Annealed strontium titanate (SrTiO3 or STO) single crystals exhibit persistent photoconductivity (PPC) at room temperature. Illumination with sub-gap light reduces the resistance by three orders of magnitude, which persists for weeks or longer. The defects responsible for this remarkable phenomenon have not been identified. In this work, we report on the importance of hydrogen and oxygen during the annealing process that is used to induce PPC. The results from IR spectroscopy and two-point resistance measurements indicate that water vapor at 1200 °C yields hydrogen and oxygen-vacancy populations that result in large PPC. Deuterium substitution experiments show evidence for a two-hydrogen center that forms after exposure to light. The results suggest that light causes substitutional hydrogen to leave the oxygen site, forming metastable O-H bonds. This process liberates electrons and causes PPC.

Strontium titanate (SrTiO3 or STO) is a transparent conducting oxide material with an indirect bandgap of 3.2 eV.1 STO has been utilized for a wide variety of applications, from high temperature oxygen sensors2 to a photo-catalyst used to hydrolyze water.3 This material has the perovskite structure and can be used as a substrate for thin films, some of which display a highly conducting layer at the interface.4 

Specific annealing treatments can induce persistent photoconductivity (PPC), where samples go from insulating to conductive upon exposure to light of energy 2.9 eV or higher.5 The conductivity change is three orders of magnitude, occurs at room temperature, and is stable on the order of weeks or longer. This makes STO unique since large PPC in single crystals is normally observed only at low temperatures.6 The dramatic increase in conductivity can be observed by electrical measurements as well as the increase in free carrier absorption in the infrared (IR) region of the spectrum.7 Additionally, conductive paths can be drawn with light on a resistive sample,8 opening up photo-lithographic applications.

Figure 1 shows two-point resistance measurements of an STO sample that was ∼1 MΩ prior to light exposure. After illumination with 405 nm light, the resistance dropped to ∼1 kΩ. The resistance was measured in the dark over ∼1 year. The data were fitted empirically to a sum of two exponential functions with time constants of 17 days and 800 years. While the origins behind the detailed time dependence are not known, the long-time behavior implies that the resistance change is essentially permanent at room temperature.

FIG. 1.

Resistance of the STO sample after exposure to 405 nm light. Measurements were taken in the dark. The resistance of the sample prior to illumination was ∼1 MΩ.

FIG. 1.

Resistance of the STO sample after exposure to 405 nm light. Measurements were taken in the dark. The resistance of the sample prior to illumination was ∼1 MΩ.

Close modal

To achieve large PPC, SrO powder is placed in a sealed ampoule, along with the STO sample, and annealed at 1200 °C in vacuum. Recent results have shown that annealing in vacuum prior to the “SrO” anneal is sometimes a necessary step to condition the crystals.9 The vacuum anneal introduces oxygen vacancies,10 which are necessary for PPC. The defect responsible for PPC and the role of the SrO powder are yet to be determined. In this paper, we report evidence that the presence of hydrogen and oxygen during the 1200 °C anneal plays a critical role. These observations enable us to propose a tentative model for the mechanisms behind PPC.

The annealing recipe for inducing PPC in STO involves sealing a bulk single crystal of STO in a fused silica ampoule with 0.5 g strontium oxide (SrO) powder under rough vacuum. The sample space inside the ampoule has a length of approximately 7.6 cm and a diameter of 1.6 cm. The ampoule is then annealed in a horizontal tube furnace for 1 h at 1200 °C. Samples are promptly removed and allowed to cool in the ambient air in the dark, which takes about 10 min.

IR spectra were obtained using a Bomem DA8 vacuum Fourier transform infrared (FTIR) spectrometer. Powder spectra were taken at room temperature using a mercury cadmium telluride (MCT) detector. Low temperature spectra were taken using a Janis closed-cycle helium cryostat at a resolution of 1 cm−1, with an indium antimonide (InSb) detector. A 405 nm light emitting diode (LED) was placed inside the cryostat to allow exposure to occur without moving the sample or breaking vacuum.

