CHIN. PHYS. LETT. Vol. 27, No.

10 (2010) 107304

*
Rectifying and Photovoltage Properties of ZnO:Al/p-Si Heterojunction
MA Jing-Jing(马晶晶), JIN Ke-Xin(金克新), LUO Bing-Cheng(罗炳成), FAN Fei(范飞), XING Hui(邢辉),
ZHOU Chao-Chao(周超超), CHEN Chang-Le(陈长乐)**
Shaanxi Key Laboratory of Condensed Matter Structures and Properties, School of Science, Northwestern
Polytechnical University, Xi’an 710072

(Received 11 May 2010)

An Al-doped ZnO/p-Si heterojunction is fabricated by a laser molecular beam epitaxy technique. The abnormally
high ideality factors (𝑛 ≫ 2) of the prepared heterojunction are observed in the interim bias voltage range. A
theoretical model is proposed to understand the much higher ideality factor of the special heterojunction diode.
The ZnO:Al film shows metal-like conductivity with the electrical resistivity about 6.56 × 10−4 Ω·cm at room
temperature. The temperature dependence of the photovoltage indicates that the photovoltaic effect of the
Al-doped ZnO based heterojunction can be changed by the intrinsic metal-semiconductor transition at 120 K.

PACS: 73. 40. Lq, 73. 50. Pz, 73. 61. Ga DOI: 10.1088/0256-307X/27/10/107304

As a wide band gap oxide semiconductor, ZnO is with repeated grinding and sintering at 1100∘ C in
an ideal material for ultraviolet (UV) optoelectronic air. The Al-doped ZnO thin film was grown on a
applications. ZnO can be doped with various Group- lightly boron-doped silicon substrate by a laser molec-
III elements such as B, Al, Ga and In to induce n- ular beam epitaxy technique, using a KrF excimer
type conductivity.[1−4] In particular, Al is an efficient laser of 248 nm wavelength and 25 ns pulse duration.
n-type dopant for ZnO to produce high-quality, good- The pulse energy and laser frequency were kept as
stability samples with low resistivity and high trans- 140 mJ and 1 Hz, respectively. Before transferred into
parency in the visible range.[5] Considering the diffi- the deposition chamber, the Si substrate was cleaned
culty of the reproducible p-type doping of ZnO with by organic solvents and then etched in the 10% HF
high concentration and hole mobility, the ZnO-based acid for 30 s to remove the contaminations and sur-
p-n heterojunction has received much attention as an face oxides. During the deposition, pure oxygen was
alternative pathway for optoelectronics applications. introduced into the chamber and the working pres-
The heterojunctions were fabricated by depositing the sure was maintained at 1.4 × 10−3 Pa. The substrate
n-ZnO epilayer on various p-type materials, such as was kept at 450∘ C with a rotation speed of 10 loops
Si, GaN, La0.7 Sr0.3 MnO3 , and SrCu2 O2 .[6−9] Among per minute. In this way, the AZO/p-Si heterojunction
these heterojunctions, the n-ZnO/p-Si heterojunction was obtained and the thickness of the AZO film was
is not only more cost-effective for device fabrication about 135 nm measured by a SpecEI-2000-VIS ellip-
but also promising for the integration with silicon mi- someter. The XRD pattern of the heterojunction indi-
croelectronic devices. In the ZnO:Al/p-Si heterojunc- cates that the AZO film was highly textured along the
tion, the Al-doped ZnO (AZO) film was deposited on 𝑐 axis and aligned with the (004) peak of Si. The car-
p-Si substrates as a practical photo-window as well as rier concentration, determined from Hall-effect mea-
a semiconducting layer, causing a depletion region in surements, was about 6.4 × 1015 cm−3 for the Si and
the p-n heterojunction. In addition, the AZO contact 7.2 × 1020 cm−3 for the AZO film.
layer can efficiently enhance the photoresponse of the The photoluminescence measurement was per-
heterojunction due to its metal-like property.[10] How- formed on an FM-4 (Jobin Yvon) fluorescence spec-
ever, little research on this type of heterojunction has trometer with the excitation wavelength of 300 nm.
been reported.[11] It is important to understand the The heterojunction was placed in a Janis VPF 475
physical properties and the transport mechanisms of closed-circuit liquid nitrogen cryostat with a quartz
the AZO based devices prior to applications. In the window. Current versus voltage (𝐼 − 𝑉 ) measurement
present work, we fabricate the AZO/p-Si heterojunc- was performed using a Keithley 2182A nanovoltmeter
tion by a laser molecular beam epitaxy method and and a 6485 picoammeter at room temperature. The
reveal the electrical and photoelectric properties of the electrical resistivity was measured using a standard
heterojunction in detail. four-probe technique in the temperature range of 40
A nominal ZnO target doped with 2.0 wt% Al2 O3 to 300 K. The light source used in the photovoltaic ef-
was prepared using a solid state reaction technique fect experiment is an Nd:YVO4 continues-wave laser

