore determined fromand thboth ethe temultraviReomolecurequireneighbrotatioand prcous wrelaxatation tto diffgiven eestabliultravithe 30In chas bemittivitempersuch mple wapressuthe amsure on1 GPawas mspectraamorpthe ature iAt a teto decrand thalso insignifiments,state foTo cferencdetermplottedsmall,maximT < 15that onpressuview o138 Ktallizatagain.found132 KHGW as 35 s, with a factor of two uncertainty inates [119]. This analysis, however, did not indicateature dependence of the dielectric relaxation time ofs water.tational and translational diffusion motions of H2Oin liquid water with tetrahedral hydrogen bondingt hydrogen bond break and then reform with a newr the same neighbour. Therefore, thermally activatednd translational motions of molecules in disorderedinantly network structures, such as those of ultravis-and molten SiO2, occur together. Since the dielectricime, the self-diffusion time, and the structural relax-estimated from the Tg endotherm are usually foundabout one order of magnitude or less, the aboveate of the dielectric relaxation time of 30 s clearlythat heating ASW and HGW to 140 K produces ans state of water at 140 K, and their Tgs are 136 K forin1 heating rate.ast, the dielectric relaxation time of HDA and LDAetermined directly by measuring the dielectric per-nd loss, and , spectra, as their states at highes seem to be stable for long enough time to allowrements [53,54,110]. For this study, the HDA sam-ade by amorphizing ice Ih at 130 K by raising the1.2 GPa at a rate of 0.1 GPa h1, and then keepingized state near 1 GPa and 130 K for 1 h. The pres-amorph at 130 K was slowly decreased from 1.2 tothereafter its dielectric relaxation spectra at 130 Kred in the 0.01 Hz to 1 MHz range [53,54]. Typicalthe dielectric loss, , for the pure and KOH-doped-ice at 1 GPa are shown in Fig. 6(A). These show thatmoves to higher frequencies when HDAs temper-reased, showing that the relaxation becomes faster.rature near 150 K, the height of the peak begins, indicating that the sample has begun to crystallizecess becomes faster on heating above 150 K. Thistes that the solid formed on crystallization does noty contribute to in the frequency range of measure-necessarily that the orientation polarization of thed is less.rm that crystallization occurred, the temperature dif-ween the sample and the Teflon vessel, T was alsosimultaneously in the same experiment, and it isinst the samples temperature in Fig. 6(B). It shows aly growing exothermic effect which reaches a localat 150 K. Thus slow crystallization had begun atit accelerated at T > 150 K. It has been found alreadyw heating at a pressure of 1 GPa, the fully denseorphized state crystallizes at T near 165 K [84]. Int, the temperature of one sample was decreased from2 K, which is well below the above-mentioned crys-onset temperature and the spectra were measuredspectra measured at 132 K after this cooling wasidentical to the spectra that had been measured atheating the HDA sample to 132 K, confirming that26 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443Fig. 6. (A) TKOH-doped aindicated, andof relaxationsample and th1.7 104 mfrom the datataken from Rethe spectraconclude threlaxed staAs showmetrical diCole and Cwhere (=istic dielecthas been fFor a detacalculatedspectrum,temperaturand KOH-dhas no dischas a veryThese findithe relaxatiaddition ofThe temArrheniusis shown bat 130 K andiel of HGhe reis thInseramot leade liquolingandshohe dielectric loss spectra for pure amorph (open symbols) andFig. 7. TThe line[53,54].in whichpath thafrom thsuperco140 Kicantlymorph (symbols filled with a cross) at 1 GPa at the temperaturesan analysis in terms of the ColeCole symmetrical distributiontimes (dashed line). (B) The temperature difference between thee sample cell T, as observed on heating of the pure HDA andole fraction KOH-doped HDA. A baseline has been subtracted, and the curves have been shifted vertically for clarity. Data arefs. [53,54].did not change on thermal cycling. Therefore, weat the HDA sample is thermodynamically in the fullyte already at T close to 130 K.n in Fig. 6(A), the spectra are described by a sym-stribution of relaxation times function given by theole [122]: *() = + (0 )/[1+(idiel)1],2f) is the angular frequency and diel is the character-ric relaxation time. (Note that this type of distributionound for water also in the range 273323 K [123].iled discussion see Ref. [54].) The diel value wasfrom the reciprocal of the peak frequency in the i.e., diel = (2fpeak)1, and it is plotted against thee in Fig. 7. At 150 K, diel is 30 5 ms for both pureoped water, which shows that this amount of dopingernible effect on diel of HDA at 1 GPa, although itlarge effect on the relaxation of time of ice Ih [124].ngs are consistent with the general observation thaton time of liquid water is not greatly affected by thesmall amounts of electrolytes.perature dependence of diel is described well by therelation, i.e., log10(diel) varies linearly with 1/T, asy the solid line in Fig. 7. The relaxation time is 5 sd 30 ms at 150 K, which is surprisingly shorter thatW and ASW, which has been estimated as 30 s atthe calorimThese resuultraviscouIn this cbeen foundof 10, despof HDA wof LDA isknown thatwith a fullyatom distriof a structua comparisice rules aroxygen atogen bond.)relative to tbond structHDA mayrules are noture dependof its structto each oxyhydrogen bthe crystallthe decreasHDA to LDof hydrogeices.We concthat it becolaxation time of pure HDA at 1 GPa plotted against the temperature.e best fit of the Arrhenius equation. Data are taken from Refs.t is the phase diagram of the ices. The temperaturepressure regionrphization studies of ice Ih have been performed is indicated. Thes to an ultraviscous liquid state is shown by arrows. The pathid water to ultraviscous state shown by the dashed line requiresat 1 GPa. This path has not been experimentally achieved.ambient pressure. Moreover, diel at 130 K is signif-rter than 100 s to 1 ks, a value normally taken to beetric relaxation time at Tg of glass-forming liquids.lts therefore indicate that HDA at 1 GPa becomes ans liquid well before it crystallizes on heating to 150 K.ontext, it is noteworthy that diel of LDA [110] hasto be longer than that of HDA by more than a factorite the fact that its density is 25% less than thathich should decrease diel. More significantly dielin between that of HDA and ice Ic [110]. Now, it isamong the different states of water, diel of the statebonded hydrogen-bonded structure, with hydrogenbuted according to the ice rules is longer than dielre that does not obey the ice rules, as is known fromon of diel of ice Ih and that of liquid water. (Thee that there are two hydrogen atoms adjacent to eachm, and there is only one hydrogen atom per hydro-By analogy, therefore, the higher diel value for LDAhat of HDA would indicate formation of a hydrogenure in LDA in which the ice rules are obeyed. Thusappear to be a densified state of water in which icet obeyed. It is possible that the crystal-like tempera-ence of thermal conductivity of LDA is a reflectionure in which there are two hydrogen atoms adjacentgen atom, and there is only one hydrogen atom perond, as is the case for thermal conductivity for alline ices whose value is plotted in Fig. 3. Therefore,e in the density on the irreversible transformation ofA is also accompanied by a change in the mannern bonding from that in water to that in the crystallinelude that diel of 30 5 ms for HDA at 150 K showsmes an ultraviscous high-density liquid on heating.G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 27It is in an equilibrium but metastable state, the one that wouldbe obtained by supercooling water at a pressure of 1 GPa acrossits phase bdiagram inmodynamiby heatingthe (low-dethe findingity with teminelastic netering dataTg-related pThus it doeto LDA. Tequilibriumuid states,be vitrifieddensities diIt shoulendothermin Fig. 5(Aof purpose[126,127] athe ageingproton-ordthe obtaineenthalpy inDSC scan slost on annor proton-osured gainHGW, unathermal cymally annechange in tas has beenin the DSCsoftening oit [93].There huse of DSCet al. [130]ice XII recoture in theirconcludedvations byices V anda criterion[29,104] cotheir Fig. 4and probabincrease in[126] studyFig. 9 alsoarea with thHanda et ain the peakTo resolve this aspect further, it seems necessary to determinethe relaxation time of LDA at ambient pressure by dielectric ormethce Sastudis wl-dop-ordscopfor icof thviewA [2re haA anavey meundd inat Tis themeabsA safewed bytiononsforThenic sexpf cohad135eculbe oe intproionicatlizatocesse, thlizatA. Tlar thilein thmalnothandlaint brewhooursangeoundary with ice VI, which is shown in the phaseFig. 1, and discussed later here. Since LDA is not ther-cally connected to the (low-density) water obtainedASW and HGW [83], it seems that it does not formnsity) water on heating. This is also evident fromof the crystal-like variation of its thermal conductiv-perature as shown in Fig. 3, from the analysis of theutron scattering data [80,81] and inelastic X-ray scat-[82] and from the thermodynamic, vibrational androperties [37,83,110112], as discussed earlier here.s not seem that the low-density liquid water is relatedhis puts into question the conjecture of HDALDAand the transition between their corresponding liq-in particular the conjecture that HDA and LDA maystates of two distinct liquid states of water whoseffer by 25% [44].d be noted that a DSC feature resembling the Tg-of the type observed for ASW, HGW and LDA shown) has also been observed in the calorimetric studiesly aged samples of ice Ic [107,125], ices V and VInd of tetrahydrofuran-ice clathrate [107,128]. Duringat a low temperature, these crystalline solids undergoering, which should be an exothermic process. Whend partially proton-ordered state is heated, it gainsa time-, and temperature-dependent manner and thehows a peak whose area is equivalent to the enthalpyealing. An estimate for the extent of orientational-rdering in the ices has been obtained from the mea-in enthalpy on heating. But in the case of ASW andnnealed samples have shown the Tg-endotherm oncling over a limited temperature range, and isother-aled samples for different periods have shown littlehe onset temperature of the endotherm or its height,discussed in Ref. [129]. So, the endotherm observedscan is unambiguously attributable to the glass-f ASW and HGW. A recent study seems to confirmas been a further development, which came from thescans in the calorimetric study of ice XII. Salzmannfound that both the unannealed ice XII and annealedvered at ambient pressure show an endothermic fea-DSC scans obtained by heating at 30 K min1. Theythat this is in line with Handa et al. [126] obser-Cp measurements of the unannealed and annealedVI, and therefore Salzmann et al. [130] used it asto speculate that similar features observed for LDAuld have been possibly made on ice XII. However,in Ref. [130] shows that the DSC scans peak heightly also the area do not show a systematic change withthe annealing time. They also repeated Handa et al.sfor ice V by DSC and the scans for ice V in theirshow no systematic change of the peak height ore annealing time. This is contrary to the findings byl. [126] who had found clear and systematic changesheight and area on annealing of ices V and VI [127].NMRSinearliervationof HCprotonspectrofoundsourceIn ourfor LDThein HD[133] hLDA bhave fodetecteice Icthe basenhancthat itsthe LDthan aobtaintranslain LDAits tranrange.harmoTheies is owhichlater of molwouldBut thshow atallizatany indcrystalless prthat cacrystalfor LDmolecuworthwfoundof therIn aFisherbe expby firsing asneighbshort-rods.lzmann et als [130] conclusion has a bearing on theies of LDA [29,104,131], another more recent obser-orth mentioning here. In a paper on proton-orderinged ice V, Salzmann et al. [132] have concluded thatering in pure ice V could not be confirmed by Ramany and . . .similar endothermic events were alsoe XII. This conclusion creates doubts regarding thee endothermic feature observed for ices V and XII., this restores the interpretation of the DSC feature9,104,131].ve also been attempts to observe molecular mobilityd LDA by neutron scattering methods. Koza et al.performed a detailed and elegant study of HDA andasuring the intensity at = 0 (see their Fig. 2). Theythat a pronounced drop-off in this intensity was notthe temperature range of crystallization of LDA to> 135 K, where they had expected to find its Tg. Onat such a drop-off of intensity is a fingerprint for annt in molecular mobility of the sample, they deducedence indicates that there is no molecular mobility inmple in the 135155 K range at time scales of lessnano-seconds. Further studies of the energy scansthe incoherent scattering technique confirmed thatdiffusion of molecules over intermolecular distancesa time scale of less than 4 ns does not occur duringmation to ice Ic on heating through the 130150 Ky also found that LDA as well as ice Ic behave asolids.eriment time scale in the neutron scattering stud-urse longer than in the DSC studies of 30 K min1,showed its Tg to be first 129 K [29,104,131], andK [130]. Therefore, one expects that the enhancementar mobility in LDA in the neutron scattering studiesbserved at a much lower temperature than 130 K.ensity at = 0 in Fig. 2 of Ref. [133] also does notnounced drop-off at lower temperatures. Since crys-is usually a thermally activated process, the lack ofion of molecular mobility in the temperature range ofion indicates that LDAs crystallization is a diffusion-s and thus unrelated to the viscosity of the material. Ine JohnsonMehlAvrami equation would not fit theion kinetics data of LDA, and this aspect can be testedhere seem to be difficulties in resolving the issue ofranslational diffusion in LDA and perhaps it may beto determine if the harmonic solid behaviour of LDAese studies [133] is consistent with LDAs coefficientexpansion in the 75135 K temperature aspect of the study, Johari [134] has argued thatDevlins [135] findings of isotopic exchange caned by a mechanism in which H2O clusters diffuseaking hydrogen bonds from the neighbours, diffus-le and then reforming hydrogen bonds with other. It is also uncertain whether or not the