Science - USA (2020-07-10)

100-hour period, in which the operation volt-
agewas improved from 0.9 to ~1.0 V by in-
troducing an anode buffer layer (ABL) (fig.
S12). The ABL can effectively enhance device
performance and durability. This durability
test can be further improved by exploring
optimized electrode materials and device
technology in future engineering efforts.
Theoretical calculations further reveal the
proton conduction mechanism, as illustrated
in Fig. 4 (see Fig. 4A for the distribution of
charge density differences of the NCO/CeO 2
interface). At the NCO/CeO 2 interface, the un-
balanced charge distribution between the
Ce–O and Co–O layers induces a LEF within
the interface region. Oxygen vacancies of CeO 2
and sodium vacancies of NCO can also cause
an in-plane LEF, for which interfacial poten-
tial is verified by KPFM (Fig. 2F) and the LEF

by the current-voltage (I-V)curveoftheNCO/

CeO 2 thin-film device (Fig. 2G). Proton trans-

port states and paths are illustrated in Fig. 4,

B and C. Proton intercalation leads to a pro-

nounced spectral weight shift of the Co 3d and

Ce 4f bands toward higher energy. Correspond-

ingly, distributed orbitals cover a wide range

of Fermi energies (EF) from−3to+1eV,in-

dicating the proton transport path with a

lower barrier and resistance. Nudged elastic

band calculations were carried out to under-

stand proton migration pathways. The energy

diagram of H diffusion at the interface (Fig.

4D) reveals the lowest energy barrier between

protons and NCO. This is the optimal migra-

tion pathway for a proton moving from CeO 2

to the NCO surface. During the transport pro-

cess, the proton driven by the LEF could in-

corporate with a lattice oxygen ion and move

through the oxygen vacancy of CeO 2 (i: initial

state) to form the H–OCe bond (bond length of

0.98 Å). Meanwhile, the LEF is confined in the

heterostructure interface, which drives pro-

tons from the positively charged CeO 2 to the

negatively charged NCO surface. (ii: transition

state) (Fig. 4E and table S3). When the proton

is close to NCO, Na experiences a large dis-

tortion that exerts a strong repulsive force

on the proton. This causes deformation of

the H···OCo bond (bond length 1.04 Å) and the

H···OCe bond (1.70 Å). In addition, the energy

barrier is too high for protons to be transported

in the NCO and CeO 2 structure, so they are

confined in the NCO/CeO 2 interface. Finally,

protons spontaneously flow along the proton

channel that results from the LEF-induced

metallic state of NCO surface (iii: final state),

of which the active energy (0.15 eV) is much

lowerthanthoseofthebulkCeO 2 (3.17 eV)

and NCO (3.89 eV). Such enhanced proton

conduction is attributed to a synergic effect

between the LEF and the metallic state of the

NCO surface, which substantially accelerates

proton transport.

Our approach may be used to improve proton

transport by acceleration through a field-induced

metallic state in the NCO/CeO 2 semiconduc-

tor heterostructure. The desired high proton

conductivity, 0.1 to 0.3 S cm−^1 (at 370 to 520°C),

has been achieved with this system, and the

NCO/CeO 2 fuel cell has exhibited a power den-

sity of 1000 mW cm−^2 at 520°C. Our work offers

a general methodology to develop functional

semiconductor heterostructures with distinc-

tive interfacial structure and properties. More

broadly, this work lays the theoretical and ex-

perimental foundations for designing and ex-

ploiting novel materials in energy applications.

REFERENCES AND NOTES

1. C. Duanet al.,Nature 557 , 217–222 (2018).


2. S. Choiet al.,Nat. Energy 3 , 202–210 (2018).


3. C. Duanet al.,Nat. Energy 4 , 230–240 (2019).


4. J. B. Goodenough,Nature 404 , 821–823 (2000).


5. C. Duanet al.,Science 349 , 1321–1326 (2015).


6. R.Lan,S.W.Tao,Adv. Energy Mater. 4 , 1301683– 1301689
   (2014).


7. I. H. Kimet al.,J. Mater. Chem. A 7 , 21321–21328 (2019).


8. Y. Zhouet al.,Nature 534 , 231–234 (2016).


9. Y. Zhenget al.,Adv. Mater. 29 , 1700396 (2017).


10. T. Xia, W. Zhang, J. Murowchick, G. Liu, X. Chen,Nano Lett. 13 ,
    5289 – 5296 (2013).


11. M. Bañobre-Lópezet al.,Chem. Mater. 17 , 1965– 1968
    (2005).


12. B.Y.Wanget al.,J.Mater.Chem.A 4 , 15426– 15436
    (2016).


13. A. Liscioet al.,Adv. Funct. Mater. 16 , 1407–1416 (2006).


14. F. Bocquet, L. Nony, C. Loppacher, T. Glatzel,Phys. Rev. B 78 ,
    035410 (2008).


15. C. Melzer, C. Siol, H. von Seggern,Adv. Mater. 25 , 4315– 4319
    (2013).


16. R. Berthelot, D. Carlier, C. Delmas,Nat. Mater. 10 , 74– 80
    (2011).


17. K. Takadaet al.,Adv. Mater. 16 , 1901–1905 (2010).


18. M. I. Asghar, S. Jouttijärvi, R. Jokiranta, A.-M. Valtavirta,
    P. D. Lund,Nano Energy 53 , 391–397 (2018).


19. B. Zhuet al.,Nano Energy 19 , 156–164 (2016).


20. A. J. Jacobson,Chem. Mater. 22 , 660–674 (2010).


21. V. Esposito, E. Traversa,J. Am. Ceram. Soc. 91 , 1037– 1051
    (2008).

SCIENCEsciencemag.org 10 JULY 2020•VOL 369 ISSUE 6500 187

Fig. 4. Proton transport path.(A)Charge density differences at the NCO/CeO 2 interface. LEF, local electric
field. (B) Spin-resolved density of states for fresh NCO-CeO 2 and hydrogenated NCO-CeO 2 .(C) Spin-resolved
partial density of states of H diffusion at the NCO-CeO 2 interface. (D) Energy diagram of H diffusion
at the NCO/CeO 2 interface. (E) Corresponding charge transfer behavior for the initial (i), transition (ii), and
final (iii) states of H absorption at the NCO/CeO 2 interface with a Na vacancy.

RESEARCH | REPORTS