Sound velocities and elastic properties of Fe-bearing

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. B9, PAGES 20,819-20,825, SEPTEMBER 10, 1998 Sound velocities and elastic properties of Fe-bearing wadsleyite and ringwoodite S. V. Sinogeikin Department of Geology, University of Illinois, Urbana T. Katsura Bayerisches Geoinstitut, Universit it Bayreuth, Bayreuth, Germany J. D. Bass Department of Geology, University of Illinois, Urbana Abstract. The sound velocities and single-crystal elastic moduli of I phase (wadsleyite) and ¾ phase (ringwoodite) of (Mg,Fe)2SiO4 with Fe/(Fe+Mg) ratios of and --4).09, respectively, have been determined at ambient conditions by Brillouin spectroscopy. Both compressional and shear wave aggregate velocities decrease with increasing Fe content in both phases, but the magnitude of this decrease is different for the two phases. The adiabatic bulk modulus, Ks, of Fe-bearing I -Mg2SiO4 (Ks = 170_+2 GPa) is indistinguishable from that of the Mg end-member within experimental uncertainty, whereas Ks of ¾-(Mg,Fe)2SiO4 increases rapidly with increasing iron content. The shear moduli of both phases decrease with increasing Fe content. Our measurements indicate that the velocity and impedance contrasts between olivine and [ -(Mg,Fe)2SiO4 are independent of Fe content for Mg-rich compositions, but the contrast for the ¾-(Mg,Fe)2SiO4 transition increasesignificantly with increasing Fe content. The new data support a previous estimate of 40ñ10% for the olivine content of the upper mantle and suggest that less than 50% (Mg,Fe)2SiO4 is sufficiento account for the observed impedance contrasts at depths of both 410 'km and 520 'km. Unless the effect of Fe on elastic properties is accounted for, it is difficult to account for both the 410 and 520 km discontinuities with a single olivine content. 1. Introduction impedance contrasts across phase transitions. This, in turn, will alter estimates of the olivine fraction of the upper mantle. The polymorphic phase transformations of olivine, We previously reported measurements of the velocities and (x-(mg,fe)2sion, to the higher-density 13 phase (wadsleyite) elastic properties of natural, polycrystalline ¾ phase containing and from 13 phase to ¾ phase (ringwoodite) are widely viewed 25 mol øa Fe [Sinogeikin et al., 1997]. From velocity as likely causes of rapid, or discontinuous, changes in seismic measurements on polycrystalline samples, one obtains excellent wave speeds at about 410 and 520 km depth. Within the values of the bulk and shear moduli, but such measurements context of a chemically homogenous model of the mantle, the are limited in that they do not yield any direct information on magnitudes of observed velocity increases can be used to the individual elastic moduli, Co.. In this report, we present the calculate the olivine content of the mantle, provided that first Brillouin scattering measurements of acoustic velocities laboratory elasticity data on the relevant phases are available and single-crystal elastic moduli of 13- and ¾-(Mg,Fe)2SiO4 [e.g., Weldnet et al., 1984; Zha et al., 1998]. Previous with Fe contents thoughto be typical of actual mantle values. attempts to constrain the mantle olivine content by this These measurements allow us to qualitatively assess the effect approac have relied primarily on data for the pure Mg endof Fe on velocity and impedance contrasts across the (x-- 13 members, because the dependence of sound velocity on Fe and ¾ phase transitions. content for the 13 and ¾ polymorphs had not been measured for compositions of direct relevance to the mantle. Fe changes not only the absolute values of velocities but also the velocity and 2. Experiment Single crystals of 13- and ¾-(Mg,Fe)2SiOn were synthesized from San Carlos olivine ((Mg0.9 nfe0.086)2sion) at-18 GPa and Now at Okayama University, Institute tbr Study of Earth Interior, 1600øC with a uniaxial split sphere apparatus at the Misasa, Tottori, Japan. 2Also at Bayerisches Geoinstitut, Universitat Bayreuth, Bayreuth, Bayerisches Geoinstitut, Universitat Bayreuth. Owing to the Germany. thermal gradient inside the pressure cell, three distinctive zones of crystals were obtained: [3, 13-%', and ¾. The long Copyright 1998 by the American Geophysical Union duration of the experiments (60 min) allowed the growth of large anhedral crystals with maximum dimensions up to 500 Paper number 98JB grn and a slight redistribution of Fe between 13 and ¾ phases /98/98JB Chemical compositions were determined by electron probe 20,819