presence oforder arising from hydrogen-bonded OH4 tetrahedra28 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443has an effect on the interpretation of scattering measurements interms of molecular mobility [133]. This can be tested by study-ing SiO2 aorder, but oStudies23 mm-thrange [136support ofthis findingslice rathertures severaout that adTm/dP oftraversed thcase for LDthat Tg of mMore expethis issue.dielectric rFinally,of 1 0.1solid waterthe film athas concludphous solidand low-dedata [137],of water isthat PASW205 5 K.1020 mg[137] werequent largeinterpret. Hthat have bdoing so, wHGW andbrucker et aASW priortion of thathe variousunsinterednent Tg-enTg-endothefor detailsOne asan unusuastructural-ureported infor ascertaiBriefly, theTg and thmeasured f[137] is width of its structuraby heatingtemperature sensitivity of the relaxation time with increase intemperature may make the endotherm somewhat broader for aheattribuhicetricreae is uith tate, irepibleof re asfor hrecaandp valcturaor that oASWwasamcaselainethe vef. [1e ancryissu[137bsertureder gituredof stlid oicethanen tue. MnedlueneSWtallizt in ocien. Theit isceptidothernta fods vesnd GeO4 glasses which have a similar short-rangef SiO4 and GeO4 tetrahedra.have shown that a blunt indentor penetrates throughick sample of LDA on heating through the 142 K], and therefore LDA deforms by viscous flow. Inan alternative conclusion, it has been suggested thatis an effect described as Indeed a tungsten wire canrapidly through crystalline ice samples, at tempera-l degrees below bulk melting point [137]. We pointwire under a load traverses through ice Ih becauseice Ih is negative. The ice refreezes after the wire hasrough it, thus leaving the ice unsliced. This is not theA, which has no melting point, and it is well knownaterials usually increases with increase in pressure.riments and/or interpretation are needed to resolveThis is particularly important in view of the recentelaxation time measurements of LDA [110].we consider a recent ultrafast microcalorimetry studym thick film of vapour-deposited, porous amorphous(PASW) by calorimetric scans obtained by heatinga rate of (1.3 0.2) 10 5 K s1 [137]. The studyed that most of the earlier findings on sintered amor-water (ASW), hyperquenched glassy water (HGW)nsity amorph (LDA) are inconsistent with the newand consistent with a disputed conjecture that the Tg165 K and cannot be measured. It was also reporteds enthalpy relaxation time is greater than 10 5 s atWe note that, in contrast to earlier studies in whichsamples were used, experiments on PSAW in Ref.performed on 12m thick films and the conse-surface effects in the porous samples are difficult toowever, there are certain basic aspects of the studyeen overlooked. We briefly describe these here. Ine, unlike in Ref. [137], do not use the terms ASW,LDA interchangeably. As discussed above, Hall-l. [26] had already shown the importance of sinteringto obtaining the DSC scans. In the Introduction sec-t paper they reviewed the earlier studies in whichtime-dependent thermal effects observed on heatingASW or PASW were either confused with a promi-dotherm, or else taken for a lack of observation of arm in the DSC scans. This paper may be consulted[26].pect that has been overlooked in Ref. [137] islly large increase in the temperature-width of thenfreezing endotherm (not the Tg endotherm asRef. [137]) for toluene, which was used as a test liquidning the merits of the ultrafast calorimetry technique.width of the endotherm, taken from its designatede peak position, for vapour-deposited porous tolueneor heating at 1.3 10 5 K s1 rate in Fig. 3, Ref.30 K. We point out that this width is 10 times the3 K observed for bulk toluene [138] in the range ofl-unfreezing at normal Tg of 117.5 K, as determinedat 10 K min1. One realizes that a decrease in thehigherthe distures, wcalorimfold intoluenated welaborof 14 Ka possbutionthe sam140 KWeices Ictheir Cas struin Cp fthan thwhensamplePASWIn thatas expabove6 of RobscursampleThein be oconjecof watconjectinuityare vathat ofis lessgest, ththis issbroadeuid, tofor PAit cryscontenis suffiationeven ifthe exthe ento discthe rate as well as shift it to a higher temperature. Buttion of relaxation times is also less at higher tempera-h would make the width smaller. This is a well-knownc effect and has been discussed in Ref. [139]. The 10-se in the temperature-width of the Tg-endotherm fornusual and seems to indicate extrinsic effects associ-he ultrafast measurement technique used. To furtherf we use this factor to scale the corresponding widthorted for the Tg endotherm for ASW [26], and ignoredecrease in the width due to decrease in the distri-elaxation times, or assume the decrease to be aboutfor toluene, the expected width for PASW would beeating at 1.3 105 K s1 rate.ll that Cp of ASW is within 2% of the Cp values forIh up to a temperature of 125 K [99,140,141], andues are expected to remain close to each other as longl-unfreezing of ASW does not occur. The net increasee 136 K Tg-endotherm is found to be only 8% higherf ice at its maximum value reached at 147 K [141],was heated at 30 K min1. It would be less if thes not fully amorphous. This is important because theple was stated to be at least 50% amorphous [137]., the sample would show a very broad Tg-endotherm,d above, with Cp rise of significantly less than 8%alue for the ice. The noise in the data reported in Fig.37] is comparable with that, and therefore it wouldy slow increase of the wide Cp endotherm before thestallizes.e of the apparent absence of Cp-rise was considered] in order to infer that the expected Cp-rise is enoughvable. But in that consideration, interpolated paths ofCp in its plots against T in the supercooled regionven in Ref. [142] were used. The three Cp T pathsthere had been originally used for analysing the con-ates between HGW and ambient water [142] and theynly if the Cp of HGW is 2 J mol1 K1 higher thanIh at 153 K, as shown in Fig. 4 of Ref. [142]. If itthat, as Chonde et al.s data [137] apparently sug-he paths in that Fig. 4 are inappropriate for resolvingoreover, as the ultrafast microcalorimetry techniquethe structural-unfreezing endotherm for the test liq-, it would also broaden the corresponding endothermand thus further reduce the Cp-rise of PASW beforees. It has also been stated that [137], . . .the ASWur vapour-deposited ice films is at least 50%, whichtly high to observe endothermic heat capacity vari-amount of PASW in the sample is not known, buttaken to be 70%, the reduced fraction of PASW andonal broadening of the Tg-endotherm would makeermic signal smaller and hence even more difficultby ultrafast technique used. Above all, we find thatr the relaxation time of toluene used for testing thealidity were compared by mistakenly using at leasthigher values than those in the literature they hadG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 29Recent studies of the dielectric relaxation time of ASWand HGW have made it clear that their Tg is less than 140 K[118,119].conjecture[26] had fostudying itmay also bsintered AS5. Characpressure-aIt has becan be constate isothemay be rettemperaturof crystalsa shock wageologicaland Faheytalline silicmechanications for otof the propa shock prphous. In 1become amthere have bmechanicamill, in whHemley ettals at 300300 K, indiwork structhas been gX-ray andBragg peabroadeningerences toof materia[150].Duringarrangemenapplied precorrespondmoduli. Buon amorphsure collapenergy incthe molecumolecules iphous in thleads to fophases of iformed aftappear amoIn this context, it should be noted that recent studies of HDAand LDA have shown a considerable scattering of neutrons atanglehetescalin twas36],ty, huggeisteno soas oscoe ant 1.1inedg ofdoubthatnd thbe ca, buaxathasssurres hecent 1254].ollaptiesaturhis chen] wassureerfo]. Sif ice2, thlow1.3t thare, anic inte wre ofdiesclather sensitd toleasebeThere is no experimental evidence to support thethat ASWs Tg is 165 K. Since Hallbrucker et al.und it necessary to sinter the ASW sample prior tos DSC scans in order to reveal the Tg-endotherm, ite necessary to perform accurate measurements onW samples using ultrafast microcalorimetry.teristic changes duringmorphization of iceen known since the 1960s that a variety of materialsverted from their crystalline solids to an amorphousrmally by uniaxial compression and most of thoseained in the apparently amorphous state at very lowes at ambient pressure. The pressure-induced collapsemay result from natural or man-made impact, fromve generated by it, or from a sustained stress as inoccurrences. Historically, as early as 1963, Skinner[143] had reported that stishovite, a form of crys-a, becomes amorphous by the simple procedure ofl grinding, and others had reported similar observa-her crystals soon thereafter [144]. In his compilationerties of silica, Primak [145] had noted that underessure of more than 20 GPa, SiO2 becomes amor-972, Brixner [146] reported that Gd2(M2O4) crystalsorphous when subjected to pressure. Since 1981,een a number of systematic studies of shear-inducedl amorphization of metal crystals in a high-speed ballich crystals were subjected to high impact [147,148].al. [149] reported that -quartz and coesite crys-K transform to amorphous solids at 2535 GPa andcating the thermoelastic instability of tetrahedral net-ures at high compression. Their amorphous structureenerally deduced from the observation that theirneutron diffraction features lack sharp features orks, or their vibrational features show exceptionalof the peaks. A brief description with relevant ref-earlier studies of pressure-amorphization of a hostls is given in the Introduction of a recent articlethe course of the collapse of a crystal, topologicalt of atoms or molecules is forcibly altered by thessure, which in the case of ices Ih and Ic wouldto a value greater than the value of their Youngst, more significantly, while the density decreasesization of a crystal, the density increases on pres-se and subsequent amorphization, while the internalreases in both cases. When the resulting change inlar arrangement leads to random displacement ofn the structure, the collapsed solid may appear amor-e X-ray and neutron diffraction spectra. But when itrmation of nanometer size crystals of high-pressurece, with a very large net surface area, the producter the collapse has a higher density and would stillrphous in the X-rays and neutron diffraction spectra.smallto thescopicearlierwateret al. [diversihave sis consonly twknownas mesmediatstructupressuHDA adetermsurizinseemsknowntion, awould1 GPtric relVHDAent prepressumore rLDA a[153,1sure cpropertemperIn tered w[27,28ial prethose pal. [71tures oin SiOatively0.93 tothoughpressudynamclathrapressuBut sturan iceA furthigh-drevertewas recannots, i.e., at low q values [36] that has been attributedrogeneous disordered structures of HDA at a meso-e. This is similar to a feature that had been observedhe neutron scattering studies at small angles whenconfined to the nanopores of a polymer [151]. Kozahave linked this low angle scattering feature to theeterogeneity and kinetics of the HDAs and LDA, andsted that there may be only one HDA. The lattert with an earlier conclusion [34,152] that there arelid amorphs, one is known as LDA and the other isvery high-density amorph (VHDA), both are seenpically homogeneous phases. Other HDAs of inter-nsities are therefore mesoscopically heterogeneousDensity of VHDA has been measured at ambientd 77 K by recovering a sample that had formed whenGPa had been heated to 165 K [74]. It has also beenby using the piston displacement data during pres-LDA at 125 K to 1.5 GPa [153,154]. In both cases, ittful that the ultimate density had been reached. It isthe volume does not change at glassliquid transi-erefore the density of the ultimately formed VHDAlose to that of the ultraviscous water at 140 K andt this density is not known, even though its dielec-ion time is known [53,54]. Although the density ofbeen determined by the buoyancy method at ambi-e and also piston displacement measurements at highave been reported, these values have been revised andt data indicate that VHDA produced on pressurizing5 K does not reach the limiting high-density valueWhichever state of HDA has been formed by pres-se of ices Ih and Ic, most studies have shown thatof this state change irreversibly with pressure ande.ontext we recall that HDA was accidentally discov-ice Ih contained in a piston cylinder apparatus at 77 Ks found to irreversibly collapse on raising the uniax-on it to 1.5 GPa. The experiments were similar tormed on ice clathrates a few years earlier by Ross etnce tetrahedral hydrogen bonding in the open struc-s Ih and Ic is much weaker than the covalent bondinge structure of these ices at 77 K collapses at a rel--pressure of 1 GPa with a density increase from1 g cm3 at 77 K and 1 GPa [27,28]. It was earliert at 77 K, ice Ih slowly melted irreversibly at 1 GPad the slow crystalliquid transformation was thermo-nature. If so, it was expected that tetrahydrofuran iceould show a similar transformation at a much lower0.30.4 GPa of its extrapolated phase boundary [56].of volume and dielectric properties of tetrahydrofu-hrate showed no indication of its pressure collapse.tudy showed that it was not possible to recover they phase of the ice clathrate at zero pressure, as itthe original crystalline structures when the pressured [155]. It is still unclear why a high-density amorphobtained from pressure collapse of an ice clathrate,30 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443whose structure is as bulky as that of ices Ih and Ic, and, if ahigh-density amorph does form, why it cannot be recovered atambient prIt is tostate (of unneous, highices, mixtumostly nondecreased,the state renoteworthytion of crysa crystal ifthan the baices Ih anduct is a riglike that ofor non-relaof changinglong as thechange in iorigin, occX-ray andrecovered ain the detaas well aswas conclutime condisingle HDAIce Ic has aHDA, butice Ih [31].In additment data, pby measurvelocity ansure of thedisplacemeshown in Fity at 77 Kin Fig. 8(C5 MHz freqTo deterthe crystalwater, whiforming icmicron-siza relativelymillimeter-micrometesure needecrystal sizeis