Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

Sensors and Actuators B 204 (2014) 250–272

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical


journal homepage: www.elsevier.com/locate/snb

Nanoscale metal oxide-based heterojunctions for gas sensing:


A review
Derek R. Miller, Sheikh A. Akbar ∗ , Patricia A. Morris
Department of Materials Science and Engineering, The Ohio State University, Columbus, OH 43212, United States

a r t i c l e i n f o a b s t r a c t

Article history: Metal oxide-based resistive-type gas sensors are solid-state devices which are widely used in a number
Received 2 June 2014 of applications from health and safety to energy efficiency and emission control. Nanomaterials such
Received in revised form 17 July 2014 as nanowires, nanorods, and nanoparticles have dominated the research focus in this field due to their
Accepted 19 July 2014
large number of surface sites facilitating surface reactions. Previous studies have shown that incorporat-
Available online 29 July 2014
ing two or more metal oxides to form a heterojunction interface can have drastic effects on gas sensor
performance, especially the selectivity. Recently, these effects have been amplified by designing hetero-
Keywords:
junctions on the nano-scale. These designs have evolved from mixed commercial powders and bi-layer
Nanostructure
Heterojunction films to finely-tuned core–shell and hierarchical brush-like nanocomposites. This review details the var-
Gas sensor ious morphological classes currently available for nanostructured metal-oxide based heterojunctions
Metal oxide and then presents the dominant electronic and chemical mechanisms that influence the performance of
Semiconductor these materials as resistive-type gas sensors. Mechanisms explored include p–n and n–n potential bar-
Nanomaterial rier manipulation, n–p–n response type inversions, spill-over effects, synergistic catalytic behavior, and
microstructure enhancement. Tables are presented summarizing these works specifically for SnO2 , ZnO,
TiO2 , In2 O3 , Fe2 O3 , MoO3 , Co3 O4 , and CdO-based nanocomposites. Recent developments are highlighted
and likely future trends are explored.
© 2014 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
2. Gas sensing mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
3. Heterostructure classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
3.1. Mixed composite structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
3.2. Bi-layer and multi-layer films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
3.3. Structures decorated with second-phase particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
3.4. Mixed 1D–1D structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
3.5. Core–shell structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
4. Mechanisms responsible for enhanced sensing performance in heterostructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
4.1. Role of the interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
4.1.1. Effects of p–n nanojunctions on gas sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
4.1.2. Effects of n–n and p–p nanojunctions in gas sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
4.1.3. Separation of unlike charge carriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
4.1.4. n–p response type inversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
4.2. Synergistic and complimentary behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
4.2.1. Complimentary decomposition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
4.2.2. Catalyzed spill-over effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
4.3. Manipulation of microstructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264

∗ Corresponding author. Tel.: +1 614 292 6725.


E-mail address: akbar.1@osu.edu (S.A. Akbar).

http://dx.doi.org/10.1016/j.snb.2014.07.074
0925-4005/© 2014 Elsevier B.V. All rights reserved.
D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272 251

5. Remaining challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265


5.1. Long-term stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
5.2. Reproducibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
5.3. Characterizing the dispersion state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
5.4. Understanding the mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
5.5. Useful analogues to catalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
5.6. Integration of nanostructured heterojunction materials into silicon microcircuit technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
6. Future outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Biographies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271

1. Introduction such as decrease in activation energy [15], targeted catalytic activ-


ity [16] and synergistic surface reactions [17]; and geometrical
Solid-state resistive-type metal oxide gas sensors (MOGS) have effects such as grain refinement [18], surface area enhancement
been widely used in applications ranging from health and safety [19], and increased gas accessibility [20]. Understanding the mech-
(e.g., medical diagnostics, air quality monitoring, food processing, anisms that control the sensing behavior in these heterostructures
and detection of toxic, flammable and explosive gases), to energy will be essential for future advances in this field. A simple summary
efficiency and emission control in combustion processes. These diagram of these mechanisms is shown in Fig. 1. Detailed explana-
sensors are attractive because of their high sensitivity, low cost, tions and examples of these mechanisms can be found in Sections
simplicity, and compatibility with modern electronic devices. 4 and 5.
MOGS are typically designed to be small, portable, and to inte- No review focused specifically on nanostructured oxide hetero-
grate easily with the environment or structure in their application. junction materials for gas sensing has previously been published.
However, resistive-type MOGS generally have the disadvantage of Therefore, the authors present here an overview of the past devel-
poor selectivity between gases, where many gas species may cause opments to understand the current progress of the state-of-the-art
changes in the resistance, making it impossible for a single sen- in gas sensing using heterostructures. Several other excellent
sor to properly identify the gas. This problem can be solved, in reviews on gas sensor materials [21], nanostructures in gas sensing
part, by using a suite of coordinated sensors, known as an “elec- [2,22–25], p-type oxides in gas sensing [26], and gas sensing funda-
tronic nose,” to simultaneously analyze several (up to 36 or more) mentals [27–29] may further aid in background and understanding
sensor signals at once [1] using calibrated software to correctly of the content presented here. In the present review, we first
identify the gases present [2]. However, further enhancements present the many morphological classes of heterostructures inves-
in the selectivity of the materials are necessary to simplify the tigated in the literature, then present some of the mechanisms that
electronic nose design and eliminate false-positive or interfering lead to the resulting enhanced performance, concluding with future
signals. directions to explore. It is the hope of the authors that compil-
Many recent studies have shown that the selectivity and other ing and comparing the information and results from these studies
important sensing parameters of resistive-type MOGS can be will lead the reader to a more coherent understanding of influ-
improved through the use of composite metal oxides [3–7]. encing factors and help facilitate further advances in the sensor
The physical interface between two dissimilar materials is often field.
referred to as a heterojunction. The material incorporating these
two components is known as a heterostructure. By creating inti- 2. Gas sensing mechanism
mate electrical contact at the interface between two dissimilar
semiconducting materials, the Fermi levels across the interface The nature of the fundamental operating principles of gas
can equilibrate to the same energy, usually resulting in charge sensing are described briefly here. In resistive-type sensors, a metal
transfer and the formation of a charge depletion layer. This is oxide is deposited across two or more electrodes which measure
the basis for unique effects that can lead to increased sensor the change in the electrical resistance of the oxide in the presence of
performance, and will be discussed later. Additional anomalies the analyte gas. The resistance of the oxide may increase or decrease
can occur by having two dissimilar semiconducting materials in on exposure to the gas depending on the dominant charge car-
close proximity, with both interfaces exposed to the atmosphere. rier and type of gas interacting with the surface, as described in
A gas or liquid may react with one material most readily, and Table 1. Consider an n-type semiconductor such as SnO2 . Under
this reaction byproduct may then react with the second mate- normal (ambient) operating conditions, there is significant oxygen
rial in order to complete the reaction, referred to as a synergistic adsorption on the metal oxide surface. Oxygen molecules in air dis-
reaction. Nano-heterostructures are often used in gas sensors sociate and each oxygen atom accepts an electron from the material
because their small dimensions and high surface-to-volume ratios (if n-type) to complete the bond, decreasing the electron density in
amplify the above effects. Dopant oxides may also be used sim- the material and increasing the resistance of the oxide. The oxy-
ply to control the microstructure and morphology of the material gen only adsorbs onto the surface and thus the electrons are only
during processing. Though microstructure manipulation is often removed to a certain depth from the surface known as the Debye
the claim for improved performance, there are likely electronic length, (), typically on the order of 2–100 nm. The region within
and chemical effects also in play that were not thoroughly a Debye length of the surface is known as the depletion region
elucidated. because it is depleted of its normal charge carriers. The Debye
The improvements in sensing performance of these composites length may change as more or less oxygen is adsorbed on the sur-
has been attributed to many factors, including electronic effects face, which in turn causes a measurable change in the resistance.
such as: band bending due to Fermi level equilibration [8,9], charge It has been overwhelmingly shown that when crystallite dimen-
carrier separation [10], depletion layer manipulation [11–13] and sions are driven below about 20 nm, sensor response drastically
increased interfacial potential barrier energy [14]; chemical effects increases [29]. When the nanostructure dimensions allow all atoms
252 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

Surface-dependent Interface-dependent Structure-dependent

Resistance Microstructural
Catalyc Reacon manipulaon manipulaon

Enhancement Spill-over Solid-state


Surface
Interface
Conducon
Decrease
Increase
sensizaon via channel surface
mechanism effect transformaon
Synergisc
PE barrier grain size
carrier injecon narrowing area

n-n juncon p-n juncon

Optimal
Oxide Amorphous
heterostructure Noble metal Mixed Spherical 1D Core- Polycrystalline
decorated Brush oxide parcle
for mechanism decoraon
nanowires
nanocrystalline core-shell shell nanowires
coang

Fig. 1. Summary of most cited sensor performance enhancing mechanisms and the types of heterostructures that are most optimal for each mechanism type. Some types of
mechanisms and structures have been omitted for space.