Anhydrous Sr(OH)2 and SrO powders were purchased from Sigma-Aldrich. Deuterated strontium hydroxide, Sr(OD)2, was prepared by placing SrO powder and heavy water (D2O) in a sealed container (humidity chamber) for 2 days. D2O formed heavy water vapor in the chamber, which was readily absorbed by strontium oxide, forming Sr(OD)2. The mass of the powder was weighed before (1.0 g) and after (1.8 g) being in the humidity chamber. The corresponding increase in mass indicated that the powder absorbed ∼4 D2O molecules per SrO (supplementary material).

The presence of SrO powder during the 1200 °C anneal was found to be essential for large PPC. Other ambient conditions, such as annealing in Ar without any powder, resulted in a highly n-type material.11 The reason for this was not obvious. To investigate further, we performed the 1200 °C anneal using newly received SrO powder. Rather than being in a high resistance state after annealing and before light exposure, samples were conductive (∼300 Ω with 2 point pressed indium contacts). This indicated that SrO powder must age in the ambient atmosphere in order to yield optimal PPC.

SrO powder was mixed with KBr, to be sufficiently transparent, and pressed into pellets. The material from a newly opened bottle was compared with the material that experienced 6 days of exposure to the atmosphere. Several additional IR peaks appear in the sample exposed to air (Fig. 2) and are attributable to strontium hydroxide, Sr(OH)2 or Sr(OH)2 nH2O, as well as strontium carbonate, SrCO3,12,13 The CO32− anion has a broad absorption band centered at around 1445 cm−1, attributed to the asymmetric stretching vibration, while the 866 and 599 cm−1 lines are the bending vibrations.14 The peak at 3590 cm−1 is due to the stretching mode of OH in Sr(OH)2, while the broad absorption centered at around 2835 cm−1 is attributed to the stretching mode of H2O in Sr(OH)2 nH2O.12 This shows that SrO absorbs water and carbon dioxide, which may be released during the high temperature anneal (see supplementary material for more details of the chemistry of SrO).

FIG. 2.

IR absorption spectrum of SrO powder after exposure to the atmosphere for 6 days, mixed with KBr. Freshly opened SrO powder mixed with KBr was used as a reference.

FIG. 2.

IR absorption spectrum of SrO powder after exposure to the atmosphere for 6 days, mixed with KBr. Freshly opened SrO powder mixed with KBr was used as a reference.

Close modal

1. Evidence for PPC

Strontium hydroxide is a contaminant species in SrO powder. To test the effect of strontium hydroxide on PPC, a small amount (0.1 g) of anhydrous Sr(OH)2 was placed in the ampoule, without any SrO. To prevent premature decomposition of the strontium hydroxide powder during the sealing process with a hydrogen-oxygen torch, the powder was placed in a heat-sinked end of the ampoule. The sample was annealed via the usual procedure (Sec. II). At 1200 °C, Sr(OH)2 decomposes into SrO and H2O (supplementary material).

This sample exhibits PPC, as measured by two methods. First, light exposure caused a dramatic decrease in the transmitted light intensity through the sample, which corresponds to a large increase in the free carrier absorption (Fig. 3). Second, the 2-point resistance of the sample at room temperature, using pressed indium contacts, decreased by a factor of 400 (Table I). This result demonstrates that water vapor at high temperatures can induce PPC.

FIG. 3.

Low temperature (125 K) IR spectra of STO annealed with 0.1 g Sr(OH)2 under vacuum or 1/2 atm hydrogen or oxygen gas. Spectra before light exposure were used as the reference for the absorbance plots.

FIG. 3.

Low temperature (125 K) IR spectra of STO annealed with 0.1 g Sr(OH)2 under vacuum or 1/2 atm hydrogen or oxygen gas. Spectra before light exposure were used as the reference for the absorbance plots.

