Search for New Thermoelectric Materials through Exploratory Solid State Chemistry. TheQuaternary Phases A1+xM3-2xBi7+xSe14, A1-xM3-xBi11+xSe20, A1-xM4-xBi11+xSe21 andA1-xM5-xB i 1 1 + xS e 2 2 (A = K, Rb, Cs, M = Sn, Pb) and the Homologous SeriesAm[M6Se8]m[M5+nSe9+n]:

Antje Mrotzek1, Tim Hogan2 and Mercouri G. Kanatzidis1,*1 Department of Chemistry and Center for Fundamental Materials Research, Michigan State

University, East Lansing, MI 48824. 2 Department of Electrical Engineering, Michigan State

University, East Lansing, MI 48824.

Abstract

The compound types A1+xM3-2xBi7+xSe14, A1-xM3-xBi11+xSe20, A1-xM4-xBi11+xSe21 and A1-xM5-

xBi11+xSe22 (A = K, Rb, Cs; M = Sn, Pb) form from reactions involving A2Se, Bi2Se3, M and Se.The single crystal structures reveal that they are all structurally related so that they all belong tothe homologous series Am[M6Se8]m[M5+nSe9+n] (M = di- and trivalent metal), whosecharacteristics are three-dimensional anionic frameworks with tunnels filled with alkali ions.The building units that make up these structures are derived from different sections of the NaCllattice. In these structures, the Bi and Sn (Pb) atoms are extensively disordered over the metalsites of the chalcogenide network, giving rise to very low thermal conductivity. These phasesare all narrow gap semiconductors with 0.25 < Eg< 0.60 eV and many possess physico-chemicaland charge transport properties suitable for thermoelectric investigations.

IntroductionOur approach to discovering new thermoelectric materials centers on complex quaternary

and ternary bismuth chalcogenides with anisotropic frameworks.[1] The examples of CsBi4Te6[2],[3]

and β-K2Bi8Se13[3],[4] show that such exploratory investigations can lead to promising new

thermoelectric materials. Naturally, we enlarged our investigations to quaternary systems of thetype A/M/Bi/Se (A = K, Rb, Cs, M= Pb, Sn) and identified new materials such as

A1+xM3-2xBi7+xSe14,[5] K1+xSn4-2xBi7+xSe15,

[6] K 1-xSn4-xBi11+xSe21,[7] K1-xSn5-xBi11+xSe22,

[8], which havestructures closely related to each other and to those of β-K2Bi8Se13 and K2.5Bi8.5Se14.

[9] It turns out

they are all members of a newly identified grand homologous series Am[M6Se8]m[M5+nSe9+n][7] (M

= di- and valent metal) in both compositional and structural sense. Their structures arecomposed of [M5+nSe9+n] (NaCl111-type) and [M6Se8]m (NaCl100-type) units of variable dimensionsdefined by n and m, which link to produce anionic frameworks with alkali metal (Am) filled

Mat. Res. Soc. Symp. Proc. Vol. 691 © 2002 Materials Research Society

G5.1.1

tunnels. The series has predictive properties so that it is possible to target and prepare newmembers. Here we discuss several materials for consideration in thermoelectric investigationswith various compositions of the type A1+xM3-2xBi7+xSe14, A1-xM3-xBi11+xSe20, A1-xM4-xBi11+xSe21 andA1-xM5-xBi11+xSe22 (A = K, Rb, Cs; M = Sn, Pb).

Results and Discussion

The family of materials discussed here were prepared by fusing at 800˚C appropriatemixtures of A2Se (K, Rb, Cs, M (Pb, Sn), Se and Bi2Se3 loaded in an evacuated carbon-coatedsilica tube. Generally, silver shiny, polycrystalline ingots composed of oriented long needles canbe obtained. In this manner we synthesized many members of the families A1+xM3-2xBi7+xSe14, A1-

xM3-xBi11+xSe20, A1-xM4-xBi11+xSe21 and A1-xM5-xBi11+xSe22. A quantitative microprobe analysis usingenergy dispersive X-ray fluorescence spectroscopy was performed to obtain averagecompositions. Small variations of the ratio of the starting materials result in new compoundswith gradually evolving structural features. The large collection of compounds discovered

suggests strongly that the A2Q/MQ/Bi2Q3 system is probably "infinitely adaptive".[10] That is anew structure type forms each time the composition changes rather than mixtures of two or morephases or solid solutions.