increaseda nominallat 77 K diding that pocollapse.hanges in the properties of ice Ih on increase in the pressure showingct of collapse or amorphization of ice under a uniaxial pressure that isd to become hydrostatic by mechanical deformation of ice. (A) Volumeindicated as displacement of the piston in the pressure vessel containingrystals of millimeter size and micrometer size at 77 K. Data are takenf. [31]. (B) Increase in the limiting high-frequency permittivity at 77 K.taken from Ref. [56]. (C) Decrease in the thermal conductivity of ice0 K. Data are taken from Ref. [152]. (D) The increase in the velocity ofse sound wave at 77 K. Data are taken from Ref. [32]. Note that increaseure irreversibly transforms crystalline ice to an amorphous structure,oes not transform back to the crystal phase on removal of pressure.eless, the transformation is cyclic in that the amorphous solid can beo obtain the original crystal phase which then can be cooled and pressure-ized the pressure collapse of ice Ih, the limiting high-ncy dielectric permittivity, which contains no relax-l contribution, has been found to increase gradually inoid-shape manner from 3.1 to 3.4, as seen in Fig. 8(B).pression of the sample does not restore the original value,of the pressure-amorphized solid remains at about 3.36his increase from 3.14 to 3.40 on pressure collapse isdue to the increase in the square of the optical refractivefrom 1.73 to 1.79, as the density increases from 0.93 tocm3, but it is mainly due to the increase in the infraredessure at 77 noted that when the pressure on the collapsedknown structural details, homogenous or inhomoge-ly deformed or nano-crystalline mixture of variousre of the ices with non-crystalline solid, or even-crystalline solid, showing no Bragg reflections) isthe sample does not transform back to ice Ih andcovered at ambient pressure remains amorphous. Aparallel to this occurrence is mechanical deforma-tals in a high-speed ball mill, which also amorphizesvitrification temperature of the amorph is much lowerll-milling temperature. In either case, the collapse ofIc, occurring at low temperatures when the prod-id solid, produces a structure of fixed configuration,a nonergodic state. Therefore, only the vibrationalxational properties of the sample change as a resultstructure of the sample during its collapse. Also, astemperature of the collapsed state is kept low, thets properties on cooling and heating is vibrational inurring reversibly, as the structure does not change.neutron diffraction features of the HDA samplest ambient pressure at 77 K have shown differencesils of the structure factor studied by the same groupby different groups, as was discussed earlier [34]. Itded that lack of control of pressure, temperature andtions has led to such differences, and that there is nothat is produced by pressure collapse of ice Ih [34].lso been found to collapse under pressure to producethe pressure is slightly less than that for collapse ofion to measuring the volume from piston displace-rogress of the collapse of ice Ih has been investigateding the thermal conductivity, the ultrasonic soundd the limiting high-frequency permittivity, a mea-infrared-red polarization. The plots for the pistonnt in a pressure vessel containing ice Ih at 77 K areig. 8(A), of the limiting high-frequency permittiv-in Fig. 8(B), of the thermal conductivity at 130 K) and of the velocity of transverse sound waves ofuency at 77 K in Fig. 8(D).mine if the pressure collapse of ice is influenced bysize, experiments have been performed by freezingch produced up to 0.5 mm size crystals, by trans-e Ic to microcrystalline ice Ih, and by allowing thee crystals of ice Ih formed from ice Ic to grow tolarge size. These experiments have revealed thatsize crystals begin to collapse at 1 GPa at 77 K andr-size crystals at 0.7 GPa. This shows that the pres-d to collapse the structure of ice Ih decreases as theis decreased or as the ratio of surface to bulk energy. Johari has reported [31] that in his crude attempt,y 2 mm diameter, 3 mm long single crystal of ice Ihnot collapse at 2 GPa pressure, thereby suggest-lycrystallinity of a sample has a role in its pressureFig. 8. Cthe effeexpectedecreaseice Ih cfrom ReData areIh at 13transverin pressi.e., it dNeverthheated tamorphDurfrequeationaa sigmDecomand [56]. Tpartlyindex1.17 gG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 31(vibrational) contribution to , i.e., the frequency of transla-tional vibrations in the pressure collapsed solid is lower than inice Ih and/ohigher. Thineighbour dbe greater ttion of the npressure-ambeen no stulapse of icefrequency.Yoshimof the Ramamorphizatthe most prdecreases iis increasewhile the 3prominentthis findingfurther contransition toa process oof HDA attheir paperat 1.2 GPapeak and thslight shiftdensificatiodilatometrisample as VThere iYoshimurathe shiftingcies on incp is increathan the togesting thain the 170not constan260 MPa thsamples s170260 Mhaps be maanother tecAnderssof pressuremal condu[158160]sound wavebeen foundto broad inremarkablethe volumeThe plot ofthe same ostructure ascrystals of ice Ih. This indicates that the effect of increase inthe temperature on the onset pressure for collapse is similar toect o(inis ishasde the iceperamnithe vhowcreapressnsveromnicangeg higdicais in-outlotsequechastalsasilynd tre caus coedhownaturts or icesurizt 129decrs clees na129.9 t. Itse cotudyich we ofize inopyer thealsore oflowigmr the absorptivity associated with these vibrations iss observation led to the inference [56] that the nearistance between the water molecules in HDA shouldhan in ice Ih, which was confirmed from determina-ear neighbour distance from X-ray diffraction of theorphized solid by Bosio et al. [156], but there havedies of the infrared spectra during the pressure col-Ih to confirm the inferred lowering of this vibrationalura and Kanno [157] have performed an in situ studyan spectra of ice Ih during the course of its pressure-ion. They found that the 3082 cm1 peak which isominent peak of ice Ih at ambient pressure and 77 Kn height and shifts to lower frequencies as the pressured slowly to 0.9 GPa, and then vanishes at 1.2 GPa,200 cm1 small peak broadens and become mostat 1.2 GPa. Yoshimura and Kanno [157] attributedto formation of strong hydrogen bonds in HDA. Theycluded that amorphization of ice Ih to HDA is not aa glassy state of high-density water at 1.2 GPa, but isf collapse of the structure at high pressures. Heating1.2 GPa to 218 K produced ice VI (ice IV written inis probably a transcription error) and heating HDAto 153 K and also at 0.7 GPa to 153 K made the broade high-frequency shoulder more prominent, with aof both features to a higher frequency, indicatingn of HDA. These observations were reflected in ac study [74], which led to the naming of densifiedHDA.s, however, an interesting feature in Fig. 2 ofand Kanno [157], which is worth noting. Duringof the ice Ih 3082 cm1 peak to lower frequen-rease in pressure at 77 K, the shift is highest whensed from 170 to 260 MPa. It is also much greatertal shift on raising p from 260 to 900 MPa, sug-t something other than simple compression occurs260 MPa range. But if the pressurization rate wast and more time elapsed in raising p from 170 toan at other pressures, the additional change in thetate would be due to the longer time taken in thePa range. Significance of this observation would per-de clearer in future experiments using the same orhnique.on and coworkers [34,50] have studied the progresscollapse of ices Ih and Ic by measuring the ther-ctivity in real time, and Brazhkin and coworkershave studied the velocity and attenuation of ultra-s. Thermal conductivity, , of both ices Ih and Ic hasto decrease on increase in the pressure accordingverted sigmoid shape plot, as shown in Fig. 8(C), insimilarity to the shape of the piston displacement oragainst pressure plot for the ices shown in Fig. 8(A). against pressure at 130 K in Fig. 8(C) shows aboutnset pressure of 0.8 GPa for the collapse of thethe plot of the piston displacement of micron-sizethe effsurfaceple. Thenergyconclucollapsthe temGroies ofThey sulus inunderthe (trafrom Gultrasothe chlimitinThis inimentsspreadshape phigh-fra sharpCrymore e[113] apressuand thperformhave stemperthe ploand fodepresIts ashapeLDA ibecom115 tofrom 0effect115 Kcausedperiodcollapsus to sIh, whcollapsgrain sanisotrto lowThis isstructua muchsmall sf decrease in the crystal size, or increase in the netterfacial or grain-boundary) energy of the ice sam-a remarkable finding in that it suggests that thermala role qualitatively similar to the surface energy. Weat in terms of the high pressure needed, it is easier toIh by either decreasing the crystal size or increasingture of the sample.tskaya et al. [32] have reviewed in detail the stud-elocity of sound waves performed by their group.ed that the velocity of sound waves and shear mod-se sharply as the ice Ih structure at 77 K collapsesure, and the bulk modulus increases [32]. The plot ofrse) ultrasonic velocity against pressure constructednitskaya et al.s data [32] in Fig. 8(D) shows that thevelocity increase is abrupt, and much sharper thans observed in the volume, thermal conductivity andh-frequency permittivity, as seen in Fig. 8(A)(C).tes that the pressure collapse of ice Ih in their exper-itially very sharp. This is quite the opposite of thepressure range for the collapse indicated by the broad-observed in the volume, thermal conductivity and thency permittivity measurements. The reason for suchnge has remained unclear.of ice Ih are mechanically anisotropic. They deformalong the basal plane than along a plane normal to ithus it seemed plausible that, owing to this anisotropy,uses stress concentration at some grain junctions,llapses its polycrystalline sample. But experimentson mechanically isotropic crystals of ice Ic [29,30]that its polycrystalline sample at 77 K and at higheres [50] also collapses under pressure. Fig. 9(A) showsf against pressure for ice Ih at 115 K and 129 KIc at 129 K. An LDA sample was also made bying HDA at 130 K to a pressure below 0.05 GPa.K is also plotted in Fig. 9(A). Here, the sigmoid-ease in on pressure collapse of the two ices andarly evident, but the plots also show that this shaperrower when the temperature of ice Ih is raised fromK and the onset pressure for the collapse decreasesPa at 115 K to 0.8 GPa at 129 K. (Note that a furtherars here as a jog at a pressure of 0.9 GPa for theThis jog from the smooth, sigmoid-shape curve washe inadvertent waiting at this pressure for an unknownoccurrence is significant, as it indicates that pressurentinues isothermally at a fixed pressure, and it has ledthe time-dependence of the pressure collapse of iceould be described later here.) The onset pressure forice Ic at 129 K is0.1 GPa less than that for ice Ih, theboth cases remaining large. Thus loss of mechanicalin going from ice Ih crystals to ice Ic crystals servescollapse-pressure of their structures, not increase it.borne out by the finding that the disordered bulkierLDA, which is mechanically isotropic, collapses ater pressure of 0.35 GPa at 129 K, as is seen as theoid shape decrease in in Fig. 9(A). The decrease in32 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443Fig. 9. (A) This plotted agation ranges, alowering the pplot for 115 Kfor an unknowdecreases witat 129 K thanductivity of icduring the cou0.85 GPa at whad begun. Dthe collapseparticularlywhen thermof the effepressure reis increasedcollapsed saspect disc6. Time-,extent of aThe plostepwise mraised incrat 130 K w0.8 GPa, wthe collapse leading to amorphization, and then measured withtime for 1 h as the sample was kept at 0.8 GPa. The decreaseshoGPse obowlyn, ansingd sqen band ks bect ammors timhizattly htranforin isto 0.85decreawas slparisoincreaby fillehas berangeprocesamounin , isThiamorpcurrennamicat 77 Kermal conductivity of ice Ih at 115 and 129 K and of ice Ic at 129 Kinst the pressure showing their respective collapse or amorphiza-nd the thermal conductivity of the pressure-amorphized solid onressure to ambient. The jog from the smooth sigmoid shape in thenear 0.9 GPa is caused by the inadvertent waiting at this pressuren period. The plots show that the onset pressure for amorphizationh increase in the temperature and is about 0.1 GPa lower for ice Icfor ice Ih. Data are taken from Refs. [50,152]. (B) Thermal con-e Ih at 130 K is plotted against the pressure. The vertical arrowsrse of amoprhization are for fixed pressure conditions of 0.8 andhich the sample was kept for 1 h after the pressure-amorphizationata are taken from Ref. [34].onset pressure on increase in temperature of ice Ih issignificant, as it means that this pressure decreasesal energy of ice Ih is increased. This is the oppositect observed for vitrification of liquids, because thequired to vitrify a liquid increases as the temperature. The plot in Fig. 9(A) shows that of the pressureolid decreases when the pressure is decreased, anussed here in connection with the results in Fig. 3.pressure- and temperature-dependence of themorphizationt of against the pressure in Fig. 9(B) shows theanner in which ice Ih collapses, when the pressure isementally. In this experiment, of an ice Ih sampleas first measured as its pressure was slowly raised tohich is a higher pressure than the onset pressure fordiagram [4in Ref. [83solids withnot the equtransformatransition adepends onBecause ofhas been sIh in real tperiod of sat 128 K an110 ks (30.shows a mitself occurof an approvenient timother pTobtained fo(C).In genepressure coing occurrecollapse atchanges thethe samplestate at itslapsed buttoward a mdeformed ctals of highwhose defotime. In ocstate is expafter the onstate formeicant onlytheir physicwn by an arrow. The samples pressure was raiseda and the measurements made for another 1 h. Theserved is also shown by an arrow. Finally the pressureraised to completely amorphize the ice. For com-other sample of ice Ih was studied by continuouslythe pressure at the same rate and its results are shownuares in Fig. 9(B). The study shows that once ice Ihrought to a pressure in the (collapse) amorphizationept at that pressure isothermally, the amorphizationomes time-dependent at a fixed T and p and