Close modal
TABLE I.

Summary of 2-point pressed indium resistance values taken before and after light exposure for various sample conditions.

Annealing conditionsResistance before light exposureResistance after light exposure
Sr(OH)2 4 MΩ 118 Ω 
Sr(OH)2+ ½ atm H2 300 kΩ 1.2 kΩ 
Sr(OH)2+ ½ atm O2 >200 MΩ 200 MΩ 
Sr(OD)2 99 MΩ 3.0 kΩ 
Sr(OH)2+Sr(OD)2 38 MΩ 83 kΩ 
H2>200 MΩ 96 kΩ 
CO2 33 kΩ 172 Ω 
Annealing conditionsResistance before light exposureResistance after light exposure
Sr(OH)2 4 MΩ 118 Ω 
Sr(OH)2+ ½ atm H2 300 kΩ 1.2 kΩ 
Sr(OH)2+ ½ atm O2 >200 MΩ 200 MΩ 
Sr(OD)2 99 MΩ 3.0 kΩ 
Sr(OH)2+Sr(OD)2 38 MΩ 83 kΩ 
H2>200 MΩ 96 kΩ 
CO2 33 kΩ 172 Ω 

To further explore the role of water and its components, we back-filled the evacuated ampoule with approximately 0.5 atm of either hydrogen or oxygen along with 0.1 g of Sr(OH)2 powder. The sample annealed in the hydrogen rich atmosphere showed PPC (Fig. 3). The PPC is less dramatic than the evacuated-ampoule anneal, with a 2 order of magnitude resistance change (Table I). Hydrogen is a reducing atmosphere and could cause more oxygen vacancies to be introduced,15 thereby lowering the resistance of the before-light state by introducing more free carriers. The oxygen rich atmosphere, in contrast, did not show PPC (Fig. 3). This is consistent with previous results, indicating that oxygen vacancies are necessary for PPC.9 Additional oxygen during the anneal suppresses the formation of oxygen vacancies, resulting in a resistive material that is not photo-sensitive.

2. Hydrogen vibrational modes

The predominant hydrogen line in STO, labeled HI, has an O-H bond-stretching peak near 3500 cm−1 (Refs. 16–18). While there is some debate in the literature, recent first-principles calculations attribute the HI line to (VSr-H) single acceptor complexes.19 The HI peak is often accompanied by multiple higher energy satellite lines, but the relative intensity of these lines varies from sample to sample.17 The results of calculations suggest that these modes may arise from passivated strontium vacancies (VSr-2H).20 The satellites and the main line shift similarly to higher wavenumbers with decreasing temperature, with the main line ∼3510 cm−1 at 125 K.21 

Deuterated strontium hydroxide was used in two anneals with a vacuum atmosphere. One sample had 0.1 g of Sr(OD)2, and the other contained a mixture of 0.05 g of Sr(OD)2 and 0.05 g of Sr(OH)2. Otherwise, the samples were annealed via the standard procedure. Both samples displayed PPC.

When hydrogen is replaced with deuterium, the main O-H line and four satellite lines shift downward in frequency by a factor of 1.35 (Fig. 4). This is close to the expected frequency ratio shift for a harmonic oscillator consisting of a hydrogen (deuterium) atom bound to an oxygen atom

μOD/μOH=1.37,
(1)

where μOH and μOD are the reduced masses of O-H and O-D molecules, respectively. All the observed satellite lines correspond to those identified by Grone et al.22 After light exposure, the intensity of the main line decreases, while those of the satellite lines increase.

FIG. 4.

Low temperature (125 K) IR spectra comparing the hydrogen and deuterium lines of an annealed sample. The main O-H line and 3 satellites show the expected frequency ratio of 1.35. Spectra were baseline subtracted by a linear fit.

FIG. 4.