to be within a Debye length of the surface, the entirety of the mate- 3. Heterostructure classes
rial is depleted by the gas analyte, and the response is maximized.
If a reducing gas is introduced in this case, it reacts with the surface Before presenting the sensing mechanisms using different het-
adsorbed oxygen, pulling it from the surface, and simultaneously erostructures, it is necessary to introduce the various classes of
donating an electron back into the semiconductor that causes a heterostructure combination types that have arisen. This will aid
decrease in the resistance. P-type materials would exhibit reverse understanding of how the structure and dispersion state of the con-
behavior as shown in Table 1. stituents can affect the solid–gas interaction. For simplicity, three
The adsorption properties of the oxygen and reaction rates of different structure–architecture types will be represented in this
analyte gases are temperature-dependent, so most MOGS are also review by the nomenclature in the following general examples:
equipped with a heater that allows operation at a pre-determined
optimal operating temperature. Sensor response is defined in sev- • A hyphen between compounds (e.g. “SnO2 –ZnO”) represents a
eral different ways depending on the type of measurements taken, simple mixture of the two constituents, being mostly randomly
but is most often defined as the ratio of resistance in air to resis- distributed throughout the material.
tance in the presence of the gas, or Ra /Rg for an n-type material with • An “@” sign between compounds (e.g. “SnO2 @ZnO@CuO”) repre-
a reducing analyte, giving a value greater than 1. Alternatively, this sents a base material with the following oxides added over it in
becomes Rg /Ra for an n-type material with an oxidizing analyte, also some way. For example, aligned SnO2 nanorods, sputtered with
giving a value greater than 1. These are reversed in p-type materials. a layer of ZnO, and then dipped in a CuO precursor solution.
The parameter of sensitivity is defined as the degree to which the • A forward slash between compounds (e.g. “SnO2 /ZnO”) repre-
response increases as the concentration of that analyte increases, sents a well-defined partition or interface between the two
which can be found from the slope of a sensor’s calibration curve. oxides, as in a bi-layer, where the SnO2 is the top layer and the
The response and recovery times (tres and trec ) of a gas sensor are ZnO is the bottom layer.
usually defined as the time it takes for the resistance to reach 90% of • “(SnO2 –In2 O3 )/In2 O3 ” would then represent a mixture of SnO2
its steady-state value after introduction or removal of the analyte and In2 O3 nanoparticles deposited onto an In2 O3 film to form a
gas, respectively. Response, selectivity, sensitivity, operating tem- bi-layer.
perature, response time and recovery time are the most important
parameters for gas sensor performance. In most cases where the 3.1. Mixed composite structures
heterostructure improved selectivity, the sensing response to the
targeted analyte was also enhanced. Reductions in response and In composite electronic materials, it is very important to prop-
recovery times were also often reported, as were decreases in the erly characterize the dispersion state of each constituent in order
optimum operating temperature. to understand how the interfaces or heterojunctions between the

Table 1
Sensing response behavior of n-type and p-type materials to reducing and oxidizing gases.

Sensor response behavior n-type material p-type material Example gases

Reducing gases Resistance decreases Resistance increases H2 , H2 S, CO, NH4 , Ethanol, Acetone, CH4
Oxidizing gases Resistance increases Resistance decreases O2 , O3 , NOX , CO2 , SO2
Dominant charge carrier Electrons (e− ) Holes (h+ ) –
D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272 253

Fig. 2. Schematic showing the likely dispersion state of In2 O3 -based composites synthesized via the sol–gel process. Variants include core–shell (a, d and f) decorated with
second-phase particles (c and e), formation of mixed phases (d–f) and components in solid solution (b and f).
Reprinted with permission from Ref. [38].

constituents affect the behavior of the material as a sensor. The limits the insight that can be gained from the sensing results. Simi-
largest influence on the dispersion state comes from the processing lar mixed nanocrystalline films can be synthesized via advanced
routes, as well as the actual dopant used. It should be noted deposition techniques such as sputtering [44] and pulsed-laser
that even though the percent composition may be the same as deposition (PLD) [45]. These allow an even higher degree of control
another sample, the behavior can be quite different, as shown in of stoichiometry, film thickness, morphology, and growth kinetics
a direct comparison between TiO2 -doped SnO2 synthesized via by varying the gas atmosphere, power input, substrate temperature
co-precipitation as well as mechanical mixing yielding similar ran- and target material.
domized dispersions. However results showed that the optimum Several other unique morphologies of mixed oxide structures
composition for H2 sensing was 10 mol% for co-precipitation and have also been reported as shown in Fig. 3. Non-traditional
20 mol% TiO2 for mechanical mixing [30]. The easiest way to obtain morphologies have been created through simple hydrothermal
metal oxide composites is by simple mechanical mixing of exist- synthesis, including ZnSnO3 @SnO2 nanoflakes [20] and ZnO–Eu2 O3
ing oxide powders. de Lacy Costello et al. have published a series nanostrips [43] Additionally, our own work has demonstrated a
of studies [17,31,32] which showed that the addition of 25–50 wt% high-temperature gas etching process in a spinodally-decomposed
ZnO powder to SnO2 powder increased sensitivity to organic com- SnO2 –TiO2 pellet, which increased the surface area by creating
pounds, citing a synergistic mechanism in the decomposition of heavily faceted grains [46]. These structures are reported to exhibit
these molecules. In other studies, the addition of TiO2 powders to improved sensing performance [20,43,46].
SnO2 has shown to increase sensitivity to H2 [8,30,33].
The majority of oxide composites and mixtures in recent 3.2. Bi-layer and multi-layer films
literature are random dispersions of the oxide constituents syn-
thesized via various solution-based processes such as the sol–gel Heterostructures with a large, defined interface such as bi-layers
method [34–39], microemulsion [19], precipitation [9,30,40–42] and multi-layers allow for easier characterization of the electronic
and hydrothermal techniques [43]. These techniques allow for properties of the interface due to its well-defined dimensions and
more precise control of crystallite size, surface area, and stoi- contact area. It also becomes much easier to study the thermal sta-
chiometry of the resulting powders compared to using commercial bility of the interface, including diffusion across the junction and
powders. This type of synthesis route requires detailed char- possible growth of new mixed phases [47]. However, these struc-
acterization of the dispersion state in order to understand the tures may be less desirable for some applications due to lower
mechanisms at work. Many papers assume that a random dis- surface area and low gas-accessibility to the heterojunction inter-
persion of nanoparticles results, but others have found spherical face. These well-defined 2D layers can be made through several
core–shell structures, solid solutions, or presence of new phases. processing routes including, most commonly, the sol–gel method
Korotcenkov et al. detailed their best estimate at the various dis- [48–50] and sputtering [47,51], but also by more advanced tech-
persions resulting from doping In2 O3 with several other precursors niques such as electrochemical deposition with periodic potential
through the sol–gel method, as shown in Fig. 2 [38]. They concluded changes [52,53]. In horizontal 2D bi-layer structures, the analyte
that the various dopants’ effects on sensing performance are pri- gas must diffuse through the top material to reach the heterojunc-
marily a result of the different grain structures formed. Without tion. Dandeneau et al. minimized this problem by using pyrolysis
proper characterization of these structures, it is difficult to guess the to enhance and optimize the porosity of the upper CuO film [50].
degree of physical contact between each constituent, which then Cui et al. used the electrochemical deposition method to create

Fig. 3. (a) SEM image of ZnSnO3 -SnO2 nanoflakes (reprinted with permission from Ref. [20]); (b) SEM image of ZnO–Eu2 O3 nanostrips (reprinted with permission from Ref.
[43]); (c) H2 –N2 gas-etched spinodally-decomposed SnO2 –TiO2 grains [46].
254 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

Fig. 4. (a and b) TEM images of Cr2 O3 -decorated ZnO NWs; (c) high-angle annual dark field (HAADF) scanning TEM (STEM) image and EDS elemental mapping of Zn, Co, and
O (reprinted with permission from Ref. [62]); (d) SEM PdO-decorated flower-like ZnO nanostructures; (e) TEM image of PdO-decorated flower-like ZnO nanostructures, with
inset showing magnification of circled area (reprinted with permission from Ref. [61]).

many periodic interfaces and added the Au electrodes in such a Cr2 O3 nanoislands by thermal evaporation enhanced response
way as to expose the junctions directly to the atmosphere [52]. An toward trimethylamine (TMA) [5]. Fig. 4a–c provides micrographs
additional level of complexity can be introduced by making the top of the structure. Chowdhuri et al. demonstrated that in the case
layer itself a mixture of two components. Ivanovskaya et al. showed of SnO2 @CuO systems, the deposition and synthesis processes
that a [␥-Fe2 O3 –In2 O3 (1:1)]/In2 O3 film outperformed a simpler affected the performance, and also that an optimal size existed for
␥-Fe2 O3 /In2 O3 film in the detection of 100 ppm ethanol, though it the CuO nanoclusters in H2 S sensing [59]. Another unique com-
should be mentioned that the latter film had better CO sensitivity posite, ZnO@ZnS hollow dumbbell-graphene nanocomposites, was
than its tested counterparts [49]. synthesized by a two-step polymer-assisted hydrothermal method
and sulfurization treatment, and demonstrated excellent alcohol
3.3. Structures decorated with second-phase particles sensing performance as well as the ability to photocatalytically
break down methyl orange [10]. Kim et al. significantly showed
Another very common class of heterostructure consists of a that loading NiO onto hierarchical SnO2 spheres decreased humid-
host material decorated with nanoparticles of a second material. ity effects on sensing of CO due to the high affinity of NiO to
Often used in catalysis, photocatalysis, and photovoltaics, these moisture, leaving the SnO2 mostly unaffected [60]. Lou et al. used
secondary particles can act as catalysts or sensitizers and enhance hydrothermal synthesis of PdO-decorated ZnO flower-like nano-
performance [54,55]. Usually, the second phase is not suitable for structures to create a sensor with enhanced response to ethanol
comprising the host structure (due to instability, cost, synthesis and toluene which could be selected by the operating temperature
route, or conductivity), and must be applied in small, discrete (Fig. 4d and e) [61].
amounts to be most effective. The gas sensing community has Noble metal nanoparticles such as Pt, Pd, Ag, and Au have also
applied these principles with great success. A very simple method been used as second-phase particles applied to host oxides with
of synthesis involves creating the base nanostructures through great success [63–67]. The noble metal nanoparticles act as cat-
a hydrothermal treatment, and then using a solution-based pro- alysts by lowering the activation energy for the reaction, which
cess to coat the resulting powders with a metal–organic precursor improves molecular dissociation and the reaction rate [22]. In most
that then oxidizes with a proper heat treatment [12,56]. More cases, large increase in sensing response, decrease in response and
advanced processes include combined hydrothermal templating recovery times, and decrease in optimal operating temperature are
and sputtering [57], as well as thermal evaporation of the secondary observed [68,69]. However, these noble metal nanoparticles gener-
species [5,58,59]. Woo et al. were able to grow ZnO nanowires ally increase synthesis cost and have known instability issues such
directly from deposited gold electrodes via the vapor–liquid–solid as catalytic poisoning (activity reduction) from certain vapors and
(VLS) mechanism, and found that the additional decoration of tendencies to coarsen and cluster at higher operating temperatures
D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272 255