Carving-up the NaCl Lattice. All the phases discussed here have one thing in common.All are made of structural fragments that can be thought of as having been excised out of a NaCltype lattice. These fragments represent various cuts of this lattice in different orientation and ofdifferent dimensions. Usually they vary in dimension along two directions of the NaCl latticewhile the third dimension is infinite. This “carving” of the lattice results in either infinite slabsof various types or infinite rods. Figure 1 depicts the different ways the NaCl lattice can besectioned to produce the building fragments observed in the compounds, discussed here. If the

cut is made perpendicular to a certain direction (e.g. [100], [111]) the fragment may be calledNaCl100–type or NaCl111–type respectively. Sometimes the NaCl111- type fragment is referred toas Bi2Te3–type since the Bi2Te3 structure also derives through excision of infinite two-dimensional slabs cut perpendicular to the [111] direction of the NaCl lattice.

Structure Description. The quaternary phases A1+xM3-2xBi7+xSe14, A1-xM3-xBi11+xSe20, A1-

xM4-xBi11+xSe21 and A1-xM5-xBi11+xSe22 all belong to the homologous series Am[M6Se8]m[M5+nSe9+n].A1-xM3-xBi11+xSe20 and A1+xM3-2xBi7+xSe14 have n = 3 and m = 1 and 2, respectively, whereasA1-xM4-xBi11+xSe21 and A1-xM5-xBi11+xSe22 have n = 4, m = 1 and n = 5, m = 1 respectively. Figures2 and 3 show projections of their structures displaying distinct building units of the NaCl111-typeand NaCl100-type. These units are linked side by side so as to form three-dimensional anionic

frameworks with either fully or partially alkali ion filled tunnels.

G5.1.2

[100]

NaCl100-type, m=2 NaCl100-type, m=1

NaCl111-type

Figure 1. The NaCl lattice viewed down the [110] direction. The boxed areas represent cuts that result in

the building fragments observed in the homologous series. In principle, there are numerous ways to carve

this archetypal lattice to produce an great abundance of potential building blocks and consequently new

structural arrangements, not only for the Am[M6Se8]m[M5+nSe9+n] series but also for other homologies..

M6Se8-unitNaCl100-type

M8Se12-unit

NaCl111-type

m = 2 m = 1

Figure 2. Comparison of the structures of A1+xM3-2xBi7+xSe14 and A1-xM3-xBi11+xSe20. The secondary

NaCl111-type- and NaCl100-type building units are highlighted in both structures. Small white spheres: Se,

large light-gray spheres with no surrounding bonds: A, dark-gray spheres: Bi, dark-gray spheres: M.

G5.1.3

a

c

NaCl-block

Bi2Te3-block

a

Figure 3. Comparison of the structures of A1-xM4-xBi11+xSe21 and A1-xM5-xBi11+xSe22. Small white spheres:

Se, large light-gray spheres with no surrounding bonds: A, dark-gray spheres: Bi, black spheres: M. The

archetypal NaCl111-type (or Bi2Te3 type) and NaCl100-type building units are highlighted in both structures.

The structures depicted in Figure 2 possess the same [M8Se12] NaCl111-type unit (seeboxed area) which is four Bi-octahedra wide and two Bi-octahedra high and therefore resemblesa cut out of a Bi2Te3-type layer. Condensation of these units via one octahedron edge results in astep-shaped layer of the formula [M5+nSe9+n] (n = 3). The adoption of these stable structure typesis accomplished through extensive mixed occupancy disorder between A and Bi, M and Bi, andeven between all A, M, Bi atoms on selected key crystallographic sites. Those sites that areprimarily occupied with Bi atoms have distorted octahedral geometry, the other sites involving Aatoms can be of higher coordination. The distorted Bi-octahedra have interatomic distancesranging from 2.74 – 3.20 Å for K1.40Sn2.20Bi7.40Se14, 2.74 – 3.11 Å for K0.70Sn2.70Bi11.30Se20, 2.76 –3.09 Å for Rb0.36Sn2.36Bi11.64Se20 and 2.75 – 3.08 Å for Cs0.46Sn2.46Bi11.54Se20. The distances reflect

the different extend of Sn/Bi disorder in the step-shaped layers of these compounds. The twostructure types in Figure 2 differ in the size of the [M6Se8]-units that link the stepped [M5+nSe9+n]-layers (n = 3) via MSe interactions (M = Bi, Sn(Pb)) to a three-dimensional framework. In A1-

xM3-xBi11+xSe20 these blocks are three M octahedra wide and one octahedron high and thereforevery similar to those found in A1-xM4-xBi11+xSe21