Low temperature (125 K) IR spectra comparing the hydrogen and deuterium lines of an annealed sample. The main O-H line and 3 satellites show the expected frequency ratio of 1.35. Spectra were baseline subtracted by a linear fit.

Close modal

Mixed doping with both Sr(OH)2 and Sr(OD)2 shows both the O-H and O-D main line and satellites, with lower intensities (Fig. 5). Upon light exposure, an OH (OD) peak appears at 3512 (2595) cm−1. Additionally, the first satellite at 3518 cm−1 is actually an unresolved doublet,17 and the apparent shift of <1 cm−1 may be due to the change in the relative intensity of the two lines. These results suggest that there is a 2H (2D) center with an IR-active mode at 3518 cm−1 (2599 cm−1). The peaks at 3512 and 2595 cm−1 arise from HD centers.

FIG. 5.

Low temperature (125 K) IR spectra showing the deuterium and hydrogen lines and satellites after annealing with Sr(OH)2, Sr(OD)2, and a mixture of the two. Spectra before light exposure were used as the reference for the absorbance plots, which were baseline subtracted by a linear fit. Peaks that are attributed to a complex with hydrogen and deuterium are labeled HD.

FIG. 5.

Low temperature (125 K) IR spectra showing the deuterium and hydrogen lines and satellites after annealing with Sr(OH)2, Sr(OD)2, and a mixture of the two. Spectra before light exposure were used as the reference for the absorbance plots, which were baseline subtracted by a linear fit. Peaks that are attributed to a complex with hydrogen and deuterium are labeled HD.

Close modal

The results from the Sr(OH)2 anneals suggest that water vapor is crucial for PPC. To test this further, STO was annealed under water vapor without any powder. To minimize the evaporation of water under vacuum, we froze 0.04 g of water and placed the ice inside a heat-sinked ampoule with the sample. Approximately half the water remained after sealing, which corresponds to a pressure of around 10–15 atmospheres of water vapor during the anneal. This ampoule was annealed under the standard conditions (Sec. II). The sample displays PPC, which shows that water vapor alone is responsible for PPC (Fig. 6). The hydrogen lines are similar to those observed in the Sr(OH)2 anneal, except that the weak line at 3531 cm−1 is not observed and an additional satellite line is observed at 3542 cm−1.

FIG. 6.

Low temperature (125 K) IR spectra of STO annealed in water vapor. The spectrum before light exposure was used as the reference.

FIG. 6.

Low temperature (125 K) IR spectra of STO annealed in water vapor. The spectrum before light exposure was used as the reference.

Close modal

The effect of carbon dioxide during annealing was tested by sealing STO inside an ampoule with 0.5 atm of CO2 gas. Here, the sample space was larger, with a length of 12.7 cm. Otherwise, the ampoule was annealed under standard conditions (Sec. II). The sample displays PPC although somewhat weakly (Fig. 7). Here, the hydrogen lines are quite weak, with only the main line detected and increasing upon light exposure. The behavior of this peak is the opposite to what is observed when the annealing atmosphere is H2O. Since the CO2 annealing environment does not contain hydrogen, it is likely that most of the hydrogen in the sample was present as-received since hydrogen is a ubiquitous impurity in oxides and easily incorporated into STO.16 

FIG. 7.

Low temperature (125 K) IR spectra of the hydrogen line of STO annealed in CO2. The spectrum before light exposure was used as the reference.

FIG. 7.

Low temperature (125 K) IR spectra of the hydrogen line of STO annealed in CO2. The spectrum before light exposure was used as the reference.

Close modal

In this anneal, CO2 mainly acts as a source for oxygen via its partial dissociation

CO2CO+1/2O2.
(2)

A sample annealed under vacuum at the same temperature was much more conductive (∼5 Ω), due to the formation of many more oxygen vacancies, while an anneal under oxygen at this temperature resulted in a sample with high resistance (∼1 GΩ). This seems to indicate that carbon dioxide present in the annealing environment helps by gently tuning the oxygen vacancy concentration since only a small fraction of the carbon dioxide becomes oxygen. This is consistent with what we have seen before in regard to PPC's relation to oxygen vacancies. Some are needed, but too many results in an initially conductive sample, and so, the drop upon light exposure is less dramatic.