Fig. 5. Composite heterojunction electrospun nanofibers; (a) TEM image of NiO-ZnO composite fibers with NiO and ZnO both incorporated into a single fiber; (b) HRTEM
image of p–n NiO-ZnO heterojunction from fiber in red box in (a) (reprinted with permission from Ref. [75]); (c) SEM image of as-spun side-by-side p–n CuO–TiO2 composite
nanofibers; (d) TEM image of p–n CuO–TiO2 composite nanofibers, with arrows pointing to inset HRTEM images confirming crystalline lattice of CuO (top) and TiO2 (bottom)
(reprinted with permission from Ref. [78]). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

[68]. Though very promising and deserving of further study, they A special type of heterostructure consists of a 1D feature grown
will not be a focus of this review. radially from a host 1D structure. This results in a “brush-like”
nanostructure. This has been achieved both by multi-step thermal
evaporation [79–81] and by a two-step hydrothermal process [82].
3.4. Mixed 1D–1D structures
Chen et al. showed that the growth direction and orientation of the
secondary SnO2 nanorods can have an interfacial orientation rela-
A very economical way of creating 1D composites is through
tionship with the ␣-Fe2 O3 core nanorods, as shown in Fig. 6e [82].
electrospinning. Electrospinning uses a high voltage to draw a thin
This presents unique opportunities to control the crystallographic
fiber of a polymer–metal–organic solution from a needle toward
nature of the interface by using core nanorods of different crystal-
the substrate. The deposited fibers then must typically be calcined
lographic growth directions. It has been shown that the family of
over 600 ◦ C in order to remove the organics and increase crys-
exposed planes in the nanostructure can affect the gas-solid inter-
tallinity [70]. The different oxides can be deposited in sequence,
action [83]. In some cases, the growth kinetics can be manipulated
effectively forming a bi- or multi-layered film [71,72], ground
to allow control of the aspect ratio and morphology of the secondary
and mixed post-spinning to form a paste of mixed fibers, or two
features, as shown in Fig. 6a–d. A very open, porous structure allows
or more precursors can be spun in the same solution, creating
easy gas accessibility to a well-defined heterojunction interface.
true composite nanofibers [14,73–76]. A new method has also
Future work may allow the choice of crystallographic planes mak-
been demonstrated where two separate fibers can simultaneously
ing up the interface, which would allow further manipulation of
be spun side-by-side, forming a long continuous heterojunction
electronic properties.
[77,78]. Depending on the type of processing and synthesis meth-
ods, the actual heterojunction area in these composites may differ.
It is important to realize that electrospun fibers typically consist 3.5. Core–shell structures
of strings of oxide nanoparticles (Fig. 5a and b) and thus may not
transfer charge carriers as efficiently as single-crystalline nanorods Core–shell structures are among the most promising types of
or nanowires due to the higher likelihood of recombination. How- heterostructures for the future of gas sensing as well as cataly-
ever, this also means that there are more surface and defect sites sis. This morphology provides a way to maximize the interfacial
with which the analyte gas can interact, which may result in higher area between two or more materials while still minimizing the
sensitivity [70]. Electrospun fiber films are also advantageous due amount of material acting as bulk. There are several ways to
to their open porosity, which allows for very good gas accessibility. create core–shell structures, and most are relatively simple. The
Two forms of electrospun composites maximizing the heterojunc- host material structure widely varies and can be synthesized
tion area are shown in Fig. 5. by many methods including hydrothermal, thermal evaporation,
256 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

Fig. 6. (a–d) SEM images and schematics showing progression of tungsten oxide nucleation and growth on SnO2 nanowires via thermal evaporation. Scale bar 100 nm;
(reprinted with permission from Ref. [80]) (e) ␣-Fe2 O3 @SnO2 brush-like heterostructures grown via two-step hydrothermal with clear orientation relationship (reprinted
with permission from Ref. [82]); (f) ZnO nanorods grown from In2 O3 core nanowires showing both 2-fold and 6-fold orientation relationships. Scale bar 3 ␮m (reprinted with
permission from Ref. [81]); (g) Six-fold oriented ZnO nanowires growing on Ga2 O3 nanowire cores (reprinted with permission from Ref. [84]).

physical-vapor deposition (PVD), chemical-vapor deposition polycrystalline. This structure may also utilize the optimization of
(CVD), electrospinning, sol–gel methods, and others. Then, a thin the separate receptor and transducer functions using a high-surface
layer (typically <20 nm) is deposited onto most or all surfaces of area polycrystalline coating for reception and a single-crystalline
this structure via solution processes such as hydrothermal [4,66], “highway” anchored to the substrate for transduction. Alterna-
spin- or dip-coating [85,86], or advanced processes such as sputter- tively, oxide nanoparticles can be suspended in solution with a
ing (Fig. 7a) [87] and atomic-layer deposition (Fig. 7b) (ALD) [11]. secondary metal oxide precursor and subsequently dried and cal-
Solution-based processes have limited control over the amount of cined to form a spherical core–shell structure [88]. In other work,
uptake of the shell material, which then limits control of the thick- sol–gel methods are used and after heat treatment one species
ness of the outer layer. Sputtering and ALD allow for any desired tends to segregate to the surface of the nanoparticles, effectively
shell thickness to be applied simply by controlling the deposition forming a core–shell structure as well [36,89]. The main difference
time. For example, Choi et al. electrospun SnO2 nanofibers directly between core–shell and a decorated second-phase structure is the
onto a substrate and then used ALD to form a shell of ZnO, control- degree of coverage of the host material. Core–shell structures max-
lable to ∼0.45 nm per cycle (Fig. 7b) [11]. In most cases aside from imize the MI Ox –MII Ox interface, but may be disadvantageous if the
electrospun cores, the 1D core is a single crystal and the shell is MI Ox –gas interface is needed for the reaction. If both MI Ox –gas

Fig. 7. (a) HRTEM image showing a ∼2 nm NiO polycrystalline coating on the large single-crystal core SnO2 nanowire (reprinted with permission from Ref. [87]); (b) TEM–EDX
concentration line profiles of Zn, O, and Sn measured across the electrospun core–shell SnO2 @ZnO nanofiber diameter (reprinted with permission from Ref. [11]).
Table 2
Heterostructure gas sensing performance of SnO2 -based nanomaterials.
Analyte gas Composition Synthesis route Morphology Meas. temp. Response Concentration tres trec Selective Against Ref.
Ethanol 50 vol% SnO2 –ZnO Combinatorial solution Mixed multilayer 300 ◦ C 4.69a 200 ppm 72 s NA acetone, CO, H2 , NO2 , C3 H8 [48]
deposition nanoparticles
Ethanol 50 wt% SnO2 –ZnO Electrospinning SnO2 –ZnO Composite 360 ◦ C 17a, * 2500 ppm 5s 1s NA [73]
nanofibers
Ethanol 5 wt% La2 O3 –SnO2 Powder solution coating La2 O3 -coated SnO2 300 ◦ C 740a, * 1000 ppm 20 min 20 min NA [90]
nanoparticles
Ethanol ZnSnO3 @SnO2 Hydrothermal ZnSnO3 nanorods@SnO2 270 ◦ C 27.8a 50 ppm 1s 1.8 s acetone, benzene, [20]
Nanoflakes chloroform, MeOH,
formaldehyde, CO
Ethanol 1.5 mol% Fe2 O3 –SnO2 Sol–gel method Mixed nanoparticles 250 ◦ C 24i 10 ppm NA NA NA [16]
H2 (3 wt% ZnO–SnO2 )@CuO Pressed-dipped CuO-coated (ZnO–SnO2 305 ◦ C 16a 200 ppm NA NA CO [6]
powder)
H2 SnO2 @2.6 mol% ZnO PECVD-spin coating ZnO-coated SnO2 350 ◦ C 18.4b 100 ppm NA NA CO, NH3 , CH4 [85]
nanorods
H2 (0.005 mol MoO3 )–SnO2 Sol–gel method Mixed 240 ◦ C 10f 1000 ppm 5s 10 s NA [91]
H2 1 wt% Co3 O4 –SnO2 Precipitation Mixed nanoparticles 250 ◦ C 9100a 1000 ppm NA NA CO [42]
H2 S Cu2 O/SnO2 Cyclic electrochemical Thin-film horizontal RT 45f 50 ppm NA NA toluene, LPG [52]
deposition multilayers