[7] and A1-xM5-xBi11+xSe22[8] (see Figure 3). All

metal positions in this block show mixed Bi/M occupancy. In contrast, in A1+xM3-2xBi7+xSe14 thecorresponding [M6Se8]-units are two M octahedra high in the direction perpendicular to theNaCl111-type-type layers. Here the outer surface of the NaCl100-type block shows considerablemixed K/Sn occupancy. For example in the specific case of K0.70Sn2.70Bi11.30Se20 we found 23%in the K1 site and 17% in the K2 sites that correspond to the Sn1 and Bi6 site respectively. Thistype of mixed occupancy significantly impacts the electrical properties of these materials.

The phases of the type A1-xM4-xBi11+xSe21 and A1-xM5-xBi11+xSe22 also bear a close structuralrelationship both among themselves and to those in Figure 2. The structures depicted in Figure

G5.1.4

3 possess the same [M8Se12] NaCl111-type unit (n=1). The structure of A1-xM4-xBi11+xSe21 iscomposed of two distinct highlighted building units. The composition of the two phases differsonly by one “MSe” per formula. In A1-xM4-xBi11+xSe21 the Bi2Te3 (i.e. NaCl111-type) building unitsin the structure are five Bi octahedra wide along the c-direction and they are joined in an offsetfashion to form a stepped shaped layer. The connection point of these fragments is found at a

single octahedral Bi site. In contrast, the same buildings units in A1-xM5-xBi11+xSe22, which arealso offset, are joined via an octahedron edge, Figure 3.

In all four phases above the alkali metals are found in distorted tri-capped trigonalprismatic sites in the tunnels created by parallel arrangement of the two kinds of NaCl-typebuilding units to a three-dimensional framework. Invariably in these structures the observedlarge atomic thermal displacement parameters (TDP)s are very large often 3-4 times the averagevalue of the Bi/Se framework. This could be due to either “rattling” of the alkali atoms on theircrystallographic sites or, more likely, positional disorder along the tunnel since the position isonly about 1/3 occupied.

Classification of the New Quaternary Selenides. Let us now examine how all thephases discussed here are members of the same homologous series Am[M6Se8]m[M5+nSe9+n] withspecific n and m values. The close relationship of β - K2Bi8Se13, A1+xM3-2xBi7+xSe14 to

A1-xM3-xBi11+xSe20, A1-xM4-xBi11+xSe21 and A1-xM5-xBi11+xSe22 becomes more apparent when we

inspect the scheme depicted in Figure 4. The scheme places these and other phases in the samestructural context and offers an easy way to predict new phases by combining known [M6Se8]m-

blocks and [M5+nSe9+n]-layers. Besides varying n to obtain new compounds it would be worthexploring if members with m > 2 are also stable.

For the three phases of A1-xM3-xBi11+xSe20, A1-xM4-xBi11+xSe21and A1-xM5-xBi11+xSe22 theNaCl100-type block remains the same and the latter two (i.e. n = 4 and n = 5) can be easilyderived from A1-xM3-xBi11+xSe20 by successively adding MSe equivalents to the [M8Se12]-layers.The places in the scheme displaying question-marks are predicted phases which should exist andare worth targeting for synthesis. The Am[M6Se8]m[M5+nSe9+n] homology may capture only afraction of the total number of phases possible, as we have observed compositions which notbelong to the series. For example Cs1-xSn1-xBi9+xSe15 and Cs1.5-3xBi9.5+xSe15 crystallize in astructure type that does not belong to but is closely related to the members of the seriesAm[M6Se8]m[M5+nSe9+n]. These phases reveal a third dimension of structural evolution according

to the formula Am[M1+lSe2+l]2m[M1+2l+nSe3+3l+n].11

G5.1.5

+ 1 MSe

+ 1 MSe

+ 2 MSe

?

(e.g. K2Sn7Bi14Se29)M9Se13-layer (n = 4)

M8Se12-layer (n = 3)

M10Se14-layer (n = 5)

?