2 20,820 SINOGEIKIN ET AL.: ELASTICITY OF WADSLEYITE AND RINGWOODITE Table 1. Lattice Parameters, Unit Cell Volumes, and Densities of [3 and ¾ Phases of (Mg,Fe)2SiO4 Used in This Study to yellowish-green along [100] and dark bluish-green along [010]). The refractive indices were measured in yellow (-589 nm) and green (NS 14 nm) light using index-matching fluids. 13-(Mgo.92Feo.08)2SiO47-(Mgo.9 Feo.09)2SiO4 Values for the ¾ phase (Table l) are typical for spinals of this composition [Ringwood and Major, 1970]. Refractive indices ao, )k 5.705(2) 8.073(6) of the [3 phase were measured along the crystallographic a (n ) 60, A (4) and b (n ) axes. The values for n were inferred from the 2V Co,,/k 8.258(3) angle (Table 1). Several plates with parallel faces and a thickness of Vo 539.4(4) 526.2(4) p, g/cm (3) 3.701(5) prn were prepared from larger crystals within the monophase zones. The 13 and ¾ phases are exceptionally brittle, and it is Refractive Indices difficult to grind and polish faces without the samples fracturing. Therefore surfaces were prepared in the most convenient directions which required the minimum amount of //589 //514 //589 //514 grinding. Because the prepared surthces had arbitrary crystallographic orientations, very few of our velocity data a 1.711(3) 1.716(3) 1.731(2) 1.740(2) were obtained from pure mode directions. However, (5) 1.723(5) characterization of any crystallographic plane of cubic ¾ phase ¾ 1.727(3) 1.734(3) is sufficiento refine all three independent elastic constants (Co). Because the [3 phase is characterized by nine independent Numbers in parentheses represent 1 s.d. uncertainties in the last Co., measurements within a single crystallographic plane digit. Big uncertainties in densities are due to variations in constrain all nine constants only if the plane intersects the composition. crystal axes at subequal angles [Zha et al., 1997]. Brillouin scattering measurements were carried out using the Brillouin spectrometer illustrated in Figure I. Although microanalyses, yielding XF½ = [Fe]/([Fe]+[Mg]) of 0.090(5) and 0.075(10) (values in parentheses represent 1 s.d. this system has been used in all our recent work, a description of the spectrometer in its current configuration is presented uncertainties the last digit) for ¾ and [3 phases, respectively. here for the first time. The most important difference from an The lattice parameters of all samples were determined with a four-circle X ray diffractometer (Table 1). The 13 and ¾ phases are characterized by distinctive colors and are easily separated. The cubic ¾ phase is isotropic and has a deep blue color, while the orthorhombic [3 phase is earlier description [Bass, 1989] is use of the piezoelectrically scanned plane-parallel tandem Fabry-Perot interferometer [Sandercock, 1982], in a six-pass tandem configuration to analyze the scattered light. The light scattered at 90 ø from the incident beam is biaxial with a 2V angle of ø, displays significant typically collected by a f-3.5 lens. A small fraction of the birefringence (15 = 0.016(3)), and strong pleochroism (yellow initial beam ( 4øA) is redirected by a beam splitter and used to Eulerian Cradle Polarization -, 2 - FL BS Rotator 0 E Arlaser II G-T Polarizer II L L.cu I I IAnalyzer II PRl I I [ 6-pa tandem I I I U Fab -Perot I I I I / ' Intederometer I I FPC I1 I] I1 olloll [;... 1osc OMM L r %MI I II I1> / Figure 1. Schematic diagram of the Brillouin spectrometer with six-pass Fabry-Perot interferometer. Abbreviations are G-T, Glan-Thompson; BS, beam splitter; FL, focusing lens; CL, collecting lens; L, lens; M, mirror; OMM, optomechanical modulator; SF, spatial filter; R, retroreflector; PR, prism; PCU, phononcounting unit; FPC, Fabry-Perot controller; OSC, oscilloscope; MCA, multichannel analyzer; PC, personal computer.