that theorphized in 1 h, as indicated by the vertical decreasee at 0.85 GPa than at 0.8 GPa.e dependence is an important aspect of the pressure-ion mechanism, particularly as it contradicts theeld presumption that ice Ih to HDA is a thermody-sformation with an equilibrium pressure of 0.5 GPathe purpose of constructing the equilibrium phase4,161]. Merit of this phase diagram was questioned] where it was shown that HGW and LDA are twodifferent thermodynamic properties. Since HDA isilibrium state, but only one of the intermediate states,tion between ice Ih and HDA is not an equilibriumt 0.5 GPa and 77 K. It also shows that the state formedthe thermal and compression histories of the sample.its significance, the time-dependence of this processtudied in detail by continuously measuring of iceime at a fixed temperature and fixed pressure over aeveral days. In a typical experiment, ice Ih was keptd 0.8 GPa and its was measured over a period of6 h). Its value is plotted against time in Fig. 10(A). Ituch larger decrease in over time and the decreases in an asymptotic manner, but with little indicationach to a limiting low value in an experimentally con-e. Further experiments were performed on ice Ih forconditions and for different time periods. The datar six such conditions are shown in Fig. 10(B) andral, the time-dependence of a property during thellapse indicates one or several of the four follow-nces: (i) crystallites in the polycrystalline sampledifferent rates by a mechanism that continuouslyconditions required for their collapse with time, (ii)becomes a mixture of crystallites and their collapsedinitial stages, (iii) all the crystalline sample has col-the state formed is kinetically unstable and is tendingore stable state at high pressures, and (iv) highlyrystallites or else a mixture of nanometer-size crys-pressure polymorphs of the material have formedrmation or composition continuously changes withcurrences (i) and (ii), the rate of approach to a stableected to be very small in the early period immediatelyset of collapse because the amount of the collapsedd would be negligibly small. It would become signif-when most of the sample is in the collapsed state. Inal manifestation, occurrences (iii) and (iv) are analo-G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 33Fig. 10. (A). Tice Ih at 0.8 Grelation givenseveral samplmalized value(zero time) vaplotted againsData are takengous to struamorphizatto structuramelt, or toby vapour-high fictivewith time, aamorphousature. In thp and fixedtowards a lsample as wtime variesTo helpphization ois converted to the normalized value [(t)/(0)]p,T, and is plottedagainst the time, t, in Fig. 10(C). It is evident that [(t)/(0)]p,Tses wnt fodicathe pGPthats amselyprowome.)8 GPreacnditme ddecreadifferealso inwhenat 0.86meansreflectare clo[50], adensitywith tifor 0.8a timepT cowith tihermal conductivity of partially collapsed or amorphized state ofPa and 128 K. The line is calculated from stretched exponentialhere. Data are taken from Ref. [73]. (B) Thermal conductivity ofes plotted against time. Data are taken from Ref. [73]. (C) Nor-s of the measured thermal conductivity with respect to the initiallue of the samples kept at different pressures and temperatures aret time. The pressure and temperature conditions are as labelled.from Ref. [152].ctural relaxation of the state obtained by mechanicalion of crystals in the high speed ball-milling process,l relaxation of a glass formed by hyperquenching thestructural relaxation of an amorphous solid madedeposition. All the three states have high energy andtemperatures. They become denser spontaneouslys their amorphous states approach asymptotically anstructure of lower energy and lower fictive temper-e plots in Fig. 10(B), at various conditions of fixedT is also seen to decrease asymptotically with timeimiting low value. It is clearly evident that of theell as the rate of the asymptotic decrease in withwith the pT conditions.discuss specifically the pressure collapse and amor-f ice Ih, each set of measurements given in Fig. 10(B)be reachedindicate thaIh structurelapsed statet, as indicathan have sThe asyity with timdescribed b(t) = (where (0)sure experitime valueor amorphiby the smosame dataThe value() = 1.8From theseit would ta5% of thein Fig. 10(formed wostate formeTo elabora0.7 W mat 0.8 GPahigher. Theof ice Ih ortime at 0.8at 1.15 GPatransformaof the stateincreases oincrease ofthe densitywas formedith time, and tends toward [()/(0)]p,T, that isr different pT conditions. The plots in Fig. 10(C)te that [(t)/(0)]p,T values at a given time differ evenT conditions are almost the same, as for the plotsa and 130 K and 0.85 GPa and 129 K. This findingthe collapse-rate, which probably but not necessarilyorphization rate, differs even when pT conditionssimilar. (As increase in density increases of HDAcess involving only relaxation of HDA to a higheruld be inconsistent with the observed decrease in Moreover, the plot for 0.85 GPa and 129 K and thata and 127 K cross over, thus showing that there ished at which [(t)/(0)]p,T is the same for differentions, although the rate of decrease in [(t)/(0)]p,Tiffers. Clearly, a given value of of the sample canby different pTt paths. Altogether, these findingst the initial p, T conditions determine the rate of ices pressure collapse and that the amount of the col-or the extent of amorphization at a given p, T andted by the value, is determined by more variableso far been considered in such studies.mptotic nature of the decrease in thermal conductiv-e at 0.8 GPa and 128 K seen in Fig. 10(A) can bey a stretched exponential relation,) + [(0) ()] exp[(t0,)](7)is the value of at the instant when the fixed pres-ment at 0.8 GPa was begun, () the limiting longof , and 0,, the characteristic pressure collapsezation time. The fit of Eq. (7) to the data is shownoth line in Fig. 10(A). The normalized value of theis plotted against logarithmic time in Fig. 10(C).s obtained from this fit are (0) = 3 W m1 K1,W m1 K1, 0, = 90.9 ks (1500 min) and = 0.6.parameters we calculate that, at 0.8 GPa and 128 K,ke 120 h to reach a state whose value is within() value of 1.8 W m1 K1. Moreover, the plotsC) show that at 0.8 GPa and 128 K, of the stateuld not decrease to the same value as that of thed at 1.15 GPa and 129 K in the plot of Fig. 9(A).te, of the state formed at 1.15 GPa and 129 K is1 K1 in Fig. 9(A), but () of the state formedand 128 K is 1.8 W m1 K1, i.e., 1.1 W m1 K1refore, it would seem that either the collapsed statethe HDA formed after keeping for a formally infiniteGPa and 128 K would be different from that formedand 129 K, or Eq. (7) does not apply for the entiretion. In this context two findings are significant: (i) formed by slow pressurization to 1 GPa at 129 K [34]nly slightly on heating, which is due to an inherentHDAs withT and not due to a concurrent increase inon HDAs conversion to VHDA [74], because VHDAalready during slow pressurization at high temper-34 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443atures in our experiments, and (ii) the denser state recoveredat ambient pressure has different X-ray diffraction features (seeRef. [152]clude that tupon the p130 K evlikely be inwould decrsure, and thand 130 Kto that receAs partIh to the coconductivit(t) =(and the plosigmoid shlization, phrelaxation adetail earliIt shouldfitted genelar relaxatistructural rin chemica has beena broad disbution hasheterogenechemical rediffusion-ction of the rby Plonkasive kinetichomogenizglasses) doConsequenand producformation tthe amorphlevel regiosample oveRamanfeatures at[169]. As ea lower freparametersto allow dithen it is wsity of ice Iin Fig. 2, Rwith HDA1.2 GPa isat 0.7 GPa.135 K transforms to ice VII, and on depressurizing on ice VII at135 K, the sample converts to LDA without forming HDA [169].real tisothre arf chse orthiss alrice Irpha cer(mor8 Wdefoystasitio(iii)and[34]tobyicallydetethattextur dialar,Pa aacheultsrallya. Saectrontae Ih,h andthang oabolapsete cs incthel coly ocetrihizatly, rethatthe oemeof bomisvoluA),for discussion). Based on these findings, we con-he final state attained on pressurizing ice Ih dependsT conditions. If the sample is kept at 0.8 GPa anden for an almost infinite time, the final state wouldhomogeneous. However, the extent of inhomogeneityease with annealing time and with increase in pres-e state formed after long-time annealing at 1.15 GPais likely to be homogeneous and close, or identical,ntly referred to as VHDA [74].of further analysis, , the extent of conversion of icellapsed state has been calculated by using the thermaly data measured with time in the relation,(0) (t)(0) ())p,T(8)t of against time has been found to have an extendedape, resembling the shape of the extent of crystal-ase transformation, chemical reactions and structuralgainst time plots. This feature has been discussed iner [152].be noted that equations similar to Eq. (7) have beenrally to the relaxation spectra in studies of molecu-on processes, to the enthalpy and volume changes inelaxation studies, and to the extent of transformationl reaction kinetics. In all these studies, the quantityfound to be less than 1 and interpreted in terms oftribution of relaxation times [162164]. The distri-been suggested to be due to microscopic dynamicity in ultraviscous liquids and glasses [165167]. Inaction kinetics, it has been interpreted in terms of aontrolled kinetics, dispersive kinetics, or a distribu-eaction rate constants. This latter idea was developed[168], who has developed the concepts of disper-s of such (transformation) reactions [168], in whichation at a molecular level (in viscous liquids andes not occur over the time scale of the transformation.tly, a molecular level heterogeneity of the reactantsts develops within the bulk of a sample on the trans-ime scale. An extension of Plonkas theory [168] toization process of ice Ih would mean that molecularns of the amorph and ice Ih exist in the bulk of ther the transformation time scale.spectra of ice Ih at 135 K and 0.7 GPa have showntributable to the presence of both ice Ih and HDAxpected, the OH-stretching peak of ice Ih shifts toquency [170]. The spectral sampling time and otherfor measurements are usually kept the same in orderrect comparison of the spectra, and if that is the caseorth pointing out that the OH stretching peak inten-h relative to that of pure HDAs broad peak at 1.2 GPaef. [169] indicates that at least 50% ice Ih remainsat 135 K and yet the intensity for pure HDAs peak atsurprisingly not much more than that of the sampleOn pressurizing from 1.2 to 3.5 GPa, pure HDA atSinceice Ihpressutypes ocollapDuringsolid, ature ofor amoIh andwhoseor an aof 1.highlysize crcomporenceX-rayearlierdensityformedisobarTorecallin thein theiparticu0.5 Ghad reture faa textu1 GPtion sp77 K cand icamorpsibilityhandliTheple colcomplit seemIh andthermaabruptdilatomamorpabruptknowabovedisplacrangecan bein theFig. 9(ime studies of the Raman and far infra-red spectra ofermally at a fixed pressure or with slowly increasinge not available, we can only qualitatively discuss theanges occurring at various times during the pressureamorphization of ice Ih at a sustained high pressure.occurrence at, for example, 0.8 GPa and 129 K, theeady noted in general terms above, may be, (i) a mix-h and a certain unknown amount of the collapsed statewhose () is 1.8 W m1 K1, (ii) a mixture of icetain unknown amount of a collapsed state or amorph) is 0.7 W m1 K1, (iii) entirely a collapsed stateph that gradually transforms to its own () valuem1 K1 at 0.8 GPa and 129 K, or (iv) a mixture ofrmed ice Ih crystals, or else a mixture of nanometer-ls of high-pressure ices whose deformation and/orn continuously changes with time. Of these, occur-and (iv) may have produced HDAs with differentneutron diffraction and other features, as discussed, and which would have led to a further increase ina state named VHDA, when an HDA state alreadypressure collapse of ice Ih at 77 K and 1.45 GPa washeated at 1 GPa to 160 K [74].rmine the relative merits of these possibilities, weHemley et al. [171] had reported changes occurringre of the ice Ih at 77 K, as the pressure was increasedmond anvil high-pressure cell containing ice Ih. Inthey observed extensive fracturing of the sample atnd development of turbidity by the time the pressured 1 GPa, and a new phase appearing along the frac-[171]. This indicates that ice Ih exists initially withdifferent, turbid solid at least up to a pressure oflzmann et al. [130] have reported that X-ray diffrac-a of the recovered sample at ambient pressure andined distinctive features of both the amorphous solidwhich means that their sample was a mixture of theice Ih, as in (i) and (ii) above, and ignoring the pos-t some of the ice could have also formed during thef the occurrence (iii) requires that all ice Ih sam-e or amorphize abruptly. But no study has shown thatollapse or amorphization of ice Ih occurs abruptly andonceivable that near the onset of this occurrence, icecollapsed state or the amorph would have the samenductivity, thereby preventing us from detecting anycurring changes. It should also be noted that in onec study of 11.5 mm thick an ice Ih sample at 77 K,ion apparently occurred [172] first slowly and thenaching completion at 1.06 GPa [172]. But we nowwhen ice Ih is kept at a fixed pressure, that is farnset pressure, for a short period, the plot of the pistonnt against pressure in the collapse or amorphizationth ices Ih and Ic would be vertical. This vertical plottaken as an indication of a relatively abrupt decreaseme. (Note that the jog in the plot of at 115 K inor the vertical decrease in in Fig. 9(B) here couldG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 35have been mistaken as an indication of abrupt amorphization.)Since the rate of pressurization in the earlier study [172] had notbeen contrbetween prto a featurenot consideHemleyice Ih at 77and (b) thatmechanistipressure wpreferred reof differenhigh-pressuto scatter lthe high-deis probablerecrystallizat a certainformed shoby the relaamorphoustural relaxasequentiallcharacterisplots of pinot be idennanometerwould not bsame tempehave shownnot identicaThe factnot thermoof the proprial is in a tother propeby