(e.g. KSn2Bi11Se19)

KSn3Bi11Se20 (m = 1; n = 3)

M6Se10-layer (n = 1)

m = 2 m = 1

β−K2Bi8Se13

(m = 2; n = 1)

KSn3Bi7Se14

(m = 2; n = 3)

KSn4Bi11Se21

(m = 1; n = 4)

KSn5Bi11Se22

(m = 1; n = 5)KSn4Bi7Se15

(m = 2; n = 5)

Figure 4. The homologous series Am[M6Se8]m[M5+nSe9+n]. A member generating scheme illustrating

successive additions of MSe units to a M5Se9 layer. Small white spheres: Se, large light-gray spheres: K,

middle-gray spheres: M.

G5.1.6

Charge Transport Properties. By far the majority of the phases depicted in Figure 4await electrical and thermoelectric characterization and further investigation. Only a smallnumber of members has been examined so far and these only in a preliminary sense. Table 1summarizes room temperature values for thermopower, electrical conductivity and band gaps forseveral phases. All materials are n-type semiconductors.

The electrical conductivity was measured on polycrystalline ingots of several A1-xM3-

xBi11+xSe20 compounds and not surprisingly, was found to be strongly influenced by thepreparation conditions. If the molten mixture of the starting materials was rapidly cooled, highelectrical conductivity of ~1700 S/cm and 1350 S/cm was observed respectively for K1-xSn3-

xBi11+xSe20 and Rb1-xSn3-xBi11+xSe20. For a slowly cooled sample of K1-xSn3-xBi11+xSe20 a lowerroom temperature conductivity at 520 S/cm. For slowly cooled samples, in going from K to a Csanalog the electrical conductivity seems to decrease (370 S/cm for Cs1-xSn3-xBi11+xSe20).Generally, if all other conditions are the same, the Pb analogs K1-xPb3-xBi11+xSe20 (700 S/cm) andCs1-xPb3-xBi11+xSe20 (690 S/cm) possess higher electrical conductivity than the corresponding Sncompounds. A similar Sn vs Pb trend tends to persist A1+xM3-2xBi7+xSe14, see Table 1. The highconductivity of quenched samples is attributed to the generation of a large number of carrier

producing defects in the structure, whose exact nature is not known but can be speculated to beM/Se anti-site defects or Se vacancies.

As prepared these compounds possess moderate negative Seebeck coefficients with anearly linear dependence, see Table 1 and Figure 5. The negative values indicate n-typebehavior with electrons as the dominant charge carriers. With rising temperature from 300 to400 K the absolute value of the negative Seebeck coefficient increases from -45 to -62 µV/K for

K1+xSn3-2xBi7+xSe14 and -32 to –52 µV/K for K1+xPb3-2xBi7+xSe14, respectively. The Cs analogs

have slightly higher thermopower at room temperature with -48 µV/K (Cs1+xSn3-2xBi7+xSe14) and -

56 µV/K (Cs1+xPb3-2xBi7+xSe14). Such moderately low values are reasonable considering the high

electrical conductivity of these samples.Slowly cooled samples of K1-xSn3-xBi11+xSe20 exhibit a linearly rising absolute Seebeck

value from -68 µV/K at 300K to -107 µV/K at 400K. However, we observed much lower

thermopower for the quenched sample. The low values starting from -16 to -28 µV/K

correspond to the higher electrical conductivity observed for this sample, see Table 1. QuenchedRb1-xSn3-xBi11+xSe20 behaves similarly, it is highly doped and exhibits low thermopower of -26µV/K at 300 K that rises to -35 µV/K at 400 K. Annealing these samples for several hours or

days can increase the thermopower significantly to values up to –150 to –200 µV/K range. Thissuggests the quenching process creates a large number of structural defects in these structures.By comparison, the slowly cooled sample of Cs1-xSn3-xBi11+xSe20 possesses higher thermopower(-69 µV/K increasing to -92 µV/K) consistent with its lower electrical conductivity of 370 S/cm.

For K1-xPb4-xBi11+xSe21, Rb1-xSn4-xBi11+xSe21, Rb1-xPb4-xBi11+xSe21 and Cs1-xPb4-xBi11+xSe21 theelectrical conductivity of polycrystalline samples as a function of temperature are displayed in

G5.1.7

Figure 6 and the thermopower in Figure 7. Rb1-xPb4-xBi11+xSe21 showed a high conductivity withvalues starting around 1050 S/cm at 80K and falling to 530 S/cm at 300K. In comparison, the Snanalog is less conductive with an electrical conductivity of 450 S/cm at 80K and 160 S/cm at 400K. The observed temperature dependencies of the conductivity are similar in all analogs,consistent with the behavior of a degenerately doped narrow band gap semiconductor.