While it is conceivable that carbon from CO2 could diffuse into the sample, it is unlikely that it plays a role in PPC. Carbon's preferred site would be to substitute for the Ti4+ ion, and titanium is quite stable in the lattice, even at 1200 °C.23,24 CTi is an isoelectronic defect, since both carbon and titanium are group-IV elements, so would not introduce energy levels into the gap. Theoretical calculations show that CTi may result in a slight increase in the bandgap and possibly weakly hinder oxygen-vacancy diffusion in nearby sites.24 Significant amounts of CTi would be directly identifiable with local vibrational modes of the C-O stretch at 1414 cm−1 and the breathing stretch mode at 1081 cm−1,23 which we did not observe.

Table I lists a summary of the annealing conditions and 2 point resistance measurements before and after light exposure. The deuterium substituted samples start in a higher before-light resistance state. This may be due to a higher concentration of water vapor since the Sr(OD)2 powder had absorbed more water than the anhydrous Sr(OH)2 powder. The pressure of water vapor during the anneal may be correlated with the resistance value observed before light exposure. This may be because water has a very slight oxidizing effect, which will decrease the number of oxygen vacancies present.

On the basis of these measurements, we propose a model for PPC. For “hydrogen rich” anneals at 1200 °C, the sample contains (VSr-H) and HO impurities. Upon exposure to light, hydrogen moves from the oxygen site and forms (VSr-2H), liberating two electrons:

HO++(VSr-H)VO2++(VSr-2H)0+2e.
(3)

This results in a decrease in the (VSr-H) IR absorption peak (HI) and an increase in the sidebands (VSr-2H)0. For the hydrogen-poor CO2 anneal, most Sr vacancies are unpassivated. The PPC reaction is then given by

12VO2++HO++VSr23/2VO2++(VSr-H)+2e.
(4)

This leads to an increase in the (VSr-H) peak and no (VSr–2H)0 sidebands.

The proposed model is one possible explanation for the experimental observations. It is possible that some of the O-H bonds observed in the IR are not due to vacancy-hydrogen complexes. They could, for example, be acceptor-hydrogen pairs. In that case, electrons would also be liberated. The main point is that when H leaves its substitutional site, the oxygen vacancy is able to act as a shallow double donor. The formed O-H bonds are strong enough to prevent the hydrogen atoms from returning to their substitutional sites at room temperature.

In conclusion, STO annealed at 1200 °C in water vapor exhibits PPC. By altering the gas composition and measuring the O-H vibrational modes, we established that oxygen vacancies and hydrogen are essential defects for PPC. Specifically, we propose that exposure to light causes hydrogen to leave its substitutional site and form O-H bonds, liberating electrons. The O-H complexes are metastable and persist, along with the associated free carriers, for weeks or longer at room temperature. While this phenomenon was observed in STO, in principle, it could occur in any oxide where oxygen vacancies act as shallow double donors.

See supplementary material for Chemistry of SrO and Sr(OH)2 powders.

This research was funded by the National Science Foundation (DMR-1561419). Fruitful discussions with S. Limpijumnong, P. Sushko, and J. Varley are gratefully acknowledged.