* Supported by the NPU Foundation for Fundamental Research under Grant Nos NPU-FFR-JC200821 and JC201048, the

National Natural Science Foundation of China under Grant No 50702046, and the NWPU “Aoxiang Star” project.
** Email: chenchl@nwpu.edu.cn

○c 2010 Chinese Physical Society and IOP Publishing Ltd

107304-1
 CHIN. PHYS. LETT. Vol. 27, No. 10 (2010) 107304

with the wavelength of 532 nm and the power level factor 𝑛 can be evaluated by fitting the experimental
of 30 mW. For the transient photovoltage measure- data using Eq. (1). The values of 𝑛 obtained from the
ments, the laser is chopped and split by a beam split- ln 𝐼 − 𝑉 curve were 9.45 in the voltage range of 0.28–
ter mirror. The photovoltage of the heterojunction 0.88 V and 18.83 in the voltage range of 0.88–1.63 V.
was measured by a 500 MHz mixed signal oscilloscope It is known that the ideality factors in an ideal p-n
(Tektronix MSO 4054) with an input impedance of heterojunction are around 1.0 at a low voltage and up
1 MΩ. The setup of the photovoltaic voltage measure- to 2.0 at a higher voltage according to the Sah–Noyce–
ment and pulse duration of the laser are described Shockley theory.[13] Interestingly, the ideality factors
elsewhere.[12] are much higher than the values expected from the
Sah–Noyce–Shockley model in our cases. This may be
382 nm due to the presence of nonlinear metal-semiconductor
PL intensity (arb. units)

contact or defect states in the interface.

12
e2
10

Current (mA)
e1

Current (mA)
540 nm
8
e0 =18.83

6
e-1
350 400 450 500 550 600 4
Wavelength (nm) e-2 0 1 2 3 4
Fig. 1. Room temperature photoluminescence spectra of
2 Voltage (V)

the ZnO:Al/p-Si heterojunction.

