Spin Transitions in Lower Mantle Minerals?

Concentrate on ferropericlase as more likely to have a big effect

Electron configurations

• K shell 1s• L shell 2s 2p• M shell 3s 3p 3d

• s suborbitals take up to 2 electrons• p suborbitals take up to 6 electrons• d suborbitals take up to 10 electrons

• Fe2+ has 24 electrons, 2 in K, 8 in L and 14 in M with 6 in 3d

Structure of ferropericlase -- green are oxygen and blue are magnesium or iron

(note eg orbitals point towards nearest neighbor oxygens and t2g point between)

Fp

Pv

Fe(2+): 3d shell has 6 (out of 8) electrons -- prefer to be unpaired (high spin)

Note aluminous Pv can have Fe(3+) as well as Fe(2+)Volume contraction is not as great in Pv because outlying orbitals still populated

Xray emission spectroscopy

• K-shell electron absorbs an Xray photon and is ejected• A 3p electron collapses into the K-shell• The resulting 3p hole interacts with the partially filled 3d shell (the interaction is a

function of the spin state of the 3d shell)• Main peak is associated with K-beta emission -- satellite peak associated with

3d shell -- intensity of peak depends on spin polarization of 3d shell• Satellite peak disappears when all 3d is in low spin state

Fp (0.17) -- this is expected composition if Fe partitions preferentially into Fp

Lin et al, 2007 (Science -- in press)

Lin et al, 2007 (Science -- in press)

Summary of experimental results

• All experiments at room temperature (except Lin et al 2007)• Ferropericlase: transition range at 40--60GPa? Large weakening of elastic

moduli during transition. Experiments are for Fe rich specimens (Fe#=17--25) or for Fe#=6 corrected to larger value. LS phase seems more opaque (lower thermal conductivity). There is a clear increase in density between high and low spin states at room temperature. Enough data to estimate an EOS for high spin and low spin states

• Perovskite: some find two sharp transitions, some a continuous transition over a wide pressure band (likely in aluminous samples). Some find that LS state is more transparent (higher thermal conductivity)? Effect on elasticity may be mininal (but still important?)

• Theoretical calculations and experiment at high T suggest broad pressure transition range

Fp (0.17) (Lin et al, 2005)

(Fei et al 2007)Fp (0.20)

Some thermodynamics!

Crowhurst et al 2008

Xfe=0.06

Xfe=0.17

Xfe=0.17, T=300K, no modulus weakening

Red is HS and blue is LS; black line prediction of model

Xfe=0.17, T=300K, modulus weakening

Crowhurst et al 2008

Xfe=0.06

Xfe=0.17

Lin et al, 2007 (Science)

Can use observed width of transition to fix dE in thermo model at high T

Xfe=0.17, T=1800K, modulus weakening

Red is HS, blue is LS and black is prediction of model

How does this affect fit to lower mantle properties?

Xfe=0.17Modulus weakening

Red=densityBlue=VcGreen=Vs

“pyrolite” + spin transition in ferropericlase

Xfe=0.17No weakening

Red=densityBlue=VcGreen=Vs

“pyrolite” + spin transition in ferropericlase

Conclusions

• 1D seismic models are extremely well-known in most of the lower mantle and, along with advances in mineral physics, are useful for constraining the bulk composition of the Earth

• A limited range of compositions fit the seismic models (though the precision of the mineral physics estimates of shear velocity is a limiting factor)

• Recent results on the elastic properties of the spin crossover in ferropericlase result in bulk sound speed velocities and velocity gradients in the lower mantle which are apparently incompatible with the 1D seismic models.

• Perhaps anomalous elastic effects are diminished at high T?• Or perhaps Fe is not so strongly partitioned into ferropericlase (the

partitioning may be controlled by the presence of small amounts of aluminum, etc)

• Partitioning may also be a function of spin state -- just to make life more interesting

• Don’t need to rewrite all the text books just yet!

Future work

• Need high T experimental data on elasticity • Need better data for perovskite since this is the bulk

of the lower mantle • Need to look at seismic constraints on velocity

gradients since these may be most diagnostic

Fp (0.25)

Pv (0.1) -- broad transition?

Aluminous sample -- broad transition

(Crowhurst, Brown, Goncharov and Jabobesen, submitted to Science), uses ISS

(Crowhurst, Brown, Goncharov and Jabobesen, submitted to Science),