the pathone wouldas well aspressure-amtive of thediscussed i1.6 GPa anferent path1.38 g cm2.21 to 2.2path for proand 77 K. Athe temperhas been fuLDA differAs mentperformedstudies, haIh and Ic, ttaken in peTherefore, the term HDA is to be used in a generic sense refer-ring to all high-density amorphous solids formed by the collapseIh ae HDat a porphin sden1.5 Gre coA onate, te de25 tof uientimesiente Ih[60highis reed Hthesegesty bet tranecenstudlatioral pscally twdisortent wl prochanstillue as ofh prX-rais usstal shas sorphs prwn sr amg atsed4] won hahen3,15olled, and an unknown period of time was allowedessure-increase steps, and further that this could leadsimilar to the one seen in Fig. 9. Therefore, we dor (iii) above as a probable al.s [171] real-time optical observations that, (a)K, fractured under a hydrostatic pressure of 0.5 GPa,it became turbid with further increase in pressure, arecally significant because fracture of ice Ih at 0.5 GPaould occur only if it transforms to a denser phase ingions, and turbidity would develop if either crystalst refractive indices are formed, or the size of ice Ih,re ices and amorphous regions are all small enoughight. In either case, ice Ih crystals may coexist withnsity ices or an amorph as in (i) or (ii) above. Itthat ice Ih crystallites begin to deform and/or toe to nanometer size crystals of high-pressure icespressure as mentioned in (iv) above, and the statews no Bragg peaks. This occurrence is followedxation of the structure with time to a collapsed, orstate at 0.8 GPa and 129 K. We conclude that struc-tion at a fixed pressure may occur via a series ofy denser states of lower , until the final state of atic value has been reached. One expects that theston displacement and of against pressure wouldtical because the process by which the mixture ofsize crystals of high pressure phases of ice formse identical in different experiments performed at therature and the same pressurization rate. Experimentsthat such plots obtained in different experiments arel.that pressure collapse or amorphization of ice Ih isdynamic in nature is also evident from a basic analysiserties of an equilibrium state. Briefly, when a mate-hermodynamic equilibrium state, its vibrational andrties are a state function, i.e., they are not determinedused to produce that equilibrium state. Thereforeexpect that the measured values of the density, ,ultrasonic velocity, vtrans, of the HDAs obtained byorphization of ice Ih would be the same irrespec-temperaturepressure path used to obtain HDA, asn Ref. [37]. Briefly, the analysis has shown that atd 77 K, the values of the HDAs formed by dif-s are spread over a 0.05 g cm3 range, from 1.33 to3 and the vtrans values over 0.07 km s1 range, from8 km s1. This means that the pressuretemperatureducing HDAs determines its and vtrans at 1.6 GPasimilar conclusion has been reached by examiningature and pressure derivatives of and vtrans, and itrther shown that the HDA formed by pressurizings from the HDA formed by pressurizing ice Ih [37].ioned earlier here, a variety of data from experimentsby different groups, including some X-ray diffractionve shown that the different temperatures of the iceshe different pressurizing rates and the different timerforming the experiments produce different HDAs.of icesthat th160 Kthe amlookedfurthering tothere aof HDelaborand onfrom 1and coeffic1.4 tcoeffic of icat 93 Kand atficientdensifiSincewe sugscrutindistinchere, rteringa correstructuspatialare onneousconsistherma7. MeIt isto a trcrystaland higin thephousthe crywhichan amhas thuunknotion fomeltinHDA uRef. [4structieven wstate [8nd Ic, and not to a specific solid. It is also known [74]A formed at 77 K densifies by 5% when heated toressure higher than 0.8 GPa, and we have found thatization process is time-dependent, an aspect over-tudies in which conversion of LDA to HDA, and asified state of HDA has been achieved by pressuriz-Pa at 125 K [153,154]. Our analysis has shown thatmplications in interpreting such further densificationincrease in pressure as was done in Ref. [154]. Tohe data for two states of HDA, one normally madensified, in Fig. 3 of Ref. [154] show that on coolingo 77 K, of the densified HDA increases by 3.3%ndensified HDA by 2.3%, which makes expansionof the densified HDA as 7 104 K1, which isas large as that of HDA. Such a high expansionvalue for a solid seems erroneous. For comparison,increases from 0.9292 g cm3 at 173 K to 0.9340], only by 0.55% over a larger temperature range,temperatures, where the thermal expansion coef-latively high. Also, their data [153,154] show thatDA has a higher (elastic) compressibility than HDA.results are unexpected and appear counterintuitive,that the density data of HDA require experimentalfore further discussion in terms of a pressure-inducedsformation of HDA to VHDA. As mentioned earliert wide angle diffraction and small angle neutron scat-ies of HDA and LDA [36] have shown that there isn between their preparation conditions, microscopicroperties, extent of heterogeneities on a mesoscopice and transformation kinetics and further that thereo modifications that can be identified as homoge-dered structures, namely VHDA and LDA, a findingith the conclusion reached earlier on the basis of theperties of HDAs and LDA [34,152].ism of pressure-amorphizationuncertain whether ice Ih pressure collapses directlymorphous solid, to an assembly of highly distortedno identifiable form, or to a random mixture of ice Ihessure forms of ice that shows no crystal-like featuresy and neutron diffraction. However, the term amor-ed to distinguish solids showing no Bragg peaks fromtate, and for that reason the pressure collapsed ice Ih,hown no Bragg peaks has been pre-emptively called. The consequent use of the terms HDA and LDAecluded discussion in terms of a collapsed state of antructure. As mentioned earlier, the original explana-orphization of ice as a pressure-induced equilibrium0.5 GPa at 77 K may now be abandoned, becausefor construction of the phase diagram as reviewed inas not an equilibrium state. This phase diagram con-s been questioned on the basis of a variety of studiesthe HDA formed was taken to be in an equilibrium7]. Also, the use of emulsions of ice for studying the36 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443volume and enthalpy changes to determine the phase diagramof ice Ih and HDA has been shown to contain unresolved, extra-neous thermemulsifyinWe alsoory for medefects or lfor pressurple whichand a largesure decreasize. Alsoently haveseems to hcated by acan partly osure for icebut pressuupon bothamorphizatsteric conspressure. Instudy of th1.8 GPa ha[86]. But pneutron difstatically cpressure trathat alreadytransversepolarizatioused to sugtion of iceexpansionrequiremencrystals woet al. [176]Ih at low teas 2.5 GPphize at a pthan that osingle cryssure of 2in these stIh in expermedium [1solved nitrothe dielectrIh by sevewhether itexpansionNone ofvation thatwith a welTherefore,peculiar toboundariesarrangement is known to be liquid-like. It is also known thatwhen a polycrystalline sample with submicron size crystals isin thsignns oliquilk ennd thnowner-likarieslineperif thesureressuandlly ingnvely wystalprocnd ts. Tipienariessamre isof vughts ismaticdealed tncreic phllapshis lse prconeredrmaof atimeharmf anis seens grimpefecstalsallyt, die. Awelllity (odynamic effects that arise from interaction of theg agents with the ice surface [173].recall that a mechanism based on Lindemann the-lting applies to an ideal single crystal without pointattice faults. Therefore, it is not expected to be valide collapse or amorphization of a polycrystalline sam-contains point defects, dislocations, internal strainsgrain-boundary area, and whose amorphization pres-ses, as described here, with decrease in the particleattempts to collapse single crystals of ice appar-not succeeded [31]. In passing, we note that ice Icave a more defective structure than ice Ih, as indi-broader distribution of relaxation times [174], whichr, possibly, entirely explain the lower collapse pres-Ic. Moreover, Lindemann melting is instantaneous,re-amorphization has a slow kinetics that dependsT and p. As an alternative mechanism for pressure-ion, Sikka [175] has proposed that development oftraints in the crystal leads to its amorphization undercontrast, a recent high-resolution X-ray scatteringe amorph formed by pressurizing ice Ih at 77 K tos shown that HDA has crystal-like inelastic responsehonon dispersion data obtained from an inelasticfraction study of single crystal of ice at 140 K hydro-ompressed to 0.55 GPa by using fluid nitrogen asnsmitting medium [176], have led to the conclusionat 0.50 GPa pressure, a pronounced softening of theacoustic phonon branch in the [100] direction andn in the hexagonal plane occurs [176]. This has beengest that the lattice instability leads to amorphiza-Ih already under 0.5 GPa pressure. Negative thermalof inorganic crystals has also been considered as at for their pressure collapse and to infer that suchuld amorphize under pressure [177,178], and Strasslehave used the negative expansion coefficient of icemperatures to determine its Born instability pressurea. This means that a single crystal of ice may amor-ressure higher than 2.5 GPa, which is much higherbserved here, and consistent with the finding that atal of ice Ih did not amorphize at a hydrostatic pres-GPa [31]. Nevertheless, it is important to examineudies whether or not dissolution of nitrogen in iceiments using fluid nitrogen as pressure transmitting76], has an effect on the phonon dispersion data. Dis-gen at ambient pressure has been known to decreaseically measured molecular reorientation time of iceral orders of magnitude [179], and it is not certainalso has an effect on the temperature at which thecoefficient of pure ice is negative.the above given mechanisms account for the obser-pressure-amorphization of ice Ih is time-dependent,l-defined kinetics, and with a distribution of times.we consider other manners of melting that area polycrystalline sample as follows: at the grain-in a polycrystalline sample molecular or atomicheated[180],junctiosolidthe bumelt aalso kor watboundcrystalboth exratio oin preswith pices Ihof usuaial loadthus cosumabpolycrin thisform achangethe incboundnot theThefillingbe brocrystalmatheof an iproposwhen iacoustthe cotice. Tcollapelasticconsidtransfomationat thequasi-sure oBut thcontaiwhichpoint dthe cryplasticreoriento movple asinstabie temperature range far below its bulk melting pointificant premelting occurs at the three- and four-grainf the crystallites. In this process, the change in thed interfacial energy compensates for the change inergy, and thus in this incipient melting process, thee solid remain at thermodynamic equilibrium. It isthat the surface layer of ice crystals is disorderede [113]. The amount of water contained in the grainand grain junctions of micron size grains in poly-ice Ih at ambient pressure has been determined fromments [181] and calculations [180]. Since the relativesurface energy to bulk energy changes with changeand temperature, this amount is expected to changere. It is meaningful to recall that amorphization ofIc has been carried out by uniaxial loading at a rate0.10.2 GPa min1 in most experiments, and uniax-plastically deforms ice Ih crystallites anisotropically,rting the uniaxial load to a hydrostatic pressure pre-ithin less than 30 s at 77 K. As occurs generally forline samples of materials, the ice crystallites reorientess, the sample recrystallizes, new grain-boundarieshe population of the three- and four-grain junctionshis occurrence in turn would continuously changet melting conditions at the grain junctions and grain, and although it is still the process of melting, it ise as the Lindemann doubt that collapse of ices Ih and Ic leads to theoids in their bulkier crystal structures, and this mayabout if the Born [182] stability condition for theirviolated by the application of pressure. (Born hadally described the conditions for the loss of stabilitycrystal lattice, i.e., of a perfect single crystal. He hadhat a crystal lattice becomes mechanically unstablease in the hydrostatic pressure softens the transverseonon modes and the elastic modulus decreases ande occurs homogeneously throughout the crystal lat-eads to the formation of another crystal phase.) Theessure is determined by the manner in which thestants change on compression. Although Born hadthe lattice stability violation conditions for the phasetion of one crystal to another, and not for transfor-crystal to amorphous solid, which was not known, his theory has been used to calculate [183], in aonic approximation, the mechanical collapse pres-ideal ice Ih single crystal at different irrelevant to polycrystalline ices Ih and Ic, whichain boundaries and three- and four-grain junctions inurities segregate, and its individual crystals containts, impurities, and dislocations. Moreover, the state ofin a polycrystalline mass change as uniaxial loadingdeforms the ice sample, causes its crystal grains toslocations population to increase and the dislocationsll of these features, which are characteristic of a sam-as that of a material, are expected to alter the Bornor stability violation) pressure and thereby cause theG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 37extent of amorphization to depend upon the time, temperatureand pressure, and to a small degree upon the microstructure of thesample itsecrystal of icalso wouldit may seeinvolves twmelting at tand (ii) incrdisorderedIt should alHDA conveand their deven thougto that of icIn a meviolation inpressure, blower the cimportantlywould causdifferent prBorn instabcrystallinethe mechanslow componly after tthe near ide(The situatcrystallineat its own ccoexist witmeans that(stronger) isure and thice crystallonset presscrystal graiElectroshave beenthe conseqstrength ofas a result othe neighbosubsequentier, i.e., aftcollapse wstretched sand high-fthat this wregions in win size as nof the whotal stabilityice, its manical collapsthe time-depT condition may be reconciled with Borns stability crite-rion if occurrence of another molecular process that changesstalledgerentouldslocauousre atethecrysressute