0
Figure 1 shows the photoluminescence spectra of -6 -4 -2 0 2 4
the heterojunction at room temperature. There can Voltage (V)
be seen a narrow and sharp peak located in ultraviolet
Fig. 2. The 𝐼 − 𝑉 characteristics of the ZnO:Al/p-Si het-
region (382 nm) and a broad spectral band falling into
erojunction device. The inset shows a natural logarithm
the region between 480 and 600 nm. The near-band- plot of current vs bias. The curve can be simulated with
edge UV emission should be attributed to the free diode ideality factors 𝑛 = 9.45 in the bias range of 0.28 to
exciton recombination through exciton-exciton colli- 0.88 V and 𝑛 = 18.83 in the bias range of 0.88 to 1.63 V.
sion and the deep-level visible emission is caused by To obtain the Ohmic contact formation, a pair of
the presence of various point defects such as oxygen indium electrodes was plotted on the p-Si and n-ZnO
vacancies and interstitial Zn ions in the ZnO lattice. surfaces. Under accurately considered conditions, the
The distinctive contrast between the strong UV band heterojunction diode can be modeled by a series of
edge emission and weak green emission suggests that diodes or resistances in different bias voltage range:
the ZnO film has only a few defect states. the actual p-Si/n-ZnO heterojunction diode, the In/p-
Figure 2 shows the dark 𝐼 − 𝑉 characteristic of Si Schottky diode (or contact resistance), and the
the AZO/p-Si heterojunction measured at room tem- In/n-ZnO Schottky diode (or contact resistance). The
perature. The 𝐼 − 𝑉 curve exhibits a good rectifying n-ZnO thin film, with a work function (4.35 eV) larger
behavior with a rectification ratio (𝐼𝐹 /𝐼𝑅 ) of 4.6 × 102 than that of indium (4.12 eV), was heavily doped, so
at 3.0 V, indicating the characteristics of a diode. The that Ohmic rather than Schottky characteristics are
leakage current is about 2.6 × 10−5 A at the reverse exhibited in the In/n-ZnO contact. On the contrary,
bias of 4.0 V while the threshold voltage 𝑉𝐷 is 0.64 V. the p-Si wafer was lightly doped and its work function
A plot of ln 𝐼 versus 𝑉 (ln 𝐼 − 𝑉 ) is shown in the in- (4.85 eV) was larger than that of indium. The In/p-Si
set of Fig. 2. In the voltage range of 0.28–1.63 V, the junction therefore may exhibits nonlinear characteris-
current increases exponentially following the standard tics and can be considered as a reverse-biased Schottky
diode equation contact in the interim forward bias range.[14,15] Con-
[︂ (︁ 𝑒𝑉 )︁ ]︂ sequently, the device can be modeled by a series of
𝐼 = 𝐼𝑆 exp −1 , (1) diodes in the forward bias range of 0.28–1.63 V: the
𝑛𝑘𝑇
actual p-Si/n-ZnO heterojunction diode, the In/p-Si
where 𝐼𝑆 is the reverse saturation current, 𝑉 is the Schottky diode and the negligible In/ZnO contact re-
forward-biased voltage, 𝑘 is the Boltzmann constant, sistance. A summation of the two diodes, each with
𝑇 is the absolute temperature, and 𝑛 is the ideality different values of the reverse current and ideality fac-
factor, which is a quantity for describing the deviation tors, is generally used to explain the various conduc-
of the diode from an ideal p-n junction. The ideality tion mechanisms. According to the Shah–Li–Schubert

107304-2
 CHIN. PHYS. LETT. Vol. 27, No. 10 (2010) 107304

model,[16] the external

∑︀ current and voltage are given 150
by 𝐼 and 𝑉 = 𝑉𝑖 , respectively. The 𝐼 − 𝑉 charac- 100 280 K
teristic of the structure is given by 50
∑︁ ∑︁ [︂ (︁ 𝑘𝑇 )︁ (︁ 𝑘𝑇 )︁ ]︂ 0
𝑉 = 𝑉𝑖 = 𝑛𝑖 ln 𝐼 − 𝑛𝑖 ln 𝐼Si . 150
𝑞 𝑞
𝑖 𝑖
(2)
100 200 K

50

(mV)
Assuming that the diode voltages are 𝑉𝑖 ≫ 𝑘𝑇 /𝑞 and 0
rearranging the terms of Eqs. (1) and (2) yields

oc
150
100 120 K
∑︀
(𝑞/𝑘𝑇 ) 𝑛𝑖 ln 𝐼Si
ln 𝐼 = ∑︀ 𝑉 + 𝑖∑︀ . (3)
𝑖 𝑛𝑖 𝑖 𝑛𝑖
50
0
In Eq. (3), the second summand is a constant when
we consider only the linear region in the ln 𝐼 − 𝑉
150 80 K
100
curve, where 𝑛𝑖 are constants. According to the first 50
summand of Eq. (3), the ideality factor measured ex- 0
ternally is the sum of the ideality factors of the in-
dividual rectifying
-1.0 -0.5 0.0 0.5 1.0 1.5 2.0
∑︀ junctions. This result can be ex- Time (ms)
pressed as 𝑛 = 𝑖 𝑛𝑖 , where 𝑛𝑖 represent the ideality
Fig. 3. Photovoltages as a function of time measured at
factors of ZnO/p-Si heterojunction and reverse-biased
different temperatures.
In/p-Si junction. In a metal/semiconductor Schottky
diode, the ideality factor for reverse bias is given by
𝑛′ = (1 − 𝑛−1 )−1 , where 𝑛 is the ideality factor for 180 (a)
forward bias.[16] Note that the reverse-biased ideality
150
factor 𝑛′ is much larger than 2, and thus, the high
Up (mV)