3 SINOGEIKIN ET AL.' ELASTICITY OF WADSLEYITE AND RINGWOODITE 20, ß I ' I ' I ' I ' ß I ' I '! ' polarization, and 0 is the scattering angle. For a platelet, or symmetric scatteringeometry [Whit, fiem et al., 1976], as used in this study, uncertainties from the refractive index and scattering angle are reduced, and the above equation reduces to 140 V= AtoZ/ / The Brillouin spectra typically had a very high signal-tonoise ratio. In arbitrary (h,k,l) crystallographic directions with appreciable shear anisotropy, three peaks corresponding c 80 to one compressional and two shear modes were observed c: ,W (Figure 2). The Brillouin data for both 13 and ¾ phases were collected in single crystallographic planes over an angular range of 160 ø to 220 ø, with an average angular increment of 15 ø between data points (Figure 3). I, I, I 0 ' ß I ß I, I, I. The Co. for 13 and ¾ phases are given in Tables 2 and 3, along with those of the pure magnesian phases for comparison. Brillouin shift, GHz Aggregat elastic moduli for 13- and ¾-(Mg,Fe)2SiO 4 calculated using the Voigt-Reuss-Hill (VRH) averaging scheme are given Figure 2. Typical Brillouin spectrum of single-crystal in Tables 4 and 5 along with acoustic data from the literature 7-(Mg,Fe)2SiO4. Peaks marked as P, S 1, and S2 correspond to compressional and two shear acoustic modes, respectively. for these phases. 3. Effect of Fe on Elastic Properties stabilize the interferometer. A shutter or acousto-optic of 13 and ¾ Phases modulator can be used to mechanically suppresstrong elastic Our new data, combined with Brillouin measurements on scatteringenerated on sample surfaces and inside a sample due to irregularities. Fluorescence was filtered out by use of a natural polycrystalline ringwoodite from the Catherwood dispersing prism. All optics to control the polarization of light meteorite (¾-0Mg0.7sFe0.25)28iO4) [Sinogeikin et al., 1997]), put are as previously described [Bass, 1989]. Either a photo- strong constraints on the Fe dependence of Ks for [3- and ¾- multiplier tube (PM or solid state detector is used to detect (Mg,Fe)2SiO4 in the Mg-rich end, which is most relevant to the signal. mantle mineralogy. The bulk modulus of ¾-(Mg,Fe)2SiOa increases from 184 to 205 GPa as the Fe content increases The scattered light consists of an elastically scattered component with frequency to as well as Brillouin scattered from 0 to 430% and can be described by the equation components that have undergone a frequency shift Ato due to K = 184(2) + 36(5)*XFe (Figure 4). At higher Fe contents, the interaction with thermally generated phonons in the sample. behavior of the bulk modulus is not well defined due to a great The velocity V of an acoustic wave is calculated from the deal of scatter in reported values (Figure 4). In contrast, the relation: shear modulus of ¾ phase shows a modest decrease with V = o3. increasing Fe content and can be described as 2 n sin(o/2) = 120(2) - 25(8)*XFe.. Therefore Ks of ¾ phase shows a stronger dependence on Fe content, and g shows a weaker where L is the laser wavelength (514.5 nm), n is the refractive dependence, than is inferred from acoustic measurements on index of the sample for a particular phonon direction and the Fe2SiO4 end-member [Akimoto, 1972; Liebermann, 1975; Fukizawa and Kinoshita, 1982; Rigden et al., 199 lbl. 11l'' '''' ' ' /' ' ' ' ' ' ' phase f ¾ phase, ß vp _ ' - Table 2. Single-Crystal Elastic Moduli of 13-0Vlg,Fe)2SiO4 C, GPa (Mg.92Fe.08)2SiO4 a b Mg2SiO4 Mg2SiO4 c Ms1 I Ms1 ] I.! ß i I /I I I I/ Angle in scauering plane Figure 3. Measured acoustic velocities as a function of angle in the scattering plane for both 13 and ¾ phases. Solid lines show acoustic velocities calculated from the best fit singlecrystal elastic moduli. I I 348 (3) 360 (6) 371 (8) (3) 383 (4) 368 (7) (3) 273 (5) 272 (6) (2) 112 (2) 111 (4) (2) 118 (4) 123 (4) (2) 98 (4) 103 (4) 12 78(3) 75(9) 66(5) (3) 110 (6) 95 (5) (3) 105 (9) 105 (5) Numbers in parentheses represent 1 s.d. uncertainties the last digit. a This work. b Sawamoto et al. [1984]. c Zha et al. [ 1997].