D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272


H2 S CuO/SnO2 Sputtering-oxidation Single bi-layer 433◦ K 70J, * 16.6 ppm NA NA NA [47]
H2 S 6% CuO/SnO2 PLD Mixed nanocrystals 150 ◦ C 4300c 20 ppm 3s NA NA [45]
H2 S SnO2 @CuO Thermal SnO2 film@CuO clusters 150 ◦ C 8100a 20 ppm 12 s 366 s NA [59]
evaporation-oxidation
H2 S 1 mol% CeO2 –SnO2 Solution precipitation Mixed nanoparticles 300◦ K 3a 5 ppm 40 s 20 s LPG, EtOH, NOx , CO [37]
H2 S SnO2 @ZnO Thermal ZnO-coated SnO2 350 ◦ C 2.1a, * 500 ppm NA NA CO, CH4 [87]
evaporation-sputtering nanowires
H2 S (5 wt% ZnO–SnO2 )@3.68 wt% Ball mill + dipping CuO-coated ZnO–SnO2 150 ◦ C 60,443d 50 ppm 15 s 7 min NOx , LPG, CO2 , CH4 [86]
CuO powder
◦ d
H2 S 5 wt% ZnO–SnO2 Ball mill ZnO–SnO2 powder 250 C 250 50 ppm NA NA NA [86]
H2 S SnO2 Ball mill SnO2 powder 350 ◦ C 10d 50 ppm NA NA NA [86]
NO2 SnO2 @ZnO Electrospinning-ALD ZnO-coated SnO2 RT 0.4c 5 ppm ∼50 s* ∼450 s* NA [11]
nanofibers
NO2 40% ZnO–SnO2 Reverse microemulsion SnO2 with amorphous 250 ◦ C 34.5b 500 ppm NA NA NA [19]
ZnO coating
NO2 20% WO2 -SnO2 Sol precipitation Particles 200 ◦ C 186b 200 ppm NA NA NA [35]
NO2 (20% WO2 –SnO2 )@MgO Sol precipitation MgO Coated particles 150 ◦ C 418b 200 ppm NA NA NA [35]
NO2 (40 wt% In2 O3 –SnO2 )@10 wt% Co-precipitation Al2 O3 -Coated 200 ◦ C 632b 450 ppm NA NA CO [18]
Al2 O3 nanoparticles
NO2 40% In2 O3 –SnO2 Co-precipitation Particles 200 ◦ C 7.5b 1000 ppm NA NA NA [18]
CO 40% In2 O3 –SnO2 Co-precipitation Particles 250 ◦ C 16a 1000 ppm NA NA NA [18]
CO SnO2 @In2 O3 2-step Au-catalyzed-self- Brush-like SnO2 @In2 O3 300 ◦ C 1.9a 200 ppm 135 s 460 s NA [79]
catalytic
VLS
CO SnO2 @NiO Thermal NiO-coated SnO2 250 ◦ C 15.9a 500 ppm NA NA CH4 [87]
evaporation-sputtering nanowires
CO (0.002 mol MoO3 )–SnO2 Sol–gel method Mixed 140 ◦ C 30f 1000 ppm 5s 5s NA [91]
CO 50 wt% Co3 O4 -SnO2 Precipitation Mixed nanoparticles 100 ◦ C 175b 1000 ppm NA NA H2 [42]
CO (3 wt% ZnO–SnO2 )@CuO Pressed-dipped CuO-coated (ZnO–SnO2 235 ◦ C 15.4a 200 ppm NA NA H2 [6]
powder)
CH4 25 wt% In2 O3 –SnO2 Co-precipitation Coated nanoparticles 300 ◦ C 7.2a 850 ppm NA NA NA [9]
CH4 20 wt% TiO2 -(25 wt% Co-precipitation Coated nanoparticles 300 ◦ C 13.6a 850 ppm NA NA CO [9]
In2 O3 –SnO2 )
Ethyl-ene 0.3 wt% WO3 –SnO2 Solution precipitation Mixed nanoparticles 300 ◦ C 1.7a 6 ppm ∼10 min* ∼10 min* NA [41]
SO2 1 mol% NiO–SnO2 Pechini method NiO-coated SnO2 25 ◦ C 0.84d, * 18 ppm 4.5 min 15 min O2 , C3 H8 , NOx [89]
Cl2 W18 O49 –SnO2 Sequential thermal Brush-like SnO2 @W18 O49 RT 11.0l 6 ppm 4.6 min 17 min H2 S, CO, NOx , NH3 [80]
evaporation
Butanol 50 wt% ZnO–SnO2 Mechanical mixing Mixed particles 350 ◦ C 300g, * 5 ppm NA NA EtOH, propanol, hexanol, [17]
butanal, decane
◦ a, *
TMA 10 wt% ZnO–SnO2 Precipitation- Mixed SnO2 330 C 126 50 ppm 2s 5s NH3 , DMA, MA, EtOH, MeOH, [92]
hydrothermal nanoparticles/ZnO Acetone
nanorods
Note: a S = Ra /Rg ; b S = Rg /Ra ; c S = R/Rg ; d S = R/Ra ; e S = (R/Rg ) × 100; f S = (R/Ra ) × 100; g S = (I/Ia ) × 100; h S = Vg /Va ; i S = Gg /Ga ; j S = G/Ga ; k S = G/Gg ; l S = Ig /Ia .

257
*
Denotes a value not explicitly stated in the study, but approximated from a graphical plot.
258 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

and MII Ox –gas interfaces are needed, the decorated second-phase sensing properties is a p–n junction. N-type gas sensors are much
particles would be most advantageous. Huang et al. showed that more widely used than p-type gas sensors due to several consider-
the thickness of the ZnO shell on SnO2 nanorods was a key factor ations such as stability in low oxygen environments and response
in sensing behavior [85]. Additionally, they report that a complete compatibility to measuring devices [28]. However, these materials
ZnO shell showed an n–p–n response transition while a decoration can be combined in different ways to utilize the effects of the p–n
of ZnO nanoclusters showed only n-type behavior, which will be junction and improve sensing performance [5,13,14,26,93,102]. For
discussed in Section 4.1.4. reference, several materials with their respective dominant con-
Tables 2–4 provide a compilation of a large portion of recent duction types are presented in Table 5.
studies using nanostructured metal-oxide based heterojunctions When the Fermi energy (EF ) of a semiconducting material is dif-
for gas sensing applications. Data are sorted first by base material, ferent than that of a second material with which it has an electrical
second by gas analyte, and third by choice of secondary material. connection, the electrons at the higher energies will flow across
Table 2 focuses on SnO2 , Table 3 on ZnO, and Table 4 compiles data the interface to unoccupied lower-energy states until the Fermi
from TiO2 , In2 O3 , Fe2 O3 , MoO3 , Co3 O4 , and CdO-based nanomate- energies have equilibrated. This is equivalent to electron–hole
rials, segregated by horizontal dashed lines. The direct comparison recombination in the vicinity of a p–n junction. This is often called
of gas sensing measurements by different research groups is very “Fermi level-mediated charge transfer.” A schematic of this is
difficult due to various gas concentrations, response definition con- shown in Fig. 8a. This leads to a zone depleted of charge carriers
ventions, sensor design, film thickness, operating temperature, etc. at the interface called the depletion region. A potential energy (PE)
The best that can be drawn is from the comparisons to the native barrier develops at this interface due to the band bending which
oxides prepared in the same fashion and reported in the same study. is caused by the difference in original Fermi levels of the materi-
These are included in Tables 2–4 where relevant. In studies that als. Electrons must overcome this PE barrier in order to cross this
tested a range of compositions, the specific compositions listed for interface, as seen in Fig. 8b.
each study were the optimal composition found. Additionally, the
measurement temperature reported was usually the optimal oper- 4.1.1. Effects of p–n nanojunctions on gas sensing
ating temperature for that composition. “NA” is listed where no Consider an n-type ZnO nanowire sensor decorated with p-type
specific data were available, and asterisks denote values approxi- Co3 O4 nanoislands (similar to Fig. 4a–c) [5]. The normal ambient
mated from graphical plots. The reader is encouraged to refer to the resistance of the nanowires in air (Ra ) will be even higher than
original publication for additional details such as performance as without the heterojunction due to the depletion region at the het-
a function of composition, layer thickness, operating temperature, erojunction interface extending into the ZnO nanowire, decreasing
etc. the width of the charge conduction channel. A reduced cross-
sectional area available for charge conduction in the nanowire will
result in an increased resistance. Additionally, charge conduction
4. Mechanisms responsible for enhanced sensing
across the p–n interface will further contribute to the increase of
performance in heterostructures
the resistance. When an oxidizing gas such as NO2 is introduced,
any additional increase in the resistance due to adsorption is now
A survey of the literature reveals that a large number of het-
minimized. However, when a reducing gas is introduced, a large
erostructured MOGS research studies are strictly empirical, with
decrease in the resistance is possible due to the initial value of Ra
a trial-and-error type approach in finding suitable materials or
being exceedingly high, as response is typically measured as Ra /Rg
composites for a given application. Conclusions in the literature
[5]. This by itself should only be true if the depletion depth (there-
often state that the presence of the second constituent has a drastic
fore resistance) caused by the Co3 O4 nanoislands is more sensitive
effect on sensing performance, but do not give detailed mechanis-
to the atmosphere than the ZnO nanowires alone. The authors also
tic explanations. It is obviously expensive and time-consuming to
attribute the enhanced response toward ethanol above 360 ◦ C to
delve into this problem, and may seem unnecessary if the empir-
its total conversion to CO2 and H2 O, which has been reported as
ical results are already suited to the application. However, more
low as 380 ◦ C in the presence of a Co3 O4 catalyst [103]. These two
detailed analysis of these heterojunction mechanisms may even-
factors allowed the researchers to enhance selectivity to a reduc-
tually lead to an understanding that allows for bottom-up design
ing gas, e.g. , ethanol, by subsequently reducing the response to
of a material for a given sensing application, likely saving time
the oxidizing gas NO2 , which often causes interference with the
and money in the long term. For this reason, we highlight those
measurements due to its presence from vehicle emissions during
studies that have attempted to thoroughly explain their results,
roadway blood-alcohol (ethanol) BreathalyzerTM testing [5].
in the hope that future studies may gain more understanding
Similarly, Cr2 O3 was applied to ZnO nanowires and tested in the
from these efforts. At present, there is no one theory that can
presence of tri-methylamine (TMA) [62]. The study showed that
govern all heterostructure–gas interactions because of the diver-
nanoislands drastically enhanced response, while a complete coat-
sity of structures and materials available. One must understand
ing decreased the response compared to pristine ZnO nanowires
the fundamentals in surface chemistry and reactions, energy band
(Fig. 9). They concluded that the Cr2 O3 also narrowed the ZnO
structure, catalytic behavior, and microstructural development in
conduction channel as explained above in both cases, increasing
order to comprehend the scope of these mechanisms. Though many
the air resistance and creating a larger change when the reduc-
classes of dissimilar materials can form electronic heterojunctions,
ing gas TMA was introduced. In this case, Cr2 O3 was also chosen
the focus of this section is on oxide–oxide interfaces and mix-
because it has shown catalytic effects toward methylamine decom-
tures for gas sensing, as there is a wide range of mechanisms in
position. It was also shown that the dominant conduction path
play and no comprehensive review of the subject has before been
in the continous core–shell nanostructures was through the p-
published.
type Cr2 O3 layer, and thus showed p-type behavior on exposure
to TMA [62]. A simultaneous study [104] by Mashock et al. on SnO2
4.1. Role of the interface nanoparticle-coated CuO nanowires correlates well with the above
result. This nanostructure is of opposite nature, utilizing a p-type
The heterojunction interface is perhaps the most important con- core (CuO) and an n-type coating (SnO2 ). Still, a depletion region
sideration when analyzing the behavior of a semiconducting metal forms at the interface and impinges upon the conducting core. Since
oxide composite. A very common interface used to modulate gas p-type materials are dominated by surface conduction due to the
Table 3
Heterostructure gas sensing performance of ZnO-based nanomaterials.