Table 1. Charge transport properties and band gaps for some members of theAm[M6Se8]m[M5+nSe9+n] series

Compound σ300K (S/cm) S300K (µV/K) Eg (eV)

β-K2Bi8Se13 250 -200 0.59

K1+xSn3-2xBi7+xSe14 2000 -45 c)K1+xPb3-2xBi7+xSe14 630 -32 0.56Cs1+xSn3-2xBi7+xSe14 200 -48 0.40Cs1+xPb3-2xBi7+xSe14 350 -56 0.62K1+xSn4-2xBi7+xSe15 110 -70 0.39K1-xPb3-xBi11+xSe20

a) 700 -49 0.55Rb1-xSn3-xBi11+xSe20

a) 1350 -25 c)

Rb1-xPb3-xBi11+xSe20a) 1060 -46 0.54

Cs1-xSn3-xBi11+xSe20b) 370 -69 0.40

Cs1-xPb3-xBi11+xSe20b) 690 -91 c)

K1-xSn3-xBi11+xSe20a) 1700 -16 c)

K1-xSn3-xBi11+xSe20b) 520 -68 0.30

Rb1-xSn4-xBi11+xSe21 200 -51 0.50Cs1-xSn4-xBi11+xSe21 170 -69 0.52K1-xPb4-xBi11+xSe21 1200 -63 c)Rb1-xPb4-xBi11+xSe21

a) 530 -45 0.62Rb1-xPb4-xBi11+xSe21

b) 250 -63 0.62Cs1-xPb4-xBi11+xSe21 160 -56 0.62

K1-xSn4-xBi11+xSe21 1800 -52 c)K1-xSn5-xBi11+xSe22 450 -43 c)a) quenched b) slowly cooled c) could not be determined reliably due to the high absorption from mid gap states.

Variable temperature thermopower data from polycrystalline ingots of K1-xPb4-xBi11+xSe21,

Rb1-xSn4-xBi11+xSe 21, Rb1-xPb4-xBi11+xS e 2 1 and Cs1 - xPb4-xBi11+xSe 21 show negative Seebeckcoefficients and nearly linear dependence, see Figure 7. With rising temperature the Seebeckcoefficient varies from - 14 µV/K at 75K to - 72 µV/K at 400K for Rb 1-xSn4-xBi11+xSe21 and Cs1-

xPb4-xBi11+xSe21. We observed smaller thermopower for K1-xPb4-xBi11+xSe21 and Rb1-xPb4-xBi11+xSe21

G5.1.8

starting from - 4 µV/K at 75K to - 35 µV/K at 400K and from - 12 µV/K at 75K to - 45 µV/K at300K, respectively. The change from a lighter alkali metal to a heavier one results in higherthermopower for A1-xPb4-xBi11+xSe21. However, due to different doping levels in the samples frompreparation to preparation a definitive trend could not be found. We expect these compounds tobe sensitive to doping and work along these lines is in progress.

CsSn3Bi11Se20

-200

-150

-100

-50

0

280 300 320 340 360 380 400 420

KSn3Bi11Se20KPb3Bi11Se20RbSn3Bi11Se20 quenched

CsPb3Bi11Se20KSn3Bi7Se14

KPb3Bi7Se14KSn3Bi11Se20 quenched

See

beck

Coe

ffici

ent (

µV/K

)

Temperature (K)

Figure 5. Seebeck coefficient data as a function of temperature for selected members of the

A1+xM3-2xBi7+xSe14 to A1-xM3-xBi11+xSe20 families.

0

200

400

600

800

1000

1200

50 100 150 200 250 300 350 400 450

Ele

ctric

al C

ondu

ctiv

ity /

S/c

m

Temperature / K

Rb1-xSn4-xBi11+xSe21

Cs1-xPb4-xBi11+xSe21

Rb1-xPb4-xBi11+xSe21

K1-xPb4-xBi11+xSe21

Ele

ctric

al C

ondu

ctiv

ity /

S c

m-1

Figure 6. Electrical conductivity data as a function of temperature for selected members of the

family A1-xM4-xBi11+xSe21.