4 20,822 SINOGEIKIN ET AL.: ELASTICITY OF WADSLEYITE AND RINGWOODITE Table 3. Single-Crystal Elastic Moduli of ¾-(Mg,Fe)2SiO4 Cij, GPa ß a (Mg.9 Fe0.09)2S On b Mg2SiOn (3) 327 (4) (1) 126(2) (2) 112(3) 13-Mg2SiO4 yields the following tentative relationship between it and Fe content: it = 112(2) - 40(10)*XFe GPa. This is consistent with our preliminary measurements on [3 phase with XFe = 0 and 0.21 [Jackson et al., 1997]. Additional measurements on samples of 13 phase with higher Fe content, as well as on pure Mg [3 phase, are required to resolve the relatively large uncertainties the shear modulus-composition relationship for 13-(Mg,Fe)2SiO4 solid solutions. Numbers in parentheses represent 1 s.d. uncertainties the last digit. 4. Effect of Fe on Velocity Contrasts Across tx-13 a This work. b Weidner et al. [ 1984]. and 13-¾ Phase Transitions The acoustic velocities for both [3 and ¾ phases decrease with increasing Fe contents, but the magnitude of the decrease is not equal for these two phases (Tables 4 and 5 and Figure The largest uncertainty in the modulus-composition 5). As a result, the velocity contrast (AV _ /V ) between 13 relations for 13-(Mg,Fe)2SiO4 is the large scatter in the elastic and ¾ phases increases with increasing Fe content (Table 6 and moduli of the pure Mg end-member, [3-Mg2SiO4. The reported Figure 6). However, it is noteworthy that although the bulk moduli for [3-Mg2SiO4 range from 163 GPa [Gwanmesia absolute velocity difference for the tx --> 13 transition decreases et al., 1990] to 174(3) GPa [Sawamoto et al., 1984]. These with im:reasing Fe content, the relative velocity contrast, or extreme values are consistent with either a positive or negative percent change of velocity across the phase transition, remains dependence of Ks on Fe content when taken together with our nearly constant (Table 6 and Figure 6). The same is true for measurements on Fe-bearing samples. However, the most impedance contrasts even taking into account appreciable recent high-pressure studies of single-crystal [Zha et al., 1997] uncertainties elastic moduli of [3-Mg2SiO4. and polycrystalline [Li et al., 1996] [3-Mg2SiO4 yield Ks = 170 GPa, which coincides with the average of all measured bulk 5. Discontinuities moduli to date. We therefore adopt a value of Ks = 170(3) GPa for [3-Mg2SiO4, because it is consistent with the most The velocity increases across seismic discontinuities, recent (presumably most accurate) measurements on this especially the 410 km discontinuity, are often used to infer the material and is consistent with the global average of all extant olivine content of an assumed isochemical mantle by measurements. This preferred value is equal to our measured comparison with the velocity changes across phase transitions value of Ks for [3-(Mg,Fe)2SiO4 and suggests that Ks of in olivine [Dufj et al., 1995; Weidner et al., 1984; Zha et al., [3-(Mg,Fe)2SiO4 is roughly independent of Fe content, within 1997, 1998]. Li et al. [1996] and Zha et al. [1997] measured experimental uncertainty. Preliminary Brillouin measurements the acoustic velocities and elastic moduli of ct- and 13-Mg2SiO4 on [3-Mg2SiO4 and [3-(Mg0.79Fe0.2 )2SiO4 by Jackson et al. to pressures exceeding those of the 410 km discontinuity at [1997] support the suggestion that Ks of 13 phase is room temperature and found that the pressure derivatives of the independent on Fe content, while la decreases rapidly with elastic moduli of these two phases are virtually identical and increasing Fe. that the velocity contrast between ct- and 13-Mg2SiO4 decreases The shear modulus of 13-Mg2SiO4 is similarly uncertain. only slightly with increasing pressure. Zha et al. [1998] Single-crystal work on -Mg2SiO4 yielded values of g = 114- further calculated the proportion of olivine in the upper mantle 115 GPa [Zha et al., 1997; Sawamoto et al., 1984], whereas to be 30-50% depending on the choice of temperature ultrasonic measurements on polycrystalline samples yielded derivative of the shear modulus for 13 phase and assuming that values in the range 108 to 110 GPa [Gwanmesia et al., 1990; the velocity contrast does not depend on the Fe content of Liet al., 1996]. Our VRH average for 13-(Mg,Fe)2SiO4 with olivine. 7.5(1.0)% Fe is 108 (2) GPa. Combining the results of our Many authors have noted that the dependence of elastic new measurements with the average of all measurements for moduli on Fe content varies widely for different minerals [e.g., Table 4. Aggregate Elastic Properties (VRH Average) of [3-(Mg,Fe)2SiO4 13-Mg2SiO4 (Mg.92Fe.08)2SiO4 Sawarnoto Gwanrnesia Li et al Zha et al et al. [1984] et al. [1990] [1996] [1997] This Study p, g/cc [ Ks, GPa ( 1 ) 170(2), GPa (1) 115(2) V/,, km/s 9.66 [9.55] [9.51] [9.63] Vs, km/s 5.71 [5.65] [5.58] [5.74] 3.57(3) 170(3) 08(2) 9.38(4) 5.50(3) Numbers in parentheses represent I s.d. uncertainties the last digit. Density in brackets calculated from reported unit cell volume.