Analyte gas Composition Synthesis route Morphology Meas. temp. Response Concentration tres trec Selective against Ref
◦ a
Ethanol ZnO@ZnS Polymer-assisted ZnS nanoparticle-coated 210 C 23 1000 ppm 15 s 15 s NA [10]
hydrothermal- ZnO hollow dumbbells
sulfurization
Ethanol ZnO@ZnS @Graphene Polymer-assisted Graphene-coated ZnS 210 ◦ C 38a 1000 ppm 15 s 15 s Acetone, [10]
hydrothermal- nanoparticle-coated ZnO formaldehyde,
sulfurization hollow dumbbells benzene, cylcohexane
Ethanol 2:1 mol ZnO:SnO2 Electrospinning Composite nanofibers 300 ◦ C 18a, * 100 ppm 5s 6s NH3 , C2 H2 , NOx , CH4 , [74]
H2 , CO

D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272


Ethanol 25 wt% SnO2 –ZnO Mechanical mixing Mixed particles 350 ◦ C ∼82j, * 100 ppb NA NA NA [32]
Ethanol 6%CeO2 -Al2 O3 -ZnO LDH solution Particle 260 ◦ C 170a 1000 ppm 2s 10 s MeOH, toluene, [93]
formamide
Ethanol Al2 O3 -ZnO LDH solution Particle 260 ◦ C 9a 1000 ppm 14 s 87 s NA [93]
Ethanol ZnO Electrospun Nanofibers 300 ◦ C 7a 100 ppm NA NA NA [72]
Ethanol In2 O3 /ZnO Electrospun Nanofiber bi-layer 210 ◦ C 25a 100 ppm 2s 1s NA [72]
Ethanol ZnO/In2 O3 /ZnO Electrospun Nanofiber tri-layer 210 ◦ C 17a 100 ppm NA NA NA [72]
Ethanol ZnO Electrospun Nanofibers 300 ◦ C 9a 100 ppm 1s 5s NA [14]
Ethanol 4.5 wt% Cr2 O3 -ZnO Electrospun Nanofibers 300 ◦ C 24a 100 ppm 1s 5s CH4 , H2 , CO, NO, NO2 [14]
Ethanol ZnO@Co3 O4 VLS-thermal evaporation Co3 O4 nanoisland-coated 400 ◦ C 21.9a 100 ppm 100 s* 200 s* NO2 , CO, H2 , CH4 , C2 H6 , [5]
ZnO nanowires C3 H8 , i-C4 H10
Ethanol 2:8 mol CuO:ZnO Co-precipitation Mixed powder 115 ◦ C 96h, * 100 ppm 13 s 5s NA [94]
Acetone ZnO@TiO2 2-step CVD TiO2 -coated ZnO 400 ◦ C 22i 100 ppm 1 min 1 min CO, EtOH [3]
nanoplatelet film
H2 TiO2 /NiO DC Sputtering 100 nm TiO2 over 10 nm 200 ◦ C 70a 10,000 ppm NA NA NA [51]
NiO film
◦ g
H2 ZnO@SnO2 MBE/PLD SnO2 -coated ZnO 400 C 70 500 ppm NA NA NA [95]
nanowires
H2 S ZnO Traditional hydrothermal Nanorods 100 ◦ C 25a, * 100 ppm NA NA NH3 , MeOH, EtOH, [12]
butanol, acetone, ether
H2 S ZnO@3 wt% CuO Traditional ZnO porous nanorods@CuO 100 ◦ C 39a, * 100 ppm 120 s >150 s NH3 , MeOH, EtOH, [12]
hydrothermal/solution nanoparticles butanol, acetone, ether
synthesis
H2 S 2:8 mol CuO:ZnO Co-precipitation Mixed powder 115 ◦ C 20h, * 100 ppm 13 s 5s NA [94]
NO2 5% Eu2 O3 -ZnO Traditional hydrothermal Porous strips 300 ◦ C 16b 3 ppm 3 min 3 min CO [43]
NH3 2 mol% ␣-Fe2 O3 -ZnO Hydrothermal-solution Mixed nanoparticles RT 10,000l 0.4 ppm 20 s 20 s TMA, EtOH, MeOH [96]
mixing
NH3 ZnO@Cr2 O3 Powder film dip-coating Cr2 O3 -coated ZnO powder RT 13.7j 300 ppm 25 s 75 s LPG, CO2 , EtOH, H2 , Cl2 [97]
LPG ZnO@0.47 wt% Cr2 O3 Powder film dip-coating Cr2 O3 -coated ZnO powder 350 ◦ C 46j 100 ppm 18 s 42 s NH3 , CO2 , EtOH, H2 , [7]

Note: a S = Ra /Rg ; b S = Rg /Ra ; c S = R/Rg ; d S = R/Ra ; e S = (R/Rg ) × 100; f S = (R/Ra ) × 100; g S = (I/Ia ) × 100; h S = Vg /Va ; i S = Gg /Ga ; j S = G/Ga ; k S = G/Gg ; l S = Ig /Ia .
*
Denotes a value not explicitly stated in the study, but approximated from a graphical plot.

259
260
Table 4
Heterostructure gas sensing performance of TiO2 , In2 O3 , Fe2 O3 , MoO3 , Co3 O4 , and CdO-based nanomaterials.

Analyte gas Composition Synthesis route Morphology Meas. temp. Response Concentration tres trec Selective against Ref.

H2 20 wt% SnO2 –TiO2 Co-precipitation Mixed nanoparticles 400 ◦ C 9.9c, * 20 ppm 12–14 s 4.5–5 min NA [30]
H2 10 wt% SnO2 –TiO2 Mechanical mixing Mixed nanoparticles 400 ◦ C 10.5c, * 20 ppm 12–14 s 4.5–5 min NA [30]
Ethanol 20 wt% SnO2 –TiO2 Colloidal mixing Mixed nanoparticles 553◦ K 51a, * 200 ppm 10–15 s 14–20 s NA [98]
Acetone 20 wt% SnO2 –TiO2 Colloidal mixing Mixed nanoparticles 633◦ K 55a, * 200 ppm 10–15 s 14–20 s NA [98]
O2 90% CeO2 –TiO2 Sol–gel method Mixed 420 ◦ C 9.5b 103 ppm 50 s 80 s NA [36]
Ethanol In2 O3 @ZnO 2-step VLS Core–shell In2 O3 350 ◦ C 265i 400 ppm NA NA NA [99]

D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272


nanowires@ZnO
O3 (␥-Fe2 O3 –In2 O3 (9:1))/In2 O3 Sol–gel method Mixed 135 ◦ C 8670k 100 ppb NA NA NO2 [100]
nanoparticles/nanoparticles
NO2 (␣-Fe2 O3 –In2 O3 (9:1))/In2 O3 Sol–gel method Mixed 135 ◦ C 440k 5 ppm 20 s >900 s O3 [100]
nanoparticles/nanoparticles
Ethanol ␥-Fe2 O3 /In2 O3 Sol–gel method ␥-Fe2 O3 /In2 O3 bi-layer 300 ◦ C 68j, * 100 ppm 50–60 s 50–60 s NA [49]
Ethanol ␣-Fe2 O3 @SnO2 2-Step Hydrothermal Brush-like ␣-Fe2 O3 @SnO2 350 ◦ C 4.6a 10 ppm NA NA H2 , CH4 , C4 H10 [82]
Ethanol ␣-Fe2 O3 @SnO2 @4.8 wt% Au 2-Step Au-functionalized SnO2 -coated 320 ◦ C 3.11a 100 ppm 5s 5s MeOH, H2 S [66]
hydrothermal-solution ␣-Fe2 O3 nanospindles
reduction
Ethanol ␣-Fe2 O3 @SnO2 2-Step hydrothermal SnO2 -coated ␣-Fe2 O3 320 ◦ C 2.8a, * 100 ppm 9s 9s NA [66]
nanospindles
Ethanol ␣-Fe2 O3 @ZnO Hydrothermal-solution Core–shell ␣-Fe2 O3 220 ◦ C 22.1a 500 ppm 20 s 20 s H2 , CH4 [88]
mixing nanowires@ZnO

Ethanol ␣-Fe2 O3 @ZnO 2-Step Hydrothermal Core–shell ␣-Fe2 O3 280 C 17.8 a
100 ppm ∼1 min ∼1 min MeOH, CO [4]
nanospindles@ZnO
Ethanol ␣-MoO3 @SnO2 Hydrothermal-solution SnO2 nanoparticle-coated 300 ◦ C 67.76a 500 ppm <1 min* <30 s* NA [56]
␣-MoO3 nanobelts
Ethanol 1.4 wt% TiO2 -MoO3 RF sputtering Mixed nanocrystals 400 ◦ C 250j 100 ppm NA NA NA [44]
CO 20 wt% WO3 -MoO3 RF sputtering Mixed nanocrystals 200 ◦ C 390j 15 ppm 2 min 2 min NA [44]
Ethanol MoO3 -WO3 Sol–gel method Mixed nanoparticles 300 ◦ C 10NA 100 ppm NA NA CO [39]
O3 MoO3 –TiO2 Sol–gel method Mixed nanoparticles 300 ◦ C 1.7NA 100 ppb 20 s 2 min NA [39]
Ethanol Co3 O4 Solution Strings of nanoparticles 170 ◦ C 6.2b 100 ppm NA NA NA [13]
Ethanol ZnO–Co3 O4 Solution ZnO-coated Co3 O4 170 ◦ C 46b 100 ppm NA NA NA [13]
nanoparticles
Ethanol CeO2 –CdO–Al2 O3 LDH solution Mixed nanoparticle 190 ◦ C 890a 1000 ppm 32 s 297 s NH3 , alkane mix, [101]
organics
Ethanol CdO-Al2 O3 LDH SOLUTION Mixed nanoparticle 250 ◦ C 6.4a 1000 ppm NA NA NA [101]

Note: S = Ra /Rg ; S = Rg /Ra ; S = R/Rg ; S = R/Ra ; S = (R/Rg ) × 100; S = (R/Ra ) × 100; S = (I/Ia ) × 100; S = Vg /Va ; S = Gg /Ga ; S = G/Ga ; k S = G/Gg ; l S = Ig /Ia .
a b c d e f g h i j

*
Denotes a value not explicitly stated in the study, but approximated from a graphical plot.
D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272 261

Table 5
Response type behavior of several metal oxides, semiconductors, and polymers (adapted from Ref. [28]).