thprestruhighs woby thansfomecy.houte colhizatpTinsictalliconnstabpresand) thementid nonstabulateolycexpeystalhizeerimamermoh anorphifferlingreva tef difo cocanned bthaned hlf. It has also been found that dislocations in a singlee have an extended noncrystalline core [184], whichalter the conditions of pressure-amorphization. Thusm that pressure-amorphization of ice Ih and ice Ico mechanistically distinct processes: (i) incipienthe inter-granular regions in a polycrystalline sampleease in the population of dislocation cores containingarrangements of water molecules in the be noted that the much lower pressure of LDA torsion cannot be explained by the influence of defectsiffusion on the Born instability condition of LDA,h in terms of decrease in LDAs collapse is similares Ih and Ic.chanical collapse resulting from the Born stabilitya crystal, lattice faults would lower the collapseecause these faults store energy and in most casesrystals density from that of an ideal lattice. More, a variation in the population of the lattice faultse different crystallites in the sample to collapse atessures. Thus one would expect a distribution of theility pressures (of different crystallites) in a poly-sample, which would broaden the pressure range forical collapse of the sample. Hence even at a veryression rate, full amorphization would be reachedhe pressure is of a magnitude high enough to collapseal ice crystal of the highest Born instability pressure.ion may be seen as analogous to a multi-componentcomposite, in which each component would collapsesharacteristic pressure.) If this occurred then ice wouldh the amorph at formally infinite annealing time. Thisthe final state achieved would remain a mixture ofce Ih crystallites that did not collapse at a given pres-e amorph that formed by the collapse of (weaker)ites. This would also explain the observation that theure of amorphization decreases with decrease in then size in the sample [31].tatic interactions in the structure of crystalline icesfound to be co-operative in nature [185,186], withuence that breaking of some H-bonds weakens thethe others. Accordingly, breaking of some H-bondsf structural collapse of ices Ih and Ic would weakenuring H-bonds in the ice crystal. If that were to occur,pressure-amorphization of ice would become eas-er part of the ice Ih structure has collapsed, furtherould require a smaller increase in pressure. But theigmoid shape plots of volume, thermal conductivityrequency permittivity against pressure have showneakening effect is inconsequential. Therefore, thehich the structure collapses seem to be small enoughot to cause a sudden and rapidly increasing collapsele crystal. Thus although the Born criterion for crys-may remain valid for the mechanical collapse ofifestation is altered by a distribution of the mechan-e pressures in a polycrystalline mass. Nevertheless,pendent pressure-amorphization of ice Ih at a giventhe cryknowlconcurcess wand dicontinpressuWhice Ihhigh-pWe noing thethat itstals ofcrystalgivenmay trner asenthalpWitthat thamorpgivenare introf crys(ii) theBorn iof thiscationsand (vexperi77 K dBorn ibe calcfor a pin histhat cramorpan expstate to8. ThamorpAmhave dby coousuallypath inresult otaken tliquidbe formhigherproducites microstructure is included. On the basis of ourof plastic deformation of polycrystalline sample andrecrystallization, we suggest that this molecular pro-be diffusion of defects, redistribution of impuritiestions and partial melting. This mechanism wouldly alter, with time, the distribution of the collapsea given pT condition.r or not this collapse would produce highly distortedtals or extremely small, nanometer size crystals ofre forms of ices are yet to be investigated in formation of ultraviscous water [53,54] on heat-ssure collapsed state does not conflict with the viewcture may contain highly distorted and/or nanocrys--pressure ices because, (i) the melting point of suchuld be significantly lower than of large crystals, ase GibbsThomson equation, and (ii) such crystalsrm to the lower energy glassy state in the same man-hanically amorphized states do on heating by loss ofreferring to the nature of the product, we concludelapse onset pressure, the pressure-range for completeion and the characteristic amorphization time at acondition are determined by at least five effects thatto an ice Ih sample: (i) the rate of plastic deformationtes and their recrystallization under a uniaxial stress,centration of lattice faults in the crystallites, (iii) theility pressure of the crystallites and the distributionsure, (iv) the redistribution of impurities and dislo-partial melting during the period of amorphization,pressurizing rate. It is conceivable that in Joharismentioned in Ref. [31], a single crystal of ice Ih att amorphize at pressures of up to 2 GPa because theility pressure for a single large crystal, which mayd in a harmonic approximation, could be higher thanrystalline sample, and this pressure was not reachedriment [31]. On the opposite end, it is conceivables of ice approaching several nanometers in size mayat a pressure as low as 0.1 GPa. This may be tested byent in which change in properties from nanocrystalorphous state may be detected.dynamics and kinetics of pressure collapsedd of ultraviscous waterous states of a material made by different techniquesent properties. We recall that a glass is formed onlya liquid or by compressing a liquid. The occurrence isersible on cooling and heating, although the coolingmperature plane differs from the heating path as aferent extents of structural relaxation during the timeol the liquid or heat the glass. In those cases when aot be supercooled by usual methods, the glass has toy hyperquenching the liquid, i.e., by cooling at a rate105 K s1, as in the case of water [23]. The glass thusas a high fictive temperature, Tf, than a glass formed38 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443by cooling of a liquid, say at 0.1 K s1. The structure of a hyper-quenched glass relaxes with time to a lower energy state of lowerTf at a rate that increases with increase in the temperature. (Fora discussion of the subject, see for example Ref. [164].) Whensuch a glass is heated, its energy and Tf decrease. After its statehas crossed the equilibrium liquid line in the temperature planefor a certain heating rate, its energy and Tf increase, the glasssoftens and becomes an ultraviscous liquid. If the liquid crystal-lizes rapidly, the glass-softening to ultraviscous liquid may beimmediately followed by the latters crystallization and may notbe observed if crystal nuclei had already formed as a result of theheat released during structural relaxation [119]. For a materialthat does not crystallize in the ultraviscous state, this process isillustrated in Fig. 11(A).In contrast to the normal supercooling of a liquid, a varietyof technologically useful amorphous solids are made by [91],(i) vapour-deposition on a cold substrate, (ii) rapid evapourationof a solution, (iii) chemical reaction that leaves a solid prod-uct, (iv) electrodeposition and (v) mechanical deformation ofFig. 11. (A) Afictive temperspontaneous sfictive temperarrows. In thicollapse of icsolid with timliquid whichsegments of tchange.crystals in a high speed ball-mill. The internal energy, entropyand Tf of these solids are much higher than those of glassesformed byrelaxes toheating mebecomes aand may thglass-softewhich thesmuch as thoriginal staetc.We procollapsed samorphoustals, to theor to the amillustrationsolid produto have a hmetal crystin the mectheir undefence betweamorph formechanicaand the amis denser thsure higherVI, which iices Ih andbe comparethe propertaspect haswhether anby mechanby mechansure and 77as used forTo illustral rrelaxvolu(B).ml me atThea tems beatingtion60 Khiche res4,19longthatn illustration for the formation of high-energy amorphs of highatures by mechanical deformation, and/or by rapid cooling. Thetructural relaxation in a heating rate dependent manner to a lowature, low energy state and then glass transition are indicated bys case the ultraviscous liquid does not crystallize. (B) Pressuree Ih, its structural relaxation to a denser, low thermal conductivitye and on heating, its gradual transformation to an ultraviscouscrystallizes to ice XII on fast heating and finally to ice VI. Thehe plots are labelled and arrows are used to show the direction ofstructuof themolarFig. 1118.1volum125 toline hather herelaxa1501XII, walso thrate [8sure a(Notenormal cooling. When heated, their structure rapidlya lower energy amorphous structure [91]. Furtherchanically softens the solid slowly and it ultimatelyn ultraviscous liquid, which may crystallize rapidlyerefore not show, in some cases, the characteristicning endotherm [91]. The thermodynamic path alonge changes occur is irreversible, but it is cyclic in ase final amorphous solid can be converted back to thetes of vapour, solution, chemical reactants, crystal,pose that in terms of its high energy, the pressuretate of ices Ih and Ice is qualitatively similar to thestate produced by mechanical deformation of crys-glassy state formed by hyperquenching of a liquid,orphous state formed by vapour-deposition. In theof Fig. 11(A), the mechanically deformed amorphousced by mechanical deformation of crystals is shownigh internal energy, as for mechanically amorphizedals and organic crystals [187,188]. The self-diffusionhanically deformed solids is much faster than inormed state. Thermodynamically, the basic differ-en a mechanically amorphized crystal state and themed by pressure collapse of ices Ih and Ic is that thelly amorphized state is bulkier than the parent crystal,orph formed by pressure collapse of these two icesan the parent crystals, which do not survive a pres-than 0.8 GPa. But, as the amorph is bulkier than ices the stable crystal phase at the collapse-pressures ofIc, the density of the pressure collapsed state shouldd with the density of ice VI. In earlier discussion ofies of the amorph at ambient pressure and 77 K, thisbeen unfortunately overlooked. It remains to be seenamorph of lower density than ice VI can be producedical deformation of ice VI at 1 GPa, or else producedical deformation of recovered ice VI at ambient pres-K, by using a high-speed ball mill in the same mannerordinary crystals [189].rate the pressure-amorphization of ices Ih and Ic, theelaxation of the solid on heating and the conversioned state to ultraviscous water, we have plotted theme of these various states against the temperature inIn this figure, ices Ih and Ic have molar volume ofol1 and they collapse to a solid of 13.8 ml mol11 GPa at 77 K, and to 13.75 ml mol1 volume atcollapsed state becomes denser on heating accord-perature-dependent rate, and after the equilibriumen crossed on heating, the volume increases. Fur-transforms it to ultraviscous water whose dielectrictime is 1 s at 140 K [53]. On further heating torange, the ultraviscous water may crystallize to iceexists in the 0.71.5 GPa and 158212 K range, butulting crystalline phase may depend upon the heating0]. Ice XII can be thermally cycled at 1 GPa pres-the path shown by the oppositely pointing arrows.crystallization has been found to occur also to mix-G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 39tures of ices VI and XII [8] and mixtures of ices IV and XII[84], and therefore there is some ambiguity of crystallizationin this metforms to thFig. 11(B).at a temperVI phase bocool througthe 30014waters ultrthe pressurof volumeagainst temtures as thethe same asOne maheating ofof the varicause themmelt themsome casescomminutispeed ballwith a largthese particalready largfrom that oconditions,Instead theparticles ca Tg-endotexothermic1,3,5-tri--and discussolids [91]have clearltallites tothe packagBhat et al.cles of PbGon heatingals have shthe highlyto becomelarger crystmechanicainterfacial tories and tOnce the vnew crystaviscosity ofthe paths sh9. SummaThermaof unexpecnot for the waters amorphous solids, an apparently amorphoussolid of low-density, LDA, shows a crystal-like temperatureencthrathousl cod prol conobsty oflitatiis ofincrse. Wt captherantithinh.sta(ASWeneentis dethe uovathisf waelaxaeffeffracscouhigo staer fodielre ison]. Thf thadensstatDAy du110]asA sge iscryst a pperaperaon thsolids of-dens thastable state). On heating, metastable ice XII trans-e stable phase ice VI [190] at 1 GPa as is indicated inIce VI can be heated at 1 GPa and it melts to waterature slightly above 300 K. Studies of the liquid iceundary have shown that liquid water does not super-h the ice VI stability region and water at 1 GPa in0 K range is unstable. It seems now, however, thataviscous state at high pressure can be obtained viae collapse route of ice Ih, as inferred from the plotin Fig. 11(B). We also note that the plots of enthalpyperature of these phases would show the same fea-volume, with all points of transformation remainingfor the plots in Fig. 11(B).y argue that contrary to the above-given postulates,the microcrystalline or submicrocrystalline particlesous ices formed by pressure-amorphization wouldto act as a nuclei for crystal growth and would notto an ultraviscous liquid. While this may be true in, it is well known that mechanical pulverization oron of organic, inorganic and metallic crystals in high-mills creates micron and submicron size particlese population of point defects and dislocations andles become brittle. As the volume to surface area ise and the surface tension of the particles is differentf the usual crystal-nuclei formed in the equilibriumthese particles do not act as sites for crystal growth.mechanically amorphized powder of highly strainedonverts to a glassy state which on heating showsherm in the DSC scan, immediately after a broadfeature. For example, mechanically amorphizednaphthylbenzene studied by Yamamuro et al. [191],sed in connection with high enthalpy amorphous, and a number of pharmaceuticals [187,188,192]y shown conversion of mechanically deformed crys-glasses. It is also a well-recognized occurrence ining technology of pharmaceuticals. More recently,[193] have shown that ball-milled crystalline parti-eO3 become amorphous and show a Tg-endothermin the same manner as a number of other