ideality factors (𝑛 ≫ 2) of the device can be observed. 120

Furthermore, we note that the value of 𝑛 is as high
as 18.83 in the voltage range of 0.88–1.63 V, which is 90

deviated much from a conventional diode. Similar re- 60

sults were observed by Raddy et al.[17] in the n-ZnO
nanorods/p-Si devices (𝑛 > 20) and suggested that 660 (b) Oscilloscope
Resistivity (mWScm)

a
the observed 𝑛 at higher voltages is probably due to
Laser
the presence of defect states in ZnO lattice and/or the 640
In
presence of traps at the interface.[18] With the further AZO
p-Si
In

increase of the forward bias, the equivalent circuit of 620

the device turns out to be three resistances instead of
the series of diodes. As a result, the 𝐼 − 𝑉 curve in 600
Fig. 2 becomes linear when the bias voltage is larger 40 80 120 160 200 240 280 320
than 1.63 V. Temperature (K)
The time dependence of the photovoltaic singles Fig. 4. (a) Temperature dependence of peak photovoltage
(𝑉oc ) is shown in Fig. 3. The photovoltages increase 𝑈𝑝 of the ZnO:Al/p-Si heterojunction device. (b) Temper-
quickly to the maximum values at about 300 µs, and ature dependence of the electrical resistivity of the AZO
then decrease gradually. The peak photovoltages 𝑈𝑝 film. The inset shows the schematic setup for the photo-
voltage measurement.
are about 178 mV and 52 mV at 120 and 280 K, respec-
tively. For the 532 nm green laser light, the photon The temperature dependence of the peak photo-
energy is nearly 2.33 eV, which is larger than the band voltage 𝑈𝑝 is shown in Fig. 4(a). It is observed that
gap of Si (about 1.12 eV), and less than the band gap the heterojunction shows a positive temperature coef-
of AZO (about 3.37 eV). When the heterojunction is ficient of peak photovoltage (TCP) below 120 K and
irradiated by this modulated laser, the photons are a negative TCP above it. It is known that the pho-
mainly absorbed in the Si layer where the electron- tovoltages are usually related to the concentration of
hole pairs are generated. The photogenerated elec- thermal charge carriers and the width of the depletion
trons diffuse into the depletion region and then swept layer. At low temperatures, a large bias voltage is re-
to the AZO layer by the built-in field. To counter- quired to establish a current that counterbalance the
act the current produced, an electric field has to be photocurrent due to the reduction of the concentration
established between the two electrodes of the hetero- of thermal charge carriers. This actually implies an
junction, leading to the occurrence of a photovoltages. increase of the photovoltage.[19] However, this theory