Material Type of conductivity

n p n, p

Metal oxides SnO2 , ZnO, TiO2 , In2 O3 , MoO3 , MgO, Al2 O3 , NiO, CeO2 , Mn2 O3 , Co3 O4 , La2 O3 , Y2 O3 , PdO, Fe2 O3 , HgO2
Ga2 O3 , Nb2 O5 , ZrO2 , CaO, V2 O5 , Ta2 O5 , WO3 Ag2 O, Bi2 O3 , Sb2 O3 , TeO2 , CuO, Cr2 O3
Semiconductors SiC, GaN, diamond Si, InP, GaAs
Polymers Polypyrrole, polythiophene, polyaniline Trans-polyacetylene, polyphenylene

accumulation layer formed on oxygen adsorption, the effect is likely by Woo et al., and shown in Fig. 9. As Mashock et al. go on to
amplified. The researchers showed a large increase in resistance caution, this model only takes into account conduction along the
after deposition of a discrete SnO2 nanoparticle coating, as would axial direction of the nanowires, which is dominated by the avail-
be expected by the model we have presented. By doubling the able cross-sectional area for conduction, i.e. not affected by the
deposition time on a different sensor, they were able to create a depletion region. It does not take into account the changes in the
continuous nanoparticle coating forming a true core–shell struc- potential energy barrier present at the interface between particles
ture. The longer deposition time caused an even higher resistance, as it varies with the degree of oxygen adsorption. Jain et al. attribute
but also had a smaller increase in response to NH3 compared to the increase in LPG sensitivity of their SnO2 @NiO core–shell spher-
the nanowires with shorter deposition time. The authors cited two ical nanoparticles to the increased potential energy barrier height
contributing mechanisms to explain this: (1) NH3 reduces the CuO from the formation of the p–n junction [105]. Unlike 1D structures,
nanowire surface by donating an electron, which increases its resis- spherical core–shell nanoparticles force electrons to overcome sev-
tance and (2) NH3 reduces the SnO2 nanoparticles by removing eral interfacial energy barriers in series and the performance is then
adsorbed oxygen, which donates an electron and increases the car- even more dependent on the heterojunction. These two mecha-
rier concentration in the nanoparticle coating, creating a stronger nisms are described further in Section 5.4.
p–n junction. The stronger junction pushes further into the CuO
nanowire core and increases the resistance, complimenting the 4.1.2. Effects of n–n and p–p nanojunctions in gas sensing
effects of the first mechanism. When a continuous shell is created Band bending can also occur in n–n and p–p heterojunctions
by the longer deposition, the SnO2 coating becomes the primary [80,98]. Zeng et al. propose that the electron migration from TiO2 to
conductive path, and the CuO core does not participate in the reac- the SnO2 seen in Fig. 8(c and d) can help facilitate additional oxygen
tion. They propose that a discrete particle coating allows both adsorption at the SnO2 surface due to the larger electron density
mechanisms to take place, and therefore gives the best sensing [98]. Kusior et al. also proposes this mechanism in the same sys-
performance. This result is very similar to that explained above tem [8]. The interface at a p–n junction has far fewer free electrons

Fig. 8. Schematics explaining band bending at heterojunction interfaces with no adsorbed surface species. (a) p–n junction about to form between p-type Co3 O4 and n-type
ZnO. The available lower-energy valence band states (holes) stimulate electron transfer across the interface and the equilibration of Fermi energies (EF ). (b) Depletion layer
formed on both sides of p–n junction due to recombination, creating a potential barrier for electron flow. (c) n–n junction about to form between n-type SnO2 and n-type
TiO2 . The available lower-energy conduction band states stimulate electron transfer to n-SnO2 . (d) Depletion layer formed at n-TiO2 surface due to loss of electrons, and
accumulation layer formed at SnO2 surface due to added electrons, which enhances oxygen adsorption. A potential barrier was also formed at the interface. ECB : conduction
band edge energy, EVB : valence band edge energy, Egap : band gap energy, EF : Fermi energy ((c and d) adapted from Ref. [98]).
262 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

Fig. 9. TMA response (a) and resistance (b) to unmodified ZnO nanofibers, Cr2 O3 -decorated ZnO nanofibers, and ZnO–Cr2 O3 core–shell nanocables (NCs). The addition of
the p-type nanoislands caused a nearly two order of magnitude increase in air resistance and subsequently enhanced response to reducing gas TMA. The dominant charge
conduction in the core–shell nanocables was through the p-type Cr2 O3 layer, thus showing opposite response.
Reprinted with permission from Ref. [62].

due to electron–hole recombination, increasing resistance, while 4.1.3. Separation of unlike charge carriers
the interface at an n-n junction simply transfers electrons into the An additional mechanism that can be used to enhance sensing
lower-energy conduction band, forming an “accumulation layer” performance based on an interface is through charge carrier sepa-
in the SnO2 rather than a depletion layer. The accumulation layer ration. The electric field that is created across the depletion region,
can be depleted by subsequent oxygen adsorption on the surface much like in a p–n junction solar cell, acts to pull electrons in one
of the SnO2 , further increasing the potential energy barrier at the direction and holes in the opposite direction. Charge “directing”
interface, and enhancing the response. can also exist in n–n junctions through balancing of Fermi ener-
These studies show that it is important to modulate potential gies, such as in ZnO–TiO2 [3]. Charge carrier separation minimizes
energy barrier height at the heterojunction interface under ambi- electron–hole recombination and thus increases charge carrier
ent oxygen adsorption conditions. This is the key for facilitating density allowing for a more sensitive material. Zakrzewska and
the largest resistance decrease in n-type materials on introduction Radecka uniquely showed that charge carrier separation can be uti-
of reducing gases. However, for the detection of oxidizing gases it lized for enhancing H2 sensor lifetime and performance by using
has been shown to be more beneficial to artificially generate a low TiO2 added to SnO2 to photocatalytically break down organic con-
ambient resistance to allow for the largest resistance increase. Sen taminants that would otherwise degrade the sensor [106]. Here,
et al. did this using a brush-like structure with W18 O49 nanowires the working principles of photocatalysis are directly applied to the
grown on SnO2 core nanowires (Fig. 6a–d) [80]. They propose that sensor material giving it a dual functionality when illuminated with
the n-type W18 O49 donated electrons to the SnO2 core, decreasing UV light. This improved the long-term performance and stability of
the base resistance in air [80]. The potential barrier generated at the sensor.
the interface likely also increased, but this was not as much of a
factor for two reasons: the W18 O49 was not very reactive to the 4.1.4. n–p response type inversion
gas and the SnO2 core was still directly exposed to the gas and It is important to realize that the addition of a p-type material
thus not dependent upon charge conduction across the hetero- will counteract the resistance change of the n-type material for a
junction interface. The SnO2 cores were the main conduction path given gas if the structure allows it to be an independent sensing
through the material and so the injection of electrons from the material in the composite. Sometimes the p-type constituent will
W18 O49 resulted in a lower air resistance. This is another result become the dominant sensing material over a certain range of con-
that shows the importance of the architecture used to combine ditions. A few groups have reported the phenomenon of n–p or
the two types of materials, engineered to either conduct across the n–p–n response inversion behavior [51,85]. Kosc et al. reported a
interface or along the crystalline nanowires. However, one must be bi-layer TiO2 /NiO sputtered film that showed p-type conductivity
careful in interpreting these results based solely on resistance mea- naturally until a critical concentration of reducing H2 was reached
surements. The addition of a second constituent will undoubtedly [51]. The added electrons from broken oxygen bonds eventually
change the resistance just by a rule-of-mixtures such that adding fully compensated the holes in the NiO layer and then the sensor
a more conductive material will increase the conductivity. Other exhibited n-type behavior. This transition concentration was also
variables that confound these explanations include film thickness, temperature-dependent. Huang et al. reported a n–p–n response
nanostructure number density and distribution. If presenting one inversion in SnO2 @ZnO core–shell nanorods [85]. Fig. 10 shows the
of these mechanisms as an explanation, it is important to average mechanism that successfully selected H2 in the presence of other
the measurements of several “identical” samples to remove vari- reducing gases. They reported that a mixed Zn–O–Sn phase formed
ation in film depositions. A more precise measurement to remove at the heterojunction interface with intrinsic p-type behavior due
these confounding effects using single nanowires is also presented to acceptor-type doping of Zn2+ on Sn4+ sites. Hall measurements
in Section 5.4. confirmed that both the ZnO shell layer and SnO2 nanorods were
D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272 263

Fig. 10. Schematic showing the mechanism for H2 selectivity. (a) SnO2 nanorods
with continuous ZnO shell that is impermeable to gas analytes other than H2 , and
showed n–p–n transition on exposure to H2 . The key to the transition was the mixed
Sn–O–Zn p-type phase at the interface coupled with the continuous ZnO shell. (b)
Identical SnO2 nanorods except for an incomplete coating of ZnO nanoclusters. This
allowed larger gas molecules to interact directly with both n-type surfaces, and only
n-type behavior was observed.
Reprinted with permission from Ref. [85].