G5.1.9

-80

-70

-60

-50

-40

-30

-20

-10

0

50 100 150 200 250 300 350 400 450

See

beck

coe

ffici

ent (

µ V/K

)

Temperature (K)

Rb1-xSn4-xBi11+xSe21

Cs1-xPb4-xBi11+xSe21

Rb1-xPb4-xBi11+xSe21

K1-xPb4-xBi11+xSe21

Figure 7. Seebeck coefficient data as a function of temperature for selected members of the family

A1-xM4-xBi11+xSe21.

Very little work has been done yet on the K1-xSn5-xBi11+xSe22 samples. The electricalconductivity of an ingot of K0.66Sn4.82Bi11.18Se22 was measured to be 450 S/cm at roomtemperature. The thermopower of compacted samples of K0.66Sn4.82Bi11.18Se22 increases steadilyup from –13 µV/K at 80 K to –98 µV/K at 600 K.

0

0.5

1

1.5

2

0 50 100 150 200 250 300 350 400

The

rmal

Con

d. /

W m

-1K

-1

Temperature / K

K1-xSn4-xBi11+xSe21

0

0.5

1

1.5

2

0 50 100 150 200 250 300 350 400

The

rmal

Con

d. /

W m

-1K

-1

Temperature / K

Cs1-xPb4-xBi11+xSe21

0

0.5

1

1.5

2

0 50 100 150 200 250 300 350 400

The

rmal

Con

d. /

W m

-1K

-1

Temperature / K

Rb1-xPb4-xBi11+xSe21

Figure 8. Variable temperature thermal conductivity data from polycrystalline ingots of Rb1-xPb4-

xBi11+xSe21, K1-xSn4-xBi11+xSe2, and Cs1-xPb4-xBi11+xSe21.

Thermal Conductivity. A common feature of all these phases is that the thermalconductivity is extraordinarily low. The low symmetry, large lattice constants and extensivemass fluctuation disorder are inherent in these phases, and almost guarantee an extremely highthermal resistance. The temperature dependence of the thermal conductivity was measured onpolycrystalline ingots of K1-xPb4-xBi11+xSe21, Rb1-xSn4-xBi11+xSe21, Rb1-xPb4-xBi11+xSe21 and Cs1-xPb4-

G5.1.10

xBi11+xSe21. Besides K1-xPb4-xBi11+xSe21 which exhibited a slightly higher thermal conductivity (1.6W/m·K at 300K), all samples possess very low thermal conductivity around 1 Wm-1·K-1 at roomtemperature, Figure 8. The observed values for A1-xM4-xBi11+xSe21 are among the lowest reportedfor materials with potential thermoelectric interest. Optimized Bi2Te3 alloys have a thermalconductivity of ~1.5 W/m·K, which is about 50% higher than those of A1-xM4-xBi11+xSe21 (A = K,

Rb, Cs; M = Sn, Pb). Because of their very low thermal conductivity it seems worthwhile tostudy systematically A/A’ or Pb/Sn solid solutions in order to enhance the thermoelectricproperties. The thermal conductivity of K0.66Sn4.82Bi11.18Se22 is also low and it increases withrising temperature from 0.8 W/m·K (80 K) to 1.4 W/m·K (270 K).

0.2 0.3 0.4 0.5 0.6 0.7 0.8

Abs

orpt

ion,

α/S

Energy, eV

Eg ~ 0.30 eV

Figure 9. Infrared absorption spectrum of K1-

xSn3-xBi11+xSe20 obtained at room temperature.The energy band gap is indicated in thespectrum.

Energy gaps. The infrared absorption spectraof A1+xM3-2xBi7+xSe14 and A1-xM3-xBi11+xSe20 (A= K, Rb, Cs; M = Sn, Pb) were recorded atroom temperature in the range of 0.1 to 0.7 eV.The optical band gaps of K1+xSn3-2xBi7+xSe14,Rb1-xSn3-xBi11+xS e 2 0 and Cs1-xPb3-xBi11+xSe20

could not be determined reliably. However,for K1+xPb3-2xBi7+xSe14, Cs1+xM3-2xBi7+xSe14, K1-

xM3-xBi11+xSe20, Rb1-xPb3-xBi11+xSe20 and Cs1-xSn3-

xBi11+xSe20 we were able to observe opticalband gaps between ~ 0.3 eV (see Figure 9) and~0.6 eV. The narrow band gaps of thesequaternary selenides are consistent with theobserved charge transport behavior describedabove.