materi-own. Therefore, it would not be a unique case fordistorted micron or submicron crystals of the icesa glass or ultraviscous liquid rather than to grow toals. It is conceivable that the surface tension of suchlly deformed particles differs substantially from theension used in the nucleation and crystal growth the-herefore those theories may not be applicable here.iscosity of the liquid decreases on heating to T > Tg,ls nucleate and grow at a rate that depends upon thethe liquid and the liquid crystallizes, as illustrated byown in Fig. 11(A) and (B) for water at high pressures.ry and concluding remarksl conductivity of the ices shows a remarkable varietyted behaviours: It is lowest for an ice clathrate, anddependice claamorpthermasity anthermaThecapacibe quaanalysmodesdecreathe heain thenot quity witstrengtThewatera highpromining. ThbeforeIc. Remrevealsstate otural risotopeand diultraviand itsthe twthe othThepressuformed[53,54times ohigherdensityat the His likelrules [in LDAAn LDthe larThelapse athe temthe temeffectphoussampleof highsolid. Ae of thermal conductivity and a distinctly crystallinee a glass-like temperature dependence. Crystals andsolids of water of higher density may have a lowernductivity, and two crystal forms of the same den-perties, ices Ih and Ic, show unexpectedly differentductivities.erved variation of thermal conductivity and heatthe high-density amorphous solid with pressure mayvely explained by the Debye theory. A quantitativethe heat capacity shows that the frequency of phononeases and the Debye energy and the anharmonicityhile the Debye phonons can explain the decrease inacity of HDA by 5% GPa1, and also the increasemal conductivity towards a plateau value, they doatively describe the increase in thermal conductiv-the approximation of a constant phonon scatteringtes of as-made vapour-deposited amorphous solid) and hyperquenched glassy water (HGW) havergy and high fictive temperature, and both show aenthalpy decrease due to structural relaxation on heat-crease masks the onset of glass-softening endothermltraviscous liquid begins to crystallize rapidly to icel of this heat by isothermal annealing prior to heatingendotherm and yields a Tg of 136 K for the glassyter. LDA also shows enthalpy decrease due to struc-tion but it is relatively less and its Tg is 129 K. Thect on Tg and a variety of thermodynamic, vibrationaltion features have shown that glassy water, and thes water obtained by heating it, are different from LDAh temperature state. Thus, there is no evidence thattes of liquid water, one formed by heating HDA andrmed by heating ASW and HGW, are the same.ectric relaxation time of ultraviscous water at ambientestimated to be 30 s at 140 K [118,119] and of waterheating HDA at 1 GPa and 130 K is determined as 5 se dielectric relaxation time of LDA is more than 10t for HDA at 0.3 GPa and 130 K [110]. Thus Tg of theity state of water at 1 GPa is lower than theTg of lowere at ambient pressure. The increase in relaxation timeto LDA transition, despite the 25% density decrease,e to a change in structure to one which obeys the ice. This explains the much restricted dipolar mobilityit does also in (the proton-disordered) ices Ih and Ic.tructure that obeys the ice rules would also explainotope effect on the Tg of LDA [111].tal structures of polycrystalline ice Ih and ice Ic col-ressure greater than 0.7 GPa at a rate that varies withture, pressure and crystal size. The pressurizing rate,ture and the period of sustained pressure all have ane structure and properties of the high-density amor-formed by the collapse of ice. Thus the study of thethe so-called HDA refers to the study of a generic statesity and not to a specific structure or properties of ae temperature is increased, the pressure for structural40 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443collapse becomes less, an effect opposite to that observed for vit-rification of liquids for which vitrification pressure increases asthe temperthan for icethe micronwhen thermface energythere is a thit is expectnanometerpressure thdifficult tomeasuremeincrease inThe kinbution of timexponentiathe temperalapse that ltheory of monly if theand of intemay collapsample.Such efbonded crywhose Ramfeatures wIt would btals at amband to studetermineamorph. Dsamples aslead to a bface energunderstandespeciallypressure-amdetermine ihigh-speedthe naturetion, Ramacollapse isoincreasingthe amorphparticular ia high-eneically amorelatively hthey becomshow a glaThe colforms, as inof some mcollapsed sheating an140 K. On further heating at 1 GPa, this water crystallizes firstto the denser ice XII [84,190] and then to ice VI [190], or, inheamixt 1 Geivabelo-denorphd 77tatetravioolinse ofce 2ingre-amsoscthodmogthe sed bssurifromdiffeiffraf HDA stte (pnsifiet amat 2199]perased sringonfirc solarietthe cmpeThee prer hiraturixteselinehat spreof therenstallto aderssed sh-prature is increased. This pressure is lower for ice IcIh, and it is also lower when the ice crystal size is inrange. Therefore, the collapse-pressure is increasedal energy of ice is low, and decreased when sur-of ice is high. Thus, for a given collapse-pressureermal energy equivalence for the surface energy, anded that samples with crystal-grain size approachingscale would (collapse) amorphize at a much loweran 0.7 GPa. As the nanometer size crystals would bedistinguish from an amorphous solid by diffractionnts, changes in volume and thermal conductivity onpressure would need to be used for such studies.etics of pressure-amorphization of ice Ih has a distri-e constants that is expressed in terms of the stretchedl parameter. The broad pressure range and the time-,ture-, and the pressure-dependences of structural col-eads to amorphization can be reconciled with Bornsechanical collapse, by transverse phonons softening,effects of lattice faults (point defects, dislocations)rnal surfaces are included. A single crystal of ice Ihse at a much higher pressure than a polycrystallinefects may also be observable for other hydrogen-stals, e.g., resorcinol [194] and-hydroquinone [195]an spectra have shown gradual loss of crystal-likeith increase in pressure above a certain pressure.e important to recover the collapsed organic crys-ient pressure and examine their structural disorderdy their thermal properties at ambient pressure toif they also form a high-density and a low-densityetailed studies of pressure collapse of polycrystallinea function of the crystal grain size may generallyetter understanding of the role of the samples sur-y relative to its internal energy. It may also helpthe merits of the Borns theory of crystal-instability,if the Born-instability view is to be extended toorphization of crystals. It may also be important tof such crystals can be mechanically amorphized in aball mill at ambient pressure and then to investigateof the amorph formed. Real-time studies by diffrac-n and FTIR spectroscopy methods during a crystalsthermally at a fixed pressure, as well as with slowlypressure, would be required for understanding howous state is achieved by pressurizing crystals. This ismportant because pressure collapsed crystals are inrgy state and in this state they behave like mechan-rphized crystals. After spontaneous relaxation at aigh temperature to a lower energy disordered state,e analogous to the hyperquenched glassy state andss-softening range.lapsed state on heating itself may yield new crystalthe formation of ice XII [196], and new crystal formsaterials may be technologically useful. The pressuretate of ice Ih at 1 GPa structurally relaxes rapidly ond the solid at 1 GPa becomes ultraviscous water atcertainor to asolid ais concaturesa highare amsure ansame sThe ulsuperccollapSinregardpressuis a meent metwo hoond isproducby preducedAlso,from dstate o[198].clathrathe deering aspectraHDA [the temcollapcompathey cspecifiA vwhenand te[8,84].they arof othtempeand a mVI. Thcrystalnoted tat highas onebe diffCryice Ihnot uncollapthe higting conditions, to a mixture of ices XII and VI [8]ture of ices XII and IV [84]. Thus the amorphousPa is bulkier than ice VI, its stable crystal phase. Itble that high-speed ball-milling of ice VI at temper-w 130 K at 1 GPa may mechanically amorphize it tosity amorph in the same manner as ordinary crystalsized. Alternatively, recovered ice VI at ambient pres-K may be amorphized by ball-milling to obtain theas that obtained by pressure collapse of ices Ih or Ic.scous water at 1 GPa, which is difficult to obtain byg water at 1 GPa, can be obtained via the pressureice Ih.004, a number of new findings, particularly thosethe pressure, temperature and time dependence oforphization of ice Ih, have now confirmed that thereopic heterogeneity in the HDAs produced by differ-s. Moreover, it has been concluded that there are onlyeneous amorphous solids, one is LDA and the sec-o-called VHDA. It has also been shown that the HDAy pressurizing LDA differs from the HDA producedzing ice Ih [37], and X-ray diffraction of HDA pro-ice Ic differs from that produced from ice Ih [197].rent states of HDAs, more precisely distinguishedction studies, have been produced by annealing aA at 0.2 GPa and by thermally cycling the samplesudy of pressurized and annealed tetrahydrofuran iceroduced by pressurizing to 1.5 GPa at 77 K, heatingd sample to 150 K for annealing and finally recov-bient pressure at 77 K) has shown that its Raman5 K and its X-ray diffraction are similar to those for. These observations indicate the need for controllingturepressuretime profile in producing the pressuretate of reproducible properties, as well as for care inimpure samples against pure HDAs. Nevertheless,m that HDA refers to a generic state and not to aid [37].y of high-pressure crystalline phases of ice are formedollapsed state of ice produced at different pressuresratures is heated at different (fixed) high pressuresse metastable crystalline phases are intruders asoduced in the pressuretemperature stability domaingh-pressure crystalline ices, and they persist at lowes. For example, ice XII, a mixture of ices VI and XIIure of ices IV and XII, all form in the domain of iceintruders would ultimately transform to the stablephase by a thermally activated kinetics. It should beome of these crystalline ice phases and their mixturesssures and low temperatures have the same densitye HDAs, but their phonon properties are expected tot.ization on heating of the pressure collapsed state ofmultiplicity of high-pressure phases of ice is stilltood. As a consequence, the nature of the pressuretate is debated, as to whether it is a composite ofessure phases of ice at a mesoscopic scale, some ofG.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 41which may preferentially grow into larger crystals at certainpressure and temperature conditions and thus appear to produceintrudera true amoidentical educes diffefrom pressuhydrogen bcharacter owhether thea natural ccontinuitytwo extremcompressinit. Such stulapse and senough to bin the interfrom plastisample.AcknowledGPJ is ggrant. OAResearch CReference[1] V.F. PeOxford[2] C.G. SChem.[3] C.G. Sa429 (20[4] A. Poli[5] G.P. Jo[6] H. Eng[7] C. Lob[8] S. Klot218 (20[9] J.D. Be[10] M. von[11] D.W. Dvol. 2,[12] J.A. Ri325 (19[13] Y.F. MaGases,[14] B. Beristry, v[15] E.D. SlNew Y[16] W.L. MShu, R[17] L.J. FloK.N. M[18] H. LeeMoudr[19] F. Schu[20] G.P. Jo[21] E.F. Bu[22] J.A. McMillan, S.C. Los, Nature (London) 206 (1965) 806.[23] E. Mayer, J. Appl. Phys. 58 (1985) 663.. Mayer, J. Phys. Chem. 89 (1985) 3474..P. Jo. Hall. Mis. Mis.P. Jo. Mis.P. Jo.L. G001). And.P. Jo.P. Jo.M. K006).P. Jo.G.G.hysicaM. Gpace,.C. F7..J. M.S. Mucson. Jenn. Jenn995).G. De.G. R53..G. R7..G. R967.. And. And. And. Hak269..S. Caon, O. And. And. Fors.P. Jo.P. Jo.P. Jo. And99.. Eisxford.I. Kroman006).G. R7 (198. Berm.W. Aublish. And. Sug. Sug. And. AndeS. Tsphases in the stability domain of other ices, or is itrphous solid in which no two water molecules havenvironment and whose random crystallization pro-rent intruder phases. Here information obtainedre collapse of crystals without water-like tetrahedralonding would reveal not only the mesoscopic scalef the pressure collapsed solids in general, but also thatformation of low-, and the high-density amorphs isonsequence of pressure collapse or is there just aof disordered solids of varying densities between thees of low and high densities, as would be achieved byg a liquid at different pressures and then supercoolingdies may also help determine whether pressure col-ubsequent formation of different amorphs is generale understood as a phenomenon in terms of changesmolecular forces, or whether it is a feature resultingc deformation of the crystal grains in a polycrystallinegementsrateful to NSERC of Canada for a general researchis grateful for financial support from the Swedishouncil.strenko, R.W. Whitworth, Physics of Ice, Oxford University Press,, 1999.alzmann, A. Hallbrucker, J.L. Finney, E. Mayer, Phys. Chem.Phys. 8 (2006) 3088.lzmann, A. Hallbrucker, J.L. Finney, E. Mayer, Chem. Phys. Lett.06), M. Grimsditch, Phys. Rev. Lett. 52 (1984) 1312.hari, J. Chem. Phys. 122 (2005) 194504.elhardt, E. Whalley, J. Chem. Phys. 56 (1972) 2678.ban, J.L. Finney, W.F. Kuhs, Nature 391 (1998) 268.z, G. Hamel, J.S. Loveday, R.J. Nelmes, M. Guthrie, Z. Kristallogr.03) 117.rnal, R.H. Fowler, J. Chem. Phys. 1 (1933) 515.Stackelberg, H.R. Muller, Z. Electrochem. 58 (1954) 25.avidson, in: F. Franks (Ed.), WaterA comprehensive Treatise,Plenum, New York, 1973, p. 115 (Chapter 3).pmeester, J.S. Tse, C.I. Ratcliffe, B.M. Powell, Nature (London)87) 135.kogon, in: translated by W.J. Cieslewicz (Ed.), Hydrates of NaturalPennWell, Tulsa, 1981.ecz, M. Bolla-Achs, Gas HydratesStudies in Inorganic Chem-ol. 4, Elsevier, Amsterdam, 1983.oan, Clathrate hydrates of Natural Gases, 2nd ed., Marcel Dekker,ork,, H.-K. Mao, A.F. Goncharov, V.V. Struzhkin, Q. Guo, J. Hu, J..J. Hemley, M. Somayazulu, Y. Zhao, Science 297 (2002) 2247.russe, C.J. Peters, J. Schoonman, K.C. Hester, C.A. Koh, S.F. Dec,arsh, E.D. Sloan, Science 306 (2004) 469., J.-W. Lee, D.Y. Kim, J. Park, Y.-T. Seo, H. Zeng, I.L.akovski, C.I. Ratcliffe, J.A. Ripmeester, Nature 434 (2005), Nature 434 (2005) 712.hari, La Recherche 387 (2005) 25.rton, W.F. Oliver, Proc. Roy. Soc. (Lond.) Ser. A 153 (1935) 166.