5 S1NOGEIK1N ET AL.' ELASTICITY OF WADSLEYITE AND R1NGWOODITE 20,823 Table 5. Aggregate Elastic Properties (VRH Average) of 7- (Mg,Fe)2SiO lo Mg2SiO4 a (Mg.9 Fe.09)2SiO4 b (Mg.75Fe.25)2SiO4 ½ p, g/co (5) 3.878(5) Ks, GPa (3) 193(3) la, GPa (2) 113(2) Vv, km/s (2) 9.39(4) Vs, km/s (1) 5.40(3) 8 Vp 3 Numbers in parentheses represent I s.d. uncertainties in the last digit. a Weidner et al. [ 1984]. b This work. c Sinogeikin et al. [1997]. Hazen, 1993]. For example, the bulk modulus of x- (Mg,Fe)2SiO4 is almost independent of Fe content, while the bulk modulus of the 7 phase increases rapidly with increasing Fe. This underscores that the effect of Fe on elastic properties, and hence on inferred mantle properties, can be strongly dependent on structure type and is difficult to predict from elasticity systematics. 240 C922 o 200 ø180 m Fe/Fe+Mg, % Figure 5. Sound velocities for x-,[3- and 7-(Mg,Fe)2SiO4 as a function of Fe content. Solid squares, this study for both 13 and 7 phases. Open symbols, data for 7 phase; solid circles and diamonds, data for 13 phase. Sources of data are the same as on Figure 4, with the addition of (15) Fukizawa and Kinoshita[1982] and (16)Rigden et al. [199 lb]. Straight line marked "O1" is a linear fit to velocities in olivine from data compilation of Bass [ 1995]. Q_ E m Fe/Fe+Mg, % Figure 4. Bulk and shear moduli of 13 and 7 polymorphs of (Mg,Fe)2SiO4 as a function of Fe content. Solid squares, data from this study for both 13 and 7 phases. Diamonds, moduli calculated from sound velocities for 7 phase. Stars, static compression results (for clarity, symbols are offset to the left for the Mg end-member and to the right for the Fe endmember). Solid circles, moduli from acoustic studies for 13 phase. Sources of data are (1)Mao et al. [1969];(2)Mizutani et al. [1970];(3)Akimoto [1972]; (4)Liebermann[1975]; (5) Mizukami et al. [1975]; (6) Weidner et al. [1984]; (7) Kuskov [1985]; (8) Hazen [1993]; (9) Meng et al. [1994]; (10) Sinogeikin et al. [1997]; (11) Sawamoto et al. [1984]; (12) Gwanmesia et al. [1990]; (13) Liet al. [1996]; (14) Zha et al. [ 1997]. Our measurements on 13 phase suggesthat even though the shear modulus and acoustic velocities of x and 13 change significantly with increasing Fe content, the percentage change in Vp and Vs across the x- 13 phase transition (that is, the velocity contrasts) is approximately independent of Fe content. This lends supporto the calculations of the mantle olivine content by Zha et al. [ 1998], which were based upon velocities measured in the pure Mg system. They concluded that the amount of olivine required to produce the seismic velocity discontinuity at 410 km depth is 30-50%. Our new data may also bear on the interpretation of the 520 km discontinuity. This discontinuity less well constrained by short-period waveform data, which implies that it may not be a sharp (e.g., less than 10 km [Benz and Vidale, 1993]) feature, but is perhaps better viewed as a velocity gradient up to km wide [Revenaugh and dordan, 1991]. Although there is evidence suggesting that this discontinuity a global feature [Flanagan and Shearer, 1998], the lack of routine observations of a 520 km discontinuity in refraction studies suggests that the bulk of the impedance change might occur in density rather than velocity [Shearer, 1996]. Shear and compressional impedance contrasts for this discontinuity are similar with average values of about 3% [Shearer, 1990, 1996].