Concluding RemarksThe exploration of the system A2Q/MQ/Bi2Q3 (A = K, Rb, Cs; M = Sn, Pb) leads to the

suggestion that it may be "infinitely adaptive". The existence of the grand homologous seriesAm[M6Se8]m[M5+nSe9+n] that defines a large family of materials seems clear at this stage. Allmembers of the homologous series are constructed from the basic NaCl111- and the NaCl100-typemodules whose size vary according to n and m . The phases are n-type narrow band gapsemiconductors with extremely low thermal conductivity. This broad class of materials promisesto be a great new source of potentially useful TE materials. The pursuit of additional members ofthe series is currently under way.

Acknowledgment. Financial support from the Office of Naval Research and DARPA isgratefully acknowledged.

G5.1.11

References

[1] (a) Kanatzidis, M. G.; DiSalvo, F. J. ONR Quarterly Review 1996, XXVII, 14-22. (b)Chung, D.-Y.; Iordanidis, L.; Choi, K.-S.; Kanatzidis, M. G. Bull. Kor. Chem. Soc 1998,19, 1283-1293. (c) Kanatzidis, M. G. Semicond. and Semimetals, Academic Press 2001,69, 51-100.

[2] Chung, D.-Y.; Hogan, T.; Brazis, P. W.; Rocci-Lane, M.; Kannewurf, C. R.; Bastea, M.;Uher, C.; Kanatzidis, M. G. Science 2000, 287, 1024-1027.

[3] (a) Kanatzidis, M. G.; Chung, D.-Y.; Iordanidis, L.; Choi, K.-S.; Brazis, P.; Rocci, M.;Hogan, T.; Kannewurf, C. Mat. Res. Soc. Symp. Proc. 1998, 545, 233-246. (b) Brazis, P.W.; Rocci-Lane, M. A.; Ireland, J. R.; Chung, D.-Y.; Kanatzidis, M. G.; Kannewurf, C.R. Proc. Of the XVIIIth Int. Conf. On Thermoelectrics (ITC ’99), Baltimore, USA 1999,619-622.

[4] (a) Kanatzidis, M. G.; McCarthy, T. J.; Tanzer, T. A.; Chen, L.-H.; Iordanidis, L.;Hogan, T.; Kannewurf, C. R.; Uher, C.; Chen, B. Chem. Mater. 1996, 8, 1465-1474. (b)Chen, B.; Uher, C.; Iordanidis, L.; Kanatzidis, M. G. Chem. Mater. 1997, 9, 1655-1658.

[5] Choi, K.-S.; Chung, D.-Y.; Mrotzek, A.; Brazis, P.; Kannewurf, C. R.; Uher, C.; Chen,W.; Hogan, T.; Kanatzidis, M. G. Chem. Mater. 2001, 13, 756-764.

[6] Choi, K.-S.; Chung, D.-Y.; Mrotzek, A.; Brazis, P. W.; Kannewurf, C. R.; Uher C.; ChenW.; Hogan, T.; Kanatzidis, M. G., Chem. Mater. 2001, 13, 756-764.

[7] Mrotzek, A.; Chung, D-Y.; Hogan, T.; Kanatzidis, M. G. J. Mater. Chem. 2000, 10,1667-1672.

[8] Mrotzek, A.; Chung, D-Y.; Ghelani, N.; Hogan, T.; Kanatzidis, M. G. Chem. Eur. J.

2001, 7, 1915-1926.[9] Chung, D.-Y.; Choi, K.-S.; Iordanidis, L.; Schindler, J. L.; Brazis, P. W.; Kannewurf, C.

R.; Chen, B.; Hu, S.; Uher, C.; Kanatzidis, M. G. Chem. Mater., 1997, 9, 3060-3071.[10] (a) Anderson, J. S., J. Chem. Soc., Dalton Trans. 1973, 10, 1107-1115. (b) Swinnea, J.

S.; Steinfink, H., J. Solid State Chem. 1982, 41, 114-123. (c) Mercurio, D.; Parry, B. H.;Frit, B.; Harburn, G.; Williams, R. P.; Tilley, R. J. D., J. Solid State Chem. 1991, 92, 449-459.

[11] Mrotzek, A.; Iordanidis L.; Kanatzidis, M. G., Chem. Commun. 2001, 1648-1649.

G5.1.12