1.
M.
Capizzi
and
A.
Frova
,
Phys. Rev. Lett.
25
,
1298
(
1970
).
2.
W.
Menesklou
,
H.-J.
Schreiner
,
K. H.
Hardtl
, and
E.
Ivers
,
Sens. Actuators B
59
,
184
(
1999
).
3.
E. D.
Jeong
,
S. M.
Yu
,
J. H.
Yoon
,
J. S.
Bae
,
C. R.
Cho
,
K. T.
Lim
,
P. H.
Borse
, and
H. G.
Kim
,
J. Ceram. Process. Res.
13
,
517
(
2012
), ISSN 1229-9162.
4.
Y. Z.
Chen
,
D. V.
Christensen
,
F.
Trier
,
N.
Pryds
,
A.
Smith
, and
S.
Linderoth
,
Appl. Surf. Sci.
258
,
9242
(
2012
).
5.
M. C.
Tarun
,
F. A.
Selim
, and
M. D.
McCluskey
,
Phys. Rev. Lett.
111
,
187403
(
2013
).
6.
D. V.
Lang
and
R. A.
Logan
,
Phys. Rev. Lett.
39
,
635
(
1977
).
7.
W. S.
Baer
,
Phys. Rev.
144
,
734
(
1966
).
8.
V. M.
Poole
,
S. J.
Joleka
, and
M. D.
McCluskey
,
Sci. Rep.
7
,
6659
(
2017
).
9.
V. M.
Poole
,
J.
Dashdorj
,
M. E.
Zvanut
, and
M. D.
McCluskey
, “
Large persistent photoconductivity in strontium titanate at room temperature
,”
MRS Proc.
1792
(
2015
).
10.
R.
Meyer
,
R.
Waser
,
J.
Helmbold
, and
G.
Borchardt
,
Phys. Rev. Lett.
90
,
105901
(
2003
).
11.
M. D.
McCluskey
and
M. C.
Tarun
,
AIP Conf. Proc.
1583
,
319
(
2014
).
12.
M. A.
Alavi
and
A.
Morsali
,
Ultrason. Sonochem.
17
,
132
(
2010
).
13.
M. K.
Mondal
and
M.
Lenka
,
Fluid Phase Equilib.
336
,
59
(
2012
).
14.
P.
Pasierb
,
S.
Komornicki
,
M.
Rokita
, and
M.
Rekas
,
J. Mol. Struct.
596
,
151
(
2001
).
15.
J. F.
Scott
,
S. A. T.
Redfern
,
M.
Zhang
, and
M.
Dawber
,
J. Eur. Ceram. Soc.
21
,
1629
(
2001
).
16.
G.
Weber
,
S.
Kapphan
, and
M.
Wohlecke
,
Phys. Rev. B
34
,
8406
(
1986
).
17.
D.
Houde
,
Y.
Lepine
,
C.
Pepin
,
S.
Jandl
, and
J. L.
Brebner
,
Phys. Rev. B
35
,
4948
(
1987
).
18.
J. L.
Brebner
,
S.
Jandl
, and
Y.
Lepine
,
Phys. Rev. B
23
,
3816
(
1981
).
19.
J.
T-Thienprasert
,
I.
Fongkaew
,
D. J.
Singh
,
M.-H.
Du
, and
S.
Limpijumnong
,
Phys. Rev. B
85
,
125205
(
2012
).
20.
J. B.
Varley
,
A.
Janotti
, and
C. G.
Van de Walle
,
Phys. Rev. B
89
,
75202
(
2014
).
21.
A. F. W.
Klukhuhn
,
J.
Bruining
,
B.
Klootwijk
, and
J.
van der Elsken
,
Phys. Rev. Lett.
25
,
380
(
1970
).
22.
A.
Grone
,
M.
Wohlecke
,
S.
Kapphan
, and
R.
Waser
,
Ferroelectrics
107
,
97
(
1990
).
23.
M.
Al-Hadidi
,
J. P.
Goss
,
P. R.
Briddon
,
R.
Al-Hamadany
, and
M. E.
Ahmed
,
J. Phys.: Conf. Ser.
472
,
12006
(
2013
).
24.
M.
Al-Hadidi
,
J. P.
Goss
,
P. R.
Briddon
,
R.
Al-Hamadany
,
M.
Ahmed
, and
M. J.
Rayson
,
Ferroelectrics
497
,
9
(
2016
).

Supplementary Material