6 .. 20,824 SINOGEIKIN ET AL.' ELASTICITY OF WADSLEYITE AND RINGWOODITE Table 6. Velocity and Impedance Contrasts for at-j3 and Transitions as a Function of Fe Contents 1996; Zha et al., 1997]. Although high-p measurements of the elastic moduli for 3' phase were made up to 3 GPa [Rigden et al., 1991a, 1992], these fall short of actual transition zone x--> 13 Transition 13 --> 7 Transition pressures ( GPa). However, the measurements of Rigden et al. [1991a, 1992] strongly suggesthat pressure Velocity contrast, % derivatives of 13 and ¾ phases are similar, perhaps identical. The temperature derivatives of Ks ((OKs/OT)/, = GPa/K) Vv 11.5(12)+ 1.0(10)*Xve 2.4(8)+9.1 (40)*Xw were inferred from high-temperature static compression Vs 12.6(14)+ 1.9(10)*Xw 2.1(12)+8. l(40)*xw experiments ((OKr/OT), = GPa/K for both 13 [Fei et al., Impedance contrast, % 1992] and ¾ [Meng et al., 1994] phases). Temperature derivatives of the shear modulus are not known for either 13 or ApAVv 20.2(20)-0.3(10)*Xw 5.2(17)+9.7(40)*Xw ¾ phases and must be assumed from mineral physics ApAVs 21.4(24)+0.7(10)*Xw 4.9(28)+8.7(40)*Xw systematics [Anderson, 1988]. Assuming that 13 and ¾ have equal values for P and T derivatives of elastic moduli Numbers in parentheses 1 s.d. uncertainties the last digit. (OKs/OP = 4.3, OWOP = 1.4, OKs/OT = , OWOT = ), we calculate compressional and shear impedance contrasts of 5.5% for the 13 --) ¾ phase transition, similar to values obtained for ambient conditions. Thus, when we account for P and T effects, the magnitude of the 520 km The first single-crystal measurements of the elastic discontinuity found to be generally compatible with an properties of 13- and ¾-Mg2SiO4 at ambient conditions olivine content of less than 50%. The velocity and impedance [Weidner et al., 1984; Sawamoto et al., 1984] implied that contrast would have to be diminished from the values reported there is no discontinuity associated with the 13 -> ¾ transition, here to be dearly consistent with a pyrolite, olivine-rieh because the properties of 13 and ¾ phases are very similar. In mantle. Anything which tends to increase the elasticity more recent work, Rigden et al. [ 1991 a, 1992] measured V, contrast across the 13 --) ¾ phase change will lower the implied and Vs of ¾-Mg2SiO4 to 3 GPa at room temperature and found olivine content of the transition zone. For example, if the it unlikely that any seismically observable velocity pressure derivatives of the shear and/or bulk moduli of ¾ phase discontinuity at about 520 km depth could be attributed to the are larger than the values we used [Rigden et al., 199 l a], the 13 --> ¾ phase transition. Although the impedance contrast impedance contrast will increase with depth. This will lower might be sufficient for the 13 -> ¾ transformation to be observed the amount of olivine needed to explain the 520 km in seismic reflection studies at near-normal incidence discontinuity. [Shearer, 1990], it requires an olivine content well in excess of Most of the impedance change inferred for the 520 km 50%. These values may raise a contradictory situation, discontinuity occurs in density rather than velocity [Rigden et whereby the velocity change at 410 km discontinuity implies less than 50% olivine in the mantle [e.g., Duff5, et al., 1995; Zha et al., 1997], and the velocity change near 520 km suggests an olivine content of greater than 50%. Our measurementsuggest impedance contrasts for the 25 I ' I ' I ' I ' I ' I --> ¾ (1V[go.9Feo. l)2sio4 phase transition of 6.1% and 5.7% at ApVs ambient conditions for shear and compressional contrasts, respectively. These values are substantially larger than those o 20 for the Fe-free phases. As a result, the mantle olivine content implied by the ¾ transition is reduced to 40+10%, Impe d ance contrast o ( --13 transition suggesting that a value of 40-50% olivine can account for both Velocity contrast the 410 and 520 km discontinuities. We emphasize that accounting for the effect of Fe makes it easier to rationalize the AVs o velocity structure of the transition zone with a single olivine content. However, the above estimate of olivine content is o 10 highly uncertain because it is based on measurements at room ApV P and T only. I mpe,,clance contrast Although the velocity contrasts at ambient conditions are often a surprisingly good guide to those at high P-T conditions I;pv [Dufj et al., 1995], a complete calculation of the velocity and n impedance contrast must explicitly account for effects of P, T, elocity contrast and partitioning of Fe between mantle phases. We attempted 0, I, I, I, I, I to account for the P and T effects using the formalism of Duff5, and Anderson [1989] and describe here the sources of data Fe content, mol. % used and our results. Unfortunately, extrapolations of acoustic velocities to mantle P and T are restricte due to either large Figure 6. Velocity and impedance contrasts across the at to uncertainties P and T derivatives of elastic moduli for highand 13 to ¾ phase transitions as a function of Fe content. pressure phases or due to a complete lack of measured values. Velocities are calculated assuming linear dependence of Thus far, the P derivatives of elastic moduli for olivine and 13 density and elastic moduli on Fe content. Note that the lines phase have been measured to transition zone pressures, for Fe content exceeding 30% are hypothetical since yielding values of Ks ' = , and p' = [Li et al., (Mg,Fe)2SiO4 with higher Fe contents is not stable.