n-type. The ZnO shell (ca. 3 nm) was well within the Debye length
and so the depletion region extended into the n-type nanorods.
Upon exposure to low H2 concentration (<20 ppm) normal n-
type behavior was observed. Higher concentrations (20–1000 ppm)
reduced the depletion region thickness so that it only extended into Fig. 11. Conversion percentage from ethanol to various products as a function of
the p-type mixed Zn–O–Sn region, where electron injection would Fe2 O3 fraction (x = 0, pure SnO2 ) at (a) 250 ◦ C and (b) 300 ◦ C.
increase the resistance, and a p-type behavior was exhibited. H2 Reprinted with permission from Ref. [16].
concentrations above 1000 ppm then likely fully compensated the
p-type behavior by donation of electrons and creation of shallow either a dehydration or dehydrogenation reaction when catalyzed
donor states by incorporating hydrogen, and n-type behavior was by a metal oxide [16]. An acidic oxide surface favors dehydration
again observed. The other reducing gases such as CO, NH3 , and CH4 while a basic surface favors dehydrogenation. By testing a series
were not able to diffuse through the ZnO layer due to their larger of metal cation dopants, Jinkawa et al. reported that the primary
molecular size and so only n-type behavior was observed for these ethanol reaction product shifted acetaldehyde (dehydrogenation)
analytes. They comparatively studied ZnO nanoclusters not com- to C2 H4 (dehydration) with increasing electronegativity of the
pletely covering the SnO2 surface nor forming a mixed Zn–O–Sn metal cation dopant [90]. Furthermore, the dehydration product
layer, which showed only n-type behavior and was not selective C2 H4 does not increase the electron concentration in the oxide
against the larger reducing molecules [85]. upon reaction, therefore having little or no effect on the sensing
signal [107]. The complete conversion of ethanol to CO2 and H2 O
4.2. Synergistic and complimentary behavior is also more easily accomplished when dehydrogenation occurs.
This knowledge allowed Rumyantseva et al. to add basic Fe2 O3 to
Another important mechanism that should be considered in SnO2 and reduce the proportion of surface acid sites, preferentially
composite materials is non-interface dependent complimentary selecting the dehydrogenation reaction [16]. The conversion per-
behavior, often referred to as synergistic behavior. In general, this centage of these ethanol reaction products can be seen as a function
is when two different constituents in a material are each in con- of Fe2 O3 concentration in Fig. 11. Furthermore, this concept can
tact with the gas phase and each serve a different purpose that is be confirmed by referencing Tables 2–4 and noticing that an oxide
complimentary of the other. This has been reported in the litera- with a more electronegative (acidic) cation (SnO2 ) performs best for
ture most often as either decomposition of intermediate reaction ethanol sensing when an oxide with a less electronegative (basic)
products or as a spill-over effect. cation (ZnO; Fe2 O3 ) is added. The large number of ethanol sensing
studies on ZnO and Fe2 O3 is also evidence of this.
4.2.1. Complimentary decomposition Recent work in nanocomposites has also shown that having both
Many studies have shown that the successful detection of acidic and basic reaction centers with different redox properties
alcohol can be enhanced by further decomposition of an inter- in a material can more completely break down an organic vapor
mediate oxidation reaction product generated by the host oxide than one material alone [49,100]. de Lacy Costello et al. carefully
[16,17,31,90]. It has been shown that alcohols generally undergo measured the reaction products of 1-butanol catalyzed by a heated
264 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

SnO2 , ZnO or ZnO–SnO2 composite [31]. They found that SnO2 was
most effective at oxidizing butanol to butanal and ZnO was mostly
ineffective at the decomposition of butanol but readily completed
the decomposition of butanal. Thus the SnO2 –ZnO system worked
synergistically to completely break down butanol, and showed the
best sensing behavior. This and the previous work show the anal-
ogy between resistive gas sensing and catalysis. It is important for
sensor researchers to check the catalysis literature, which often
present detailed analysis of reaction pathways that also happen
at the sensor surface. A host of new catalysis and photocatalysis
literature is now also focusing on nanostructured heterojunction
materials and the possibilities they present for selecting reaction
pathways and decomposition products [108,109].

4.2.2. Catalyzed spill-over effect


The other type of complimentary behavior often mentioned in
literature is the spill-over effect. Generally, a gas molecule reacts in
a specific way with one of the composite constituents and the reac-
tion creates a secondary product that remains adsorbed onto the
surface of the other constituent and then directly affects the sensing
performance. This process is most commonly described in the use of
a CuO composite material in H2 S detection [45,59,110,111]. Specif-
ically, the H2 S reacts with CuO nanoparticles and transforms them
into CuS. The left-over hydrogen then “spills over” onto the sur-
face of the host material and acts as a reducing agent, decreasing
the resistance [111]. In this way, the CuO helps to sensitize the
host oxide to the H2 S analyte. An additional effect of this reaction
allows further performance enhancement via complete removal of
the depletion layer [45,59,110,111]. CuO forms a p–n junction with
ZnO, increasing the resistance in air, Ra . CuS is a moderate con-
ductor and thus when conversion is complete, an ohmic junction is
formed which eliminates the depletion region and allows easy con-
duction across the interface and along the nanowires (Fig. 12) [112].
Chowdhuri et al. supported this hypothesis by finding an opti-
mal thickness of the CuO cluster on a SnO2 surface for maximum
sensing response [59]. With CuO clusters thicker than 8 nm com-
plete conversion to CuS does not occur, but below 8 nm the clusters Fig. 12. (a) A significant improvement in sensor performance (higher response,
lower operating temperature) is observed when the p-type CuO nanoparticles are
do not maximize the hydrogen spill-over effect and response drops
added to the n-type SnO2 nanowires. (b) Sensing mechanism of CuO sensitization
sharply. These systems therefore make use of multiple mechanisms to H2 S gas is explained. In air, a depletion region forms at the p–n junction, reducing
simultaneously to maximize the performance. the charge conduction channel. With H2 S present, CuO is transformed to CuS and
The spill-over effect has also been described for noble metal the depletion region is removed.
nanoparticles such as Pd catalyzing the dissociation of O2 subse- Reprinted with permission from Ref. [112].
quently spilling out the O- adsorbed ions onto the metal oxide
surface [27,34]. The main advantage of using added catalysts is
to lower the activation energy needed for reaction, which reduces can react, also enhancing response. A dopant oxide or metal pre-
response and recovery time and lowers operating temperature [28]. cursor can exist in many dispersion states as shown previously
Another slight variation of this could be called the “reverse spill- in Fig. 2 [38]. Korotcenkov et al. stress that there is no univer-
over” effect, where a molecule which preferentially adsorbed onto sal law describing effects of doping on gas sensor properties, and
one oxide readily reacts with another molecule adsorbed onto the that each dopant will behave differently depending on the concen-
other oxide and the heterojunction of the two constituents facili- tration, dispersion state, and application [68]. A series of papers
tates the desorption [87,113]. by one group has shown that In2 O3 [18], ZnO [19], TiO2 [9], and
WO3 [35] can be used to effectively limit grain growth of the SnO2
4.3. Manipulation of microstructure host nanoparticles during high-temperature crystallization follow-
ing solution-based synthesis processes. They found that there was
Many studies have indicated microstructural effects as a fac- an optimal calcination temperature at which maximum crystal-
tor in sensor performance enhancement using composites. These lization of SnO2 is achieved while the dopant oxides still exists
generally are not dependent on the heterojunction interface but as amorphous coatings on the SnO2 nanoparticles, suppressing
are worth mentioning as complicating factors in interpreting data. grain growth. Lower calcination temperatures allowed higher sur-
Many dopants introduce electronic effects and simultaneously face area, but the crystallinity was insufficient for good sensor
affect the microstructure and growth kinetics [105]. Purely geo- stability [35]. Other studies also reported dopant-controlled grain
metric considerations involve grain size reduction and surface area refinement during sintering in processing techniques such as sput-
enhancement. Although grain size reduction generally leads to tering [44], colloidal solution [98] and simple powder mixing [114].
higher surface area, the mechanisms should be distinguished. A Surface-cupricated (CuO-coated) mixed ZnO–SnO2 powders also
smaller grain size allows the entire nanoparticle to exist within showed grain refinement and enhanced response [6,86]. One must
the depletion zone, enhancing sensor response. An increased sur- take care that this growth-retarding coating does not completely
face area provides more active sites with which an analyte gas shield the gas from the active base sensor material. A recent study
D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272 265