107304-3
 CHIN. PHYS. LETT. Vol. 27, No. 10 (2010) 107304

cannot account for the positive TCP observed below the anomalously high ideality factor of the prepared
120 K in our cases. To understand the abnormal ob- p-n junction diode. The transient photovoltage mea-
servation, we measured the electrical resistivity of the surements reveal that the photovoltaic effect of the
AZO film and the temperature dependence of resistiv- Al-doped ZnO based heterojunction can be changed
ity is shown in Fig. 4(b). We note that the ZnO:Al film by intrinsic metal-semiconductor transition.
with electrical resistivity about 6.56 × 10−4 Ω·cm at
room temperature undergoes a metal-semiconductor
transition (MST) at about 120 K. It shows the nega- References
tive temperature coefficient of resistivity (TCR) below [1] Zhao S Q, Liu W W, Yang L M, Zhao K and Liu H 2009
the MST, which is a characteristic of a semiconduct- J. Phys. D: Appl. Phys. 42 185101
ing behavior, and the positive TCR metal-type con- [2] Cornelius S, Vinnichenko M, Shevchenko N, Rogozin A and
ductivity behavior above MST. This is generally ex- Kolitsch A 2009 Appl. Phys. Lett. 94 042103
[3] Bhosle V, Tiwari A and Narayan J 2006 Appl. Phys. Lett.
plained by the formation of a degenerate band in heav- 88 032106
ily doped semiconductors, as suggested by Mott.[20] [4] Jung M N, Koo J E, Oh S J, Lee B W and Lee W J 2009
The observation of MST in Al-doped ZnO is consis- Appl. Phys. Lett. 94 041906
[5] Sernelius B E, Berggren K F, Jin Z C, Hamberg I and
tent with the results reported by Pradhan et al.[21] and
Granqvist C G 1988 Phys. Rev. B 37 10244
similar observations have also been reported in the [6] Fu Z X, Lin B X and Liao G H 1999 Chin. Phys. Lett. 16
case of Ga, As and Er-doped ZnO.[22−24] The MST 753
may cause the change of the thickness of the deple- [7] Park W I and Yi G C 2004 Adv. Mater. 16 87
[8] Sun Z H, Ning T Y, Zhou Y L, Zhao S Q and Cao L Z 2008
tion layer, which has a close relation to the diffuse
Chin. Phys. Lett. 25 1861
potential that is determined by the relative difference [9] Ohta H, Orita M, Hirano M and Hosono H, 2001
of the band structure of Si and AZO.[25] Below the J. Appl. Phys. 89 5720
MST, the energy offset between the Fermi levels of [10] Yang C, Li X M, Yu W D, Gao X D, Cao X and Li Y Z
2009 J. Phys. D: Appl. Phys. 42 152002
the n-AZO and p-Si diminishes due to the eliminating [11] Lei H, Liu C H, Lin B X and Fu Z X 2005 Chin. Phys. Lett.
degeneration of the AZO film. As a result, the width 22 185
of the depletion region becomes thinner and the posi- [12] Jin K X, Zhao S G, Tan X Y and Chen C L 2008 Mater. Lett.
tive TCP is observed below 120 K. 62 4452
[13] Sah C, Noycc R N and Shockley W 1957 Proc. IRE 45 1228
To verify the transient photovoltage is a photo- [14] Rhoderick E H and Kim O K 1988 Metal Semiconductor
electric effect instead of a thermoelectric effect, we Contacts 2nd edn (Oxford: Oxford University) p 129
changed the laser spot position along the two elec- [15] He J H and Ho C H 2007 Appl. Phys. Lett. 91 233105
trodes direction (shown in the inset of Fig. 4) and no [16] Shah J M, Li L Y, Gessmann T and Schubert E F 2003
J. Appl. Phys. 94 2627
variation of the photovoltage was observed. These re- [17] Raddy N K, Ahsanulhaq Q, Kim J H and Hahn Y B 2008
sults provide evidence that there is no temperature Appl. Phys. Lett. 92 043127
gradient formed between the illuminated and the non- [18] Mridha S and Basak D 2007 J. Appl. Phys. 101 083102
illuminated areas, and thus the photovoltages occur- [19] Sun J R, Shen B G, Sheng Z G and Sun Y P 2004
Appl. Phys. Lett. 85 3375
ring in the ZnO/p-Si heterojunction originate from the [20] Mott N F 1974 Metal-Insulator Transition (London: Taylor
photoelectric effect. and Francis)
In conclusion, the Al-doped ZnO/p-Si heterojunc- [21] Bamiduro O, Mustafa H, Mundle R Konda R B and Prad-
han A K 2007 Appl. Phys. Lett. 90 252108
tion has been successfully fabricated. The rectifying
[22] Lu Z L, Zou W Q, Xu M X, Zhang F M and Du Y W 2009
and photovoltage properties of the device have been Chin. Phys. Lett. 26 116102
investigated in detail. The heterojunction shows good [23] Kord K, Williams T M, Hunter D, Zhang K, Dadson J and
rectifying behavior with a rectification ratio of about Pradhan 2006 Appl. Phys. Lett. 88 262105
[24] Pradhan A K, Douglas L, Mustafa H, Mundle R, Hunter D
4.6×102 at 3.0 V in the dark at room temperature. To
and Bonner C E 2007 Appl. Phys. Lett. 90 072108
gain a better understanding of the measured electrical [25] Sze S M 1983 Physics of Semiconductor Devices (New York:
properties, a theoretical model is proposed to analyze Wiley)

107304-4