7 SINOGEIKIN ET AL.: ELASTICITY OF WADSLEYITE AND RINGWOODITE 20,825 al., 1991a, 1992], consistent with the lack of routine Liebermann, R.C., Elasticity of olivine (o,), beta (13), and spinel (¾) observations of a 520 km velocity discontinuity in refraction polymorphs of germanates and silicates, Geophys. d. R. Astron. Soc., 42, , studies [Shearer, 1996]. In addition, phase equilibria studies Mao, H. K., T. Takahashi, W.A. Bassett, J.S. Weaver, and S. at high P and T [Irifune et al., 1986] suggesthat there is Akimoto, Effect of pressure and temperature on the molar volumes appreciable partitioning of Fe into ¾ phase in an isoehemical of wiistite and of three (Fe, Mg)2SiO4 spinel solid solutions, d. mantle and that this will further increase the impedance Geophys. Res., 74, , contrast of the 13 --> ¾ transition. Meng, Y., Y. Fei, D.J. Weidner, G.D. Gwanmesia, and J. Hu, The uncertainties in our calculations and the uncertainties in Hydrostati compression of ¾-Mg2SiO4 to mantle pressures and 700 K: Thermal equation of state and related thermoelastic seismic estimates of the impedance contrasts across the 520 properties, Phys. Chen Miner., 21, , km discontinuity do not yet allow the (]V[g,Fe)2SiO4 content of Mizukami, S., A. Ohtani, and N. Kawai, High-pressure x-ray the mantle to be determined with a high degree of confidence. diffraction studies on [3- and ¾-Mg2SiO4, Ptngs. Earth Planet.. The available data do suggest, however, that less than 50% biter., 10, , Mizutani, H., Y. Hamano, Y. Ida, and S. Akimoto, Compressional 0VIg,Fe)2SiO4 is sufficient to account for the observed wave velocities of fayalite, Fe2SiO4 spinel, and coesite, d. Geophys. impedance contrasts at depths of both 410 km and 520 km. Res, 75, , Revenaugh, J., and T.S. Jordan, Mantle layering from ScS Acknowledgments. We thank H.W. Green II, R.C. Liebermann, reverberations, 2, The transition zone, d. Geophys. Res., 96, and T. Yagi for thorough review and helpt l comments. This research 19,763-19,780, was partially supported by the National Science Foundation, by the Rigden, S.M., G.D. Gwanmesia, J.D. Fitz Gerald, I. Jackson, and R.C. Alexander von Humboldt Stiftung through a Fellowship to J.D.B., and Liebermann, Spinel elasticity and seismic structure of the transition by the Bayerisches Geoinstitut. zone of the mantle, Nature, 354, , 1991a. Rigden, S.M., I. Jackson, R.C. Liebermann, and A.E. Ringwood, References Elasticity of germanate and silicate spineis at high pressure, d. Geophys. Res., 96, ,006, 1991b. Akimoto, S.I., The system MgO-FeO-SiO2 at high pressures and Rigden, S.M., G.D. Gwanmesia, J.D. Fitz Gerald, I. Jackson, and R.C. temperatures: Phase equilibria and elastic properties, Liebermann, Progress in high-pressure ultrasonic interferometry, Tectonophysics, 13, , the pressure dependence of elasticity of Mg2SiO4 polymorphs and Anderson, D.L., Temperature and pressure derivatives of elastic constants with application to the mantle, d. Geophys. Res., 93, , Bass, J.D., Elasticity of grossular and spessartite garnets by Brillouin spectroscopy, d. Geophys. Res., 94, , Bass, J.D., Elasticity of minerals, glasses, and melts, in Mineral Physics and Crystallography: A Handbook of Physical Constants, AGU Reft Shelf, vol. 2, edited by T.J. Ahrens, pp , AGU, Washington, D.C., Benz, H.M., and J.E. Vidale, Sharpness of upper-mantle discontinuities determined from high-frequency reflections, Nature, 365, , Duffy, T.S., and D.L. Anderson, Seismic velocities in mantle minerals and the mineralogy of the upper mantle, d. Geophys. Res., 94, , Duffy, T.S., C.S. Zha, R.T. Downs, H.K. Mao, and R.J. HemIcy, Elasticity of forsteritc to 16 GPa and the composition of the upper mantle, Nature, 378, , Fei, Y., H.K. Mao, J. Shu, G. Parthasarathy, W.A. Bassett, and J. Ko, Simultaneous