Science 13Mar2020

SCIENCE sciencemag.org 13 MARCH 2020 • VOL 367 ISSUE 6483 1197

GRAPHIC: KELLIE HOLOSKI/

SCIENCE

This bandgap is given by a rule Eg = 2eSmaxT,

where e is unit charge, Smax is the maximum

Seebeck coefficient, and T is the temperature

that corresponds to Smax. The Seebeck coeffi-

cient is also called the thermopower and is a

measurement of the voltage produced within

a temperature gradient (S = DV/DT, where V

is voltage). The bandgap rule means that the

most well-known thermoelectric materials

are narrow-bandgap semiconductors, such

as (Bi,Sb) 2 Te 3 (Eg ~ 0.13 eV) ( 3 ), PbTe (Eg ~

0.28 eV) ( 4 ), and GeTe-AgSbTe 2 (Eg ~ 0.39

eV) ( 5 ). The maximum ZT (ZTmax) values for

these thermoelectric materials shift to higher

temperatures as the bandgap increases (see

the figure). To fully realize their potential,

thermoelectric materials must work over

the entire, several-hundred-kelvin operating

range. One method for doing this is with a

segmented leg (see the figure), but interfacial

resistance and mismatched compatibility fac-

tors deteriorate the long-term performance

under high temperature ( 6 ).

Wide-bandgap semiconductors could

solve this temperature range issue but often

have poor electrical properties, but the wide-

bandgap SnSe (~0.86 eV) has proved to be

an excellent thermoelectric material. Its ZT

curve for SnSe covers several narrow-band-

gap thermoelectrics ( 7 – 9 ). SnSe possesses

attractive ZT values at low temperatures,

which continuously increase without satura-

tion up to 800 K. Several special features of

SnSe provide some general selection rules for

new thermoelectric materials that may work

over a wide temperature range. First, the

wide bandgap avoids the intrinsic excitation

and the ZT values are not saturated at high

temperatures. Second, layered structures can

have high in-plane transport properties that

circumvent the normally low carrier density

that plagues wide-bandgap semiconductors.

Wide-bandgap semiconductors were ne-

glected as promising thermoelectrics because

of their intrinsically low carrier density. This

deviates from the optimal carrier density (n)

owing to the ZT parameter interrelations. To

achieve high electrical transport properties

in materials with low carrier density, high

carrier mobility (m) can be found along an

in-plane direction in layered structures, and

thus layered materials can reach high electri-

cal conductivity s = nem. Furthermore, the

low carrier density allows for a high Seebeck

coefficient, and, consequently, an ultrahigh

power factor (PF = S^2 s) ( 9 , 10 ). Third, a low-

symmetry structure is connected to low lat-

tice thermal conductivity (klat), which is a

lower electronic themal conductivity (kele) ow-

ing to low carrier density, and contributes to

the total thermal conductivity (k = klat + kele).

Asymmetric crystal structures have strong

anharmonic lattice vibrations useful for low-

ering thermal conductivity, and their more

complex electronic band structure is also

attractive for thermoelectric materials. The

selection rules will not work for all materials

because of the complex interplay between the

ZT parameters but should provide at least a

rough guide for candidate materials. The se-

lection rules for identifying potentially highly

effective thermoelectrics are appropriate for

both n-type and p-type materials because of

their similar transport principles.

Within these selection rules, some prom-

ising thermoelectrics can be identified, such

as BiCuSeO ( 11 ), BiSbSe 3 ( 12 ), K 2 Bi 8 Se 13 ( 13 ),

and Sb 2 Si 2 Te 6 ( 14 ). The anisotropic transport

properties should lead to improved perfor-

mance in crystalline forms of these materials

where we expect, as for SnSe and SnS crys-

tals, higher carrier mobility ( 9 , 10 ). Moreover,

much more highly effective thermoelectric

performance from these anisotropic thermo-

electric materials could be expected through

integrating present selection rules with the

approach to reveal the intrinsically low ther-

mal conductivity ( 15 ). Finally, it must also be

mentioned that not every material with high-

range ZT values, ZTave, is going to immediately

make for a device-ready material. High ZTave

thermoelectric materials may be challenging

to ultimately turn into commercial devices,

especially at higher temperatures. Interfacial

resistivity and diffusion between the high-

performance thermoelectrics and contact

electrode that can degrade thermoelectric

performance over time are exacerbated at

high temperatures. Some of these issues may

be solved by device engineering, but the pres-

ent selection rules also provide a rough guide

for finding different types of thermoelectric

materials. j

REFERENCES AND NOTES

1. C. B. Vining, Nat. Mater. 8 , 83 (2009).


2. X. Shi, L. Chen, Nat. Mater. 15 , 691 (2016).


3. B. Poudel et al., Science 320 , 634 (2008).


4. P. F. Poudeu et al., Angew. Chem. Int. Ed. 45 , 3835 (2006).


5. S. K. Placheova, Phys. Status Solidi A Appl. Res. 83 , 349
   (1984).


6. W. Liu, S. Bai, J. Materiomics 5 , 321 (2019).


7. L.-D. Zhao et al., Nature 508 , 373 (2014).


8. C. Chang et al., Science 360 , 778 (2018).


9. L.-D. Zhao et al., Science 351 , 141 (2016).
   1 0. W. H e et al., Science 365 , 1418 (2019).


10. L.-D. Zhao et al., Appl. Phys. Lett. 97 , 092118 (2010).


11. S. Wang et al., Energy Environ. Sci. 9 , 3436 (2016).


12. D.-Y. Chung et al., Chem. Mater. 9 , 3060 (1997).


13. Y. L u o et al., Joule 4 , 159 (2020).
    1 5. Y. L e e et al., Joule 3 , 719 (2019).

ACKNOWLEDGMENTS

We acknowledge support from the National Key Research

and Development Program of China (2018YFA0702100;

2018YFB0703600), National Natural Science Foundation

of China (51772012; 51671015), Beijing Natural Science

Foundation (JQ18004), Shenzhen Peacock Plan team

(KQTD2016022619565991), National Postdoctoral Program

for Innovative Talents (BX20190028), 111 Project (B17002), and

Postdoctoral Science Foundation of China (2019M660399).

L.-D. Z. has support from the National Science Foundation for

Distinguished Young Scholars (51925101).

10.1126/science.aaz9426

1

2

3

4

Hot side

Cool side

Current fow

Segmented leg

800 K

300 K

Single leg

Layered and low-symmetry crystal

Sn

Se

n p

_ +

a

b c

+

Thermoelectric devices are

composed of p-type and

n-type semiconducting legs.

The legs combine to convert

heat into current.

To work across a wide

temperature range, a leg

can be composed of several

materials with narrow

bandgaps or one material

with a wide bandgap.

Wide-bandgap single legs [tin selenide (SnSe),

for example] should be crystals that are layered

and have low symmetry, which are more likely to

have a promising thermoelectric performance.

Comparing the fgure of merit

at diferent temperatures

The wide-bandgap material 4 (0.86

eV) has a high ZT across the entire

temperature range of the

segmented leg, whereas each

narrow-bandgap material 1 to 3

(0.13, 0.28, and 0.39 eV) works well

at diferent temperature ranges.

2

1

0

300 400 500 600 700 800

Temperature (K)

ZT Material 1

Material 2

Material 3

Material 4

Searching for thermoelectrics

Finding thermoelectric materials that are efficient across a wide range of temperatures is a challenge. A high

figure of merit (ZT) denotes a promising thermoelectric, but maybe only for a small range of temperatures.

Published by AAAS