[24] E[25] G[26] A[27] O[28] O[29] G[30] O[31] G[32] E(2[33] O[34] G[35] G[36] M(2[37] G[38] AP[39] J.S[40] M3[41] MMT[42] P[43] P(1[44] P[45] R5[46] R8[47] R3[48] O[49] O[50] O[51] B2[52] Hd[53] O[54] O[55] H[56] G[57] G[58] G[59] O5[60] DO[61] AR(2[62] R4[63] R[64] NP[65] O[66] HH[67] O[68] P[69] J.hari, A. Hallbrucker, E. Mayer, Nature (London) 330 (1987) 552.brucker, E. Mayer, G.P. Johari, J. Phys. Chem. 93 (1989) 4986.hima, L.D. Calvert, E. Whalley, Nature (London) 310 (1984) 393.hima, L.D. Calvert, E. Whalley, Nature (London) 314 (1985) 76.hari, A. Hallbrucker, E. Mayer, J. Phys. Chem. 94 (1990) 1212.hima, J. Chem. Phys. 100 (1994) 5910.hari, Phys. Chem. Chem. Phys. 2 (2000) 1567.romnitskaya, O.V. Stalgorova, V.V. Brazhkin, Phys. Rev. B 64094205.ersson, A. Inaba, Phys. Chem. Chem. Phys. 7 (2005) 1441.hari, O. Andersson, J. Chem. Phys. 120 (2004) 6207.hari, O. Andersson, Phys. Rev. B 73 (2006) 094202.oza, T. Hansen, R.P. May, H. Schober, J. Non-Cryst. Solids 3524988.hari, J. Chem. Phys. 121 (2004) 8428.M. Tielens, L.J. Allamandola, in: G.E. Morfill, M. Scholer (Eds.),l Processes in Interstellar Clouds, Reidel, Dordrecht, 1987, p. 223.reenberg, in: J.M. Greenberg, V. Pironell (Eds.), Chemistry inKluwer Academic, Dorderecht, Netherlands, 1991, p. 227.estou, H. Rickman, R.M. West, Astron. Astrophys. Rev. 5 (1993)umma, P.R. Weissman, S.A. Stern, in: E.H. Levy, J.I. Lunine,atthews (Eds.), Protstars and Planets III, Univ. of Arizona Press,, 1993, p. 1177.iskens, D.F. Blake, Science 265 (1994) 753.iskens, D.F. Blake, M.A. Wilson, A. Pohorille, Astrophys. J. 455389.benedetti, J. Phys. Condens. Matter 15 (2003) R1669.oss, P. Andersson, G. Backstrom, Nature (London) 259 (1976)oss, P. Andersson, G. Backstrom, High Temp.-High Press. 9 (1977)oss, P. Andersson, G. Backstrom, J. Chem. Phys. 68 (1978)ersson, H. Suga, Phys. Rev. B 50 (1994) 6583.ersson, G.P. Johari, H. Suga, J. Chem. Phys. 120 (2004) 9612.ersson, H. Suga, Phys. Rev. B 65 (2002) 140201.ansson, P. Andersson, G. Backstrom, Rev. Sci. Instrum. 59 (1988)rslaw, J.C. Jaeger, Conduction of Heat in Solids, 2nd ed., Claren-xford, 1959, p. 341.ersson, Phys. Rev. Lett. 95 (2005) 205503.ersson, A. Inaba, Phys. Rev. B 74 (2006), J. Phys. D: Appl. Phys. 22 (1989) 1528.hari, S.J. Jones, Philos. Mag. 54 (1986) 311.hari, E. Whalley, J. Chem. Phys. 64 (1976) 4484.hari, E. Whalley, J. Chem. Phys. 115 (2001) 3274.ersson, B. Sundqvist, G. Backstrom, High Press. Res. 10 (1992)enberg, W. Kauzmann, The Structure and Properties of Water,University Press, Oxford, UK, 1969.ivchikov, B.Ya. Gorodilov, O.A. Korolyuk, V.G. Manzhelii, O.O.tsova, H. Conrad, W. Press, J.S. Tse, D.D. Klug, Phys. Rev. B 73064203.oss, P. Andersson, B. Sundqvist, G. Backstrom, Rep. Prog. Phys.4), Thermal Conduction in Solids, Clarendon, Oxford, 1976.shcroft, N.D. Mermin, Solid State Physics, Saunders Collegeing, Philadelphia, 1976, p. 258, 501.ersson, PhD Thesis, Umea University, Umea, 1991.a, T. Matsuo, O. Yamamuro, Pure Appl. Chem. 64 (1992) 17;a, T. Matsuo, O. Yamamuro, Supramol. Chem. 1 (1993) 221.ersson, H. Suga, J. Phys. Chem. Solids 57 (1996) 125.rsson, R.G. Ross, J. Phys. C 16 (1983) 1423.e, M.A. White, J. Phys. Chem. 92 (1988) 5006.42 G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443[70] J.S. Tse, V.P. Shpakov, V.R. Belosludov, F. Trouw, Y.P. Handa, W. Press,Europhys. Lett. 54 (2001) 354.[71] R.G. Ross, P. Andersson, G. Backstrom, Nature (London) 290 (1981)322.[72] J.L. CoLett. 82[73] O. And[74] T. LoerChem.[75] R.O. P[76] X. Yu,[77] J. Klin[78] Y.P. Ha[79] W.F. G[80] D.D. K(1999)[81] A.I. KoRev. B[82] H. SchPhys. R[83] G.P. Jo[84] C.G. S(2004)[85] D. MinState P[86] M.M. KB 69 (2[87] D.D. KB 44 (1[88] K. Rott(1994)[89] O. And[90] A.J. Le[91] G.P. Jo[92] E. MayNetwo[93] I. KohlChem.[94] A. Kou[95] M.H. M[96] G.A. Bphys. 2[97] S. Klotshall, N[98] A. Hal[99] M. Sug[100] G.P. Jo[101] S. Ram[102] G.P. Jo[103] E. May[104] A. Hal[105] G.P. Jo[106] G.P. Jo(2000)[107] G.P. Jo[108] K. Hof[109] K. Hof4674.[110] O. And[111] G.P. Jo[112] G.P. Jo[113] P.V. Ho[114] G.P. Jo[115] G.P. Jo[116] W. Hag2743.[117] W. Hag545;W. Hage, A. Hallbrucker, G.P. Johari, E. Mayer, J. Chem. Phys. 104 (1996)414.[118] G.P. Johari, Phys. Chem. Chem. Phys. 7 (2005) 1091..P. Jo.W. D. Hillolecu.S. C. Oka10 (19. Kaw. Yam987).P. H35..P. Ha. Yam988).P. Jo.G. Shem..P. Jo.G. Scienc.M. K005).P. Jo. Fish.P. Jo. Ch94501. Alb. Pasolids.P. Jo. Hal.P. Jo994).J. Sk. Sch. Primffects.H. B.Y. Yransl..C. Kett. 43.J. Hature.P. Jof NATeries I007, p. Bos992).P. Jo. Loeayer,. Loerhys. 8.P. Ha23.. Bos. Yosh.V. S995).L. G997)hn, G.S. Nolas, V. Fessatidis, T.H. Metcalf, G.A. Slack, Phys. Rev.(1999) 779.ersson, G.P. Johari, J. Chem. Phys. 121 (2004) 3936.ting, C. Salzmann, I. Kohl, E. Mayer, A. Hallbrucker, Phys. Chem.Phys. 3 (2001) 5355.ohl, J. Non-Cryst. Solids 352 (2006) 3363.D.M. Leitner, J. Chem. Phys. 123 (2005) 104503.ger, Science 209 (1980) 271.nda, D.D. Klug, J. Phys. Chem. 92 (1988) 3323.iauque, J.W. Stout, J. Am. Chem. Soc. 58 (1936) 1144.lug, C.A. Tulk, E.C. Svensson, C.-K. Loong, Phys. Rev. Lett. 832584.lesnikov, J. Li, S.F. Parker, R.S. Eccleston, C.-K. Loong, Phys.59 (1999) 3569.ober, M.M. Koza, A. Tolle, C. Masciovecchio, F. Sette, F. Fujara,ev. Lett. 85 (2000) 4100.hari, J. Chem. Phys. 112 (2000) 8573.alzmann, E. Mayer, A. Hallbrucker, Phys. Chem. Chem. Phys. 65156.ceva-Sukarova, W.F. Sherman, G.R. Wilkinson, J. Phys. C: Solidhys. 17 (1984) 5833.oza, H. Schober, B. Geil, M. Lorenzen, H. Requardt, Phys. Rev.004) 024204.lug, E. Whalley, E.C. Svensson, J.H. Root, V.F. Sears, Phys. Rev.991) 841.ger, A. Endriss, J. Ihringer, S. Doyle, W.F. Kuhs, Acta Cryst. B 50644.ersson, A. Inaba, J. Chem. Phys. 122 (2005) 124710.adbetter, Proc. R. Soc. Lond., Ser. A 287 (1965) 403.hari, J. Phys. Chem. B 107 (2003), in: M.C. Bellissent-Funel, J.C. Dore (Eds.), Hydrogen-bondedrks, Kluwer, Dordrecht, 1994, pp. 355372., L. Bachmann, A. Hallbrucker, E. Mayer, T. Loerting, Phys. Chem.Phys. 7 (2005) 3210.chi, T. Kuroda, Nature (London) 344 (1990) 134.oore, R.L. Hudson, Astrophys. J. 401 (1992) 353.aratta, G. Leto, F. Spinella, G. Strazzulla, G. Foti, Astron. Astro-52 (1991) 421.z, J.M. Besson, G. Hamel, R.J. Nelmes, J.S. Loveday, W.G. Mar-ature 398 (1999) 681.lbrucker, E. Mayer, J. Phys. Chem. 91 (1987) 503.isaki, H. Suga, S. Seki, Bull. Chem. Soc. Jpn. 41 (1968) 2591.hari, J. Chem. Phys. 116 (2002) 8067., G.P. Johari, Philos. Mag. B 61 (1990) 299.hari, J. Chem. Phys. 102 (1995), J. Microsc. 160 (1985) 3.lbrucker, E. Mayer, G.P. Johari, J. Phys. Chem. 93 (1989) 7751.hari, J. Mol. Struct. 528 (2000) 249.hari, M. Beiner, C. Mcdonald, J. Wang, J. NonCryst. Solids 27858.hari, Chem. Phys. 258 (2000), G. Astl, E. Mayer, G.P. Johari, J. Phys. Chem. 95 (1991), A. Hallbrucker, E. Mayer, G.P. Johari, J. Phys. Chem. 93 (1989)ersson, Phys. Rev. Lett. 98 (2007) 57602.hari, A. Hallbrucker, E. Mayer, J. Chem. Phys. 95 (1991) 6849.hari, A. Hallbrucker, E. Mayer, J. Chem. Phys. 92 (1990) 6742.bbs, Ice Physics, Clarendon, Oxford, 1974.hari, A. Hallbrucker, E. Mayer, J. Chem. Phys. 95 (1991) 2955.hari, A. Hallbrucker, E. Mayer, J. Chem. Phys. 97 (1992) 5851.e, A. Hallbrucker, G.P. Johari, E. Mayer, J. Chem. Phys. 100 (1994)e, A. Hallbrucker, G.P. Johari, E. Mayer, J. Chem. Phys. 103 (1995)[119] G[120] D[121] NM[122] K[123] K1[124] S[125] O(1[126] Y4[127] Y[128] O(1[129] G[130] CC[131] G[132] CS[133] M(2[134] G[135] M[136] G[137] M0[138] C[139] WS[140] G[141] A[142] G(1[143] B[144] R[145] WE[146] L[147] AT[148] CL[149] RN[150] GoS2[151] L(1[152] G[153] TM[154] TP[155] Y6[156] L[157] Y[158] O(1[159] E(1hari, J. Chem. Phys. 122 (2005) 144508.avidson, R.H. Cole, J. Chem. Phys. 19 (1951) 1484., W.E. Vaughan, A.H. Price, M. Davies, Dielectric Properties andlar Behaviour, Von Nostrand, London, 1969.ole, R.H. Cole, J. Chem. Phys. 9 (1941) 341.da, M. Yao, Y. Hiejima, H. Kohno, Y. Kajihara, J. Chem. Phys.99) 3026.ada, J. Phys. Soc. Jpn. 57 (1988) 3694.amuro, M. Oguni, T. Matsuo, H. Suga, J. Phys. Chem. Solids 48935.anda, D.D. Klug, E. Whalley, J. Phys. Colloq. 48 (1987)nda, D.D. Klug, E. Whalley, Can. J. Chem. 66 (1988) 919.amuro, M. Oguni, T. Matsuo, H. Suga, J. Phys. Chem. Solids 49425.hari, J. Chem. Phys. 119 (2003) 2935.alzmann, I. Kohl, T. Loerting, E. Mayer, A. Hallbrucker, Phys.Chem. Phys. 5 (2003) 3507.hari, A. Hallbrucker, E. Mayer, Science 273 (1996) 90.alzmann, P.G. Radaelli, A. Hallbrucker, E. Mayer, J.L. Finney,e 311 (2006) 1758.oza, B. Geil, H. Schober, F. Natali, Phys. Chem. Chem. Phys. 71423.hari, J. Chem. Phys. 117 (2002), J.P. Devlin, J. Phys. Chem. 99 (1995) 11584.hari, J. Phys. Chem. B 102 (1998) 4711.onde, M. Brindza, V. Sadtchenko, J. Chem. Phys. 125 (2006).a, L.E. Busse, C.A. Angell, J. Chem. Phys. 92 (1990) 617.cheto, M.G. Parthun, A. Hallbrucker, G.P. Johari, J. Non-Cryst.171 (1994) 182.hari, Philos. Mag. 35 (1977) 1077.lbrucker, E. Mayer, G.P. Johari, Phil. Mag. B 60 (1989) 179.hari, G. Fleissner, A. Hallbrucker, E. Mayer, J. Phys. Chem. 984719.inner, J.J. Fahey, J. Geophys. Res. 68 (1963) 5595.rader, W. Dusdorf, Krist. Technol. 1 (1966) 59.ak, The Compacted States of Vitreous Silica, Studies of Radiationin Solids, Gordon and Breach, London, 1975.rixner, Mater. Res. Bull. 7 (1972) 879.armakov, Y. Yurchikov, V.A. Barinov, Phys. Met. Metall. (English) 52 (1981) 50.och, O.B. Cavin, C.J. McKamey, J.O. Scarborough, Appl. Phys.(1983) 1017.emley, A.P. Jephcoat, H.K. Mao, L.C. Ming, M.H. Manghnani,334 (1988) 52.hari, O. Andersson, in: S.J Rzoska, V.M. Mazur (Eds.), ProceedingsO ARW on Soft Matter under Exogenic Impacts, NATO ScienceI: Mathematics, Physics and Chemistry, vol. 242, Springer, Berlin,p., G.P. Johari, M. Oumezzine, J. Teixeira, Chem. Phys. Lett. 18813.hari, O. Andersson, Phys. Rev. B 70 (2004) 184108.rting, W. Schustereder, K. Winkel, C.G. Salzmann, I. Kohl, E.Phys. Rev. Lett. 96 (2006) 025702.ting, C.G. Salzmann, K. Winkel, E. Mayer, Phys. Chem. Chem.(2006) 2810.nda, J.S. Tse, D.D. Klug, E. Whalley, J. Chem. Phys. 94 (1991)io, G.P. Johari, J. Teixeira, Phys. Rev. Lett. 56 (1986) 460.imura, H. Kanno, Chem. Phys. Lett. 349 (2001) 51.talgorova, E.L. Gromnitskaya, V.V. Brazhkin, JETP Lett. 62356.romnitskaya, O.V. Stalgorova, V.V. Brazhkin, JETP Lett. 85109.G.P. Johari, O. Andersson / Thermochimica Acta 461 (2007) 1443 43[160] O.V. Stalgorova, E.L. Gromnitskaya, V.V. Brazhkin, A.G. Lyapin, JETPLett. 69 (1999) 694.[161] O. Mishima, H. Stanley, Nature 392 (1998) 164.[162] C.K. Majumdar, Solid State Commun. 9 (1971) 1987.[163] J.T. Bendler, M.F. Schlesinger, Macromolecules 18 (1985) 591.[164] G.W. Scherrer, Relaxations in Glasses and Composites, John Wiley, NewYork, 1986.[165] K. Schmidt, Rohr, H.W. Spiess, Phys. Rev. Lett. 66 (1991) 3020.[166] B. Schiener, R. Bohmer, A. Loidl, R.V. Chamberlin, Science 274 (1996)752.[167] M. Cicerone, F.R. Blackburn, M.F. Ediger, J. Chem. Phys. 102 (1995)471.[168] A. Plonka, Dispersive Kinetics, Kluwer, Dordrecht, 2001.[169] Y. Yoshimura, H.-K. Mao, R.J. Hemley, Chem. Phys. Lett. 420 (2006)503.[170] G.P. Johari, H.A.M. Chew, Philos. Mag. B 49 (1984) 647.[171] R.J. Hemley, L.C. Chen, H.K. Mao, Nature 338 (1989) 638.[172] M.A. Floriano, Y.P. Handa, D.D. Klug, E. Whalley, J. Chem. Phys. 91(1989) 7187.[173] G.P. Johari, Philos. Mag. B 80 (2000) 323.[174] M. Scheuermann, B. Geil, K. Winkel, F. Fujara, J. Chem. Phys. 124 (2006)224503.[175] S.K. Sikka, J. Phys. Condens. Matter 16 (2004) S1033.[176] T. Strassle, A.M. Saitta, S. Klotz, M. Braden, Phys. Rev. Lett. 93 (2004)225901.[177] J. Zhang, Y. Zhao, H. Xu, M.V. Zrelinskas, L. Wang, Y. Wang, T. Uchida,Chem. Mater. 17 (2005) 2817.[178] C.A. Perottoni, J.A.H. da Jornada, Science 280 (1998) 886.[179] G.P. Johari, E. Whalley, J. Chem. Phys. 75 (1981) 1333.[180] G.P. Johari, W. Pascheto, S.J. Jones, J. Chem. Phys. 100 (1994) 4548, andreferences therein.[181] G. Salvetti, E. Tombari, G.P. Johari, J. Chem. Phys. 102 (1995) 4987.[182] M. Born, K. Huang, Dynamic Theory of Crystal Lattices, Clarendon,Oxford, 1954.[183] J.S. Tse, D.D. Klug, C.A. Tulk, I. Swainson, E.C. Svensson, C.K. Loong,V. Shpakov, V.R. Belosludov, R.V. Belosludov, Y. Kawazoe, Nature 400(1999) 647.[184] J. Perez, C. Mai, R. Vassoile, J. Glaciol. 21 (1977) 361.[185] C.A. Coulson, D. Eisenberg, Proc. Roy. Soc. (Lond.) A 291 (1966)454.[186] G.P. Johari, S.J. Jones, J. Chim. Phys. Phys.-Chim. Biol. (Fr.) 82 (1985)1019.[187] M. Nagahama, H. Suga, O. Andersson, Thermochim. Acta 363 (2000)165.[188] S. Furukawa, M. Nagahama, H. Suga, High Temp.-High-Press. 32 (2000)397.[189] T.G. Richards, G.P. Johari, Philos. Mag. B 58 (1988) 445.[190] T. Loerting, I. Kohl, C. Salzmann, E. Mayer, A. Hallbrucker, J. Chem.Phys. 116 (2002) 3171.[191] O. Yamamuro, I. Tsukushi, T. Matsuo, J. Mol. Cryst. Liquid Cryst. 277(1996) 205.[192] H. Suga, J. Phys.: Condens. Matter 15 (2003) S775.[193] M.H. Bhat, I. Peral, J.R.D. Copley, C.A. Angell, J. Non-Cryst. Solids 352(2006) 4517.[194] R. Rao, T. Sakuntala, B.K. Godwal, Phys. Rev. B 65 (2002) 054108.[195] R. Rao, T. Sakuntala, A.K. Arora, S.K. Deb, J. Chem. Phys. 121 (2004)7320.[196] M. Koza, H. Schober, A. Tolle, F. Fujara, T. Hansen, Nature 397 (1999)661.[197] T. Loerting, I. Kohl, W. Schustereder, K. Winkel, E. Mayer, Chem. Phys.Chem. 7 (2006) 1203.[198] R.J. Nelmes, J.S. Loveday, T. Strassle, C.L. Bull, M. Guthrie, G. Hamel,S. Klotz, Nat.-Phys. 2 (2006) 414.[199] Y. Suzuki, Phys. Rev. B 70 (2004) 172108.Vibrational and relaxational properties of crystalline and amorphous icesIntroductionExperimental methods for measurements at high pressuresVibrational properties of the crystalline and amorphous iceThe thermal conductivity of crystalline and amorphous solid waterHeat capacity of waters high-density amorphRelaxation properties of the amorph and conversion to ultraviscous waterPreparation of waters amorphous solid and glassy statesCalorimetric behaviour, relaxation and glass-softeningDielectric relaxation of amorphous solid and glassy states of waterCharacteristic changes during pressure-amorphization of iceTime-, pressure- and temperature-dependence of the extent of amorphizationMechanism of pressure-amorphizationThermodynamics and kinetics of pressure collapsed amorph and of ultraviscous waterSummary and concluding remarksAcknowledgementsReferences


View more >