high-p, high-t X ray diffraction study of (Mg, Fe)2SiOn to 26 GPa and 900 K, d. Geophys. Res., 97, , Flanagan, M.P., and P.M. Shearer, Global mapping of topography on constraints on the composition of the transition zone of the Earth's mantle, in High-Pressure Research: Application to Earth and Planetary Sciences, Geophys. Monogr. Set., vol. 67, edited by Y. Syono, and M. H. Manghnani, pp , AGU, Washington, D.C., Ringwood, A.E., and A. Major, The system Mg2SiO4-Fe2SiO4 at high pressures and temperatures, Phys. Earth Planet. Inter., 3, , Sandercock, J.R., Trends in Brillouin scattering: Studies of opaque materials, supported films, and central modes, in Topics in Applied Physics, vol. 51, Light Scattering in Solids 111: Recent Results, edited by M. Cardona and G. Guntherodt, Springer-Verlag, New York, Sawamoto, H., D.J. Weldnet, S. Sasaki, and M. Kumazawa, Singlecrystal elastic properties of the modified spinel (beta) phase of magnesium orthosilicate, Science, 224, , Shearer, P.M., Seismic imaging of upper-mantle structure with new evidence for a 520-kin discontinuity, Nature, 344, , Shearer, P.M., Transition zone velocity gradients and the 520-km discontinuity, d. Geophys. Res., 101, , Sinogeikin, S.V., J.D. Bass, A. Kavner, and R. Jeanloz, Elasticity of natural majoritc and ringwoodite from the Catherwood meteorite, Geophys. Res. Lett., 24, , transition zone velocity discontinuities by stacking SS precursors, Weldnet, D.J., H. Sawamoto, S. Sasaki, and M. Kumazawa, Singled. Geophys. Res., 103, , Fukizawa, A., and H. Kinoshita, Shear wave velocity jump at the crystal elastic properties of the spinel phase of Mg2SiO4, d. Geophys. Res., 89, , olivine-spinel transformation in Fe2SiO4 by ultrasonic Whitfield, C.H., E.M. Brody, and W.A. Bassett, Elastic moduli of measurements in situ, J. Phys. Earth, 30, , Gwanmesia, G.D., S. Rigden, I. Jackson, and R.C. Liehermann, Pressure dependence of elastic wave velocity for [3-Mg2SiO4 and the composition of the Earth's mantle, Science, 250, , Hazen, R., Comparative compressibilities of silicate spineis: Anomalous behavior of (Mg, Fe)2SiO4, Science, 259, , Irifune, T., T. Sekine, E.A. Ringwood, and W.O. Hibberson, The NaCI by Brillouin scattering at high pressure in a diamond anvil cell, Rev. Sci. lnstrum., 47, , Zha, C.S., T.S. Duffy, H.K. Mao, R.T. Downs, R.J. HemIcy, and D.J. Weidner, Single-crystal elasticity of [3-Mg2SiO4 to the pressure of the 410 km seismic discontinuity in the Earth's mantle, Earth Planet. Sci. Lett., 147, E9-EIS, Zha, C.-S., T.S. Duffy, R.T. Downs, H.-K. Mao, R.J. HemIcy, and D.J. Weidner, Single-crystal elasticity of the ct and 13 Mg2SiOn polymorphs at high pressure, in Properties of Earth and Planetary eclogite-garnetite transformation at high pressure and some Materials at High Pressure and Temperature, Geophys. Monogr. geophysical implications, Earth. Planet. Sci. Lett., 77, , Ser., vol. 101, edited by M.H. Manghnani and T. Yagi, pp. 9-16, AGU, Washington, D.C., Jackson, J. M., S.V. Sinogeikin, J.D. Bass, and T. Katsura, Elasticity of mantle silicates: Pyroxene and 13-(Mg, Fe)2SiO4, Eos Trans., 78 J. D. Bass and S. V. Sinogeikin, Department of Geology, (46), Fall Meet. Suppl., F803, University of Illinois, 245 NHB, 1301 W. Green St., Urbana, IL Kuskov, O., Equations of state and high-pressure phase relationships ( bass hercules.geology.uiuc.edu; sinogeik uiuc.edu.) tbr {x- and ¾-Fe2SiO4 and FeSiO3, Geochem. Int., 22, , Li, B., G.D. Gwanmesia, and R.C. Liehermann, Sound velocities of oilvine and beta polymorphs of Mg2SiO4 at Earth's transition zone pressures, Geophys. Res. Lett., 23, , T. Katsura, Institute for Study of Earth Interior, Okayama University, Misasa, Tottori 68201, Japan. (Received January 20, 1998; revised May 14, 1998; accepted May 21, 1998.)