showed that Cr2 O3 -doped ZnO sensor response to LPG reached a 5.3. Characterizing the dispersion state
maximum at 0.47 wt% and decreased at higher concentrations due
to increased masking of the ZnO surface active sites [7]. As mentioned above, the dispersion state of the constituents in
the nanocomposite must be carefully characterized to investigate
5. Remaining challenges the electronic mechanisms that determine the sensor performance.
The dispersion state should also be characterized in terms of repro-
The recent works reported here have begun to uncover new and ducibility between “identical” sensors. Any inhomogeneity in a
innovative ways that nanostructured metal-oxide based hetero- “random” mixed powder or sol–gel material may change the degree
junction materials can enhance gas sensor performance. However, of heterojunction contact area and affect the average potential
these complex configurations and designs present new challenges barrier height between grains. Techniques such as elemental dot-
that must be overcome. Typical drawbacks and challenges in metal- mapping and line scans using HRTEM-EDS are very effective for
oxide gas sensors are highlighted in previously-mentioned reviews these purposes, as shown previously in Figs. 4c and 7b above and
[2,21,28], so here we will focus on those most relevant to nanostruc- here in Fig. 13. X-ray photoelectron spectroscopy (XPS) has also
tured heterojunction materials. been used in many studies to estimate the degree of electronic
interaction between different constituents, which manifests itself
as shifts in the electron binding energy [4,35]. The dispersion state
5.1. Long-term stability is also relevant in systems of decorated second-phase particles
which may coarsen and grow on the surface of the core nano-
It is well-known that metal oxides are largely chemically sta- structure. Heterojunctions with epitaxial relationships such as in
ble relative to organic and hybrid sensor materials [28]. However, brush-like morphologies should also be studied to determine the
problems with signal drift during sensor operation are widespread crystallographic relationships of the interface as this can affect
in MOGS. As researchers move toward smaller nanoparticles and the potential barrier height and degree of electron scattering. The
attempt to utilize sharp interfaces, thermal and long-term tempo- area fraction of the core material surface exposed to the atmo-
ral stability of sensor response will become even more challenging. sphere versus that which makes up the heterojunction interface
It is generally accepted that nanoparticles of smaller dimensions should also be quantified. This can be important when considering
are more likely to coarsen over time at high operating tempera- mechanisms such as potential barrier height that are dependent on
tures than their larger, more stable counterparts. When one starts the solid–solid interface, compared to synergistic mechanisms that
to consider several metal oxide species segregated in regions less need both oxide–gas interfaces readily available.
than 50 atoms wide operating at 350 ◦ C for months at a time,
it is easy to see the challenge. Any small amount of diffusion
5.4. Understanding the mechanism
across these interfaces can form intermetallic or mixed oxide
compounds which would severely affect the electronic properties
Throughout the relevant literature there are dozens of different
of the potential barrier. Several studies have in fact shown evi-
explanations for improvements in gas sensing performance with
dence of mixed phases forming at the heterojunction interface
the addition of a second constituent. The variations in materials
[47,85,115]. The formation of these mixed phases, if undesired,
synthesis, processing methods, testing methods, and characteriza-
can be mitigated by using low-temperature synthesis techniques
tion make it nearly impossible to draw fundamentally supported
and lowering the operating temperature, in order to limit diffusion
conclusions using comparisons between works from more than one
and/or reactions [21]. Fortunately, many of these heterojunction
institution. The beginnings of a real fundamental understanding of
architectures have also lowered optimum operating tempera-
these materials must start with the simplest possible experiments,
tures. Higher stability can also be achieved by using 1D features
minimizing or elimating complicating or confounding factors. To
such as nanorods and nanowires which generally have a higher
this end, the authors propose a new focus on single-nanowire elec-
degree of crystallinity and larger planar facets than spherical
trical and sensing measurements.
nanoparticles [2,22].
There are two main competing mechanisms that must be
considered when designing heterostructures via resistance manip-
5.2. Reproducibility ulation. The first is the change in the potential energy barrier
encountered by an electron crossing a heterojunction interface.
Reproducibility in performance from one sensor to another Many papers propose that using p–n oxide combinations make this
under identical processing conditions is not often reported but is energy barrier more sensitive to the given analyte gas, thus making
essential when considering scaling up to large-scale production of the sensor more responsive [7,97,105]. The measured resistance is
commercial devices. This becomes exceedingly difficult when deal- related to the interfacial potential energy barrier (V) by [56]:
ing with more than one constituent. Fabrication methods must be  qV 
able to effectively control the thickness and diameter of each layer R = R0 exp (1)
kT
or crystalline particle to within a few nanometers. As stated above,
the Debye length of these materials is generally of the same order where R is the resistance of the material, R0 is the initial resistance, q
of magnitude as the feature radius or thickness. Therefore, if the is the charge of an electron, V is the potential energy barrier height,
shell of a core–shell material is occasionally made a few nanometers k is Boltzmann’s constant, and T is the temperature of the sensing
larger than the Debye length, the sensor response may be signifi- material.
cantly degraded due to the appearance of a bulk region. The highest The second mechanism is the narrowing of the charge conduc-
degree of control during fabrication is given by plasma deposition tion channel along the length of a nanowire via the n–p depletion
processes such as sputtering, PLD and ALD. However, much cheaper region impinging on the core. Many other papers propose that this
and higher-throughput methods such as spin- or dip-coating may is the main mechanism for the improved performance, as the oxide
be good enough for the application depending on the sensitivity nanoparticle coating makes the resulting depletion region bound-
toward the Debye length. New ways of carefully controlling the ary more sensitive than that of the depletion region created directly
particle size and coating thickness in high-throughput processes on the core by the adsorption and desorption of oxygen [14,62,104].
will be essential in the development of commercial nanostructured Here, the measured resistance would depend mainly on the diam-
heterojunction sensors. eter of the non-depleted core of the nanowires, which we may call
266 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

Fig. 13. HRTEM–EDS elemental mapping of (a–d) 0.02 CuO–ZnO and (e–h) 0.04 CuO–ZnO hollow spheres.
Reprinted with permission from Ref. [110].

the conduction channel diameter. Not including changes in resis- voltage at a perfect SnO2 –CuO interface is approximately 1.3 eV
tivity due to charge carrier concentration, this can be approximated for electrons moving from SnO2 into CuO. This built-in voltage
by the equation for resistance in a wire: increases the resistance in the material and cannot be decreased
except by a forward bias voltage, which is not easily applied to
L
R= (2) a mixture of these phases. In air, as shown, oxygen adsorption
Dcond
at the CuO surface will increase the hole concentration due to
where R is the resistance of the sensor,  is the resistivity of the removal of electrons for bonding, forming an accumulation layer
material, L is the length of the nanowire, and Dcond is the diameter where the CuO becomes more heavily p-type. This should be the
of the non-depleted charge conduction channel along the core of only cause of an overall larger potential energy barrier at this
the nanowire. interface since a 0.65 eV increase in the SnO2 is only half of the
This raises the question: “Which mechanism is most useful built-in voltage already established. The increased barrier in air
for a gas sensor?” Consider the possibility of aligning a mat of on CuO (Vair in Fig. 14) may be more or less than 0.65 eV as no
oxide nanowires either parallel or perpendicular to the current flow equivalent studies to Kar et al. have been found for other types of
direction. These architectures each allow the conduction channel nanowires in oxygen atmospheres, so these numbers are only sim-
mechanism and interface potential energy barrier mechanism to ple estimates. If CuO’s surface states cause 0.65 eV of upward band
dominate, respectively. A thick ZnO nanowire mat aligned roughly bending, it must first overcome approximately 0.3 eV of down-
parallel to the current has been shown to have much lower resis- ward bending from the depletion region, leaving resulting potential
tance than the same mat measured perpendicular to the alignment, energy barrier only capable of 0.35 eV of modulation. If adsorption
as should be expected [116]. Unfortunately, no gas sensor measure- of oxygen is much larger than SnO2 , the resulting energy barrier
ments were performed. A more precise experiment was carried capable of modulation may be significantly larger than that of a
out by Khan et al. [117], which found that the “. . .potential bar- SnO2 –SnO2 interface. Iwamoto et al. found that p-type metal oxides
rier modulation of multiple nanowires was more efficient than the had higher oxygen desorption than n-type oxides as measured by
modulation of the surface depletion of the single ZnO nanowire temperature-programmed desorption (TPD) [119]. However, this is
in gas sensing [to H2 ].” No similar studies involving heterostruc- only measuring desorption and does not necessarily find the total
tures have yet been performed, to the authors’ knowledge. Fig. 14 amount of adsorbed surface oxygen. Although this should make p-
illustrates these two different mechanisms as they occur in pris- type sensors more responsive, the surface-dominated conduction
tine SnO2 nanowires and CuO nanoparticle-coated SnO2 nanowires. paths in p-type oxides are a hindrance and generally limit the per-
The potential energy barrier modification occurs at the interfaces formance [26]. More work is needed in this area to develop a more
and is shown by the band diagrams in air and in a reducing atmo- robust model for interactions between these materials and analyte
sphere. The charge conduction channel constriction occurs along gases.
the length of the nanowires and is shown at the bottom of Fig. 14 As discussed above, Na et al. [5] propose that the large built-in
where the depletion region is represented by the lighter color faded voltage developed at the p–n interface generates a higher resis-
regions. tance in air and leads to a larger response for reducing gases by
Kar et al. [118] recently used ultrafast pump probe spectroscopy the response convention Ra /Rg but also citing a possible catalytic
and XPS valence band photoemission spectroscopy to show that mechanism introduced by the p-type constituent. Several others
the potential energy barrier at the surface of a SnO2 nanowire in including Mashock et al. [104] cite the mechanism of using the
air was approximately 0.65 eV due to oxygen adsorption. They fur- nanoparticle coating to enhance the constriction of the charge con-
ther determined that the depletion region accounted for a major ducting channel in the nanowire core. Also shown in Fig. 14, p-type
fraction of the cross-sectional area of a 200 nm diameter nanowire. nanoparticles on an n-type core compliment the changes in the
The potential energy barrier at the interface will then be decreased depletion region of the core induced by oxygen adsorption. Oxy-
in the presence of a reducing gas that removes adsorbed oxygen. gen creates an accumulation region in the p-type nanoparticles
Alternatively, one can see that the potential energy barrier change at which makes it more heavily p-type and makes a stronger deple-
the interface of a p–n junction such as CuO–SnO2 should be smaller tion effect at the interface, driving the depletion region further into
than 0.65 eV using very simplified calculations. The natural built-in the core nanowire in conjunction with oxygen adsorbed on the
D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272 267

Fig. 14. Mechanisms responsible for the resistance changes in nanowire-based gas sensors. The left side shows pristine n-type SnO2 nanowires with the interfacial mechanisms
depicted at top and the conduction channel mechanism depicted at bottom. The right side similarly shows these mechanisms in n-type SnO2 nanowires coated with discrete
p-type CuO nanoparticles. ECB is the conduction band minimum, EF is the Fermi level, EVB is the valence band maximum, Vbi is the built-in voltage caused by band misalignment,
Vair is the increase in the potential energy barrier on oxygen exposure, Vred is the potential barrier in reducing atmosphere, LD is the Debye length, or depth of the depletion
region from the surface, and Dcond is the diameter of the non-depleted region available for charge conduction through the core. (Band gaps and Fermi level positions drawn
approximately to scale.)

exposed regions of the n-type core. Heterostructures using n-type additional interfaces can be quantified by measuring across two or
coated n-type nanowires can also be considered in a similar man- more nanowires. These measurements should then be performed
ner. Both of these mechanisms should be considered in any study in various atmospheres varying the concentration of oxidizing and
utilizing a randomly-oriented mixture of coated nanowires. Addi- reducing gases to probe changes in depletion layer depth of the
tionally, these resistance-based mechanisms will not always agree various heterostructures. These experiments should begin to form
with experimental results and other factors such as increased sur- a mechanistic model for the role of different heterojunction inter-
face area, increased defect sites, and catalytic activity toward the faces and structures in gas sensing applications.
analyte gas should be considered. To the authors’ knowledge, no study has actually shown that
A single-nanowire measurement would be much the same as the point contact between single p- and n-type oxide nanoparticles
parallel nanowires except that it would completely eliminate any or nanowires behaves like p–n junctions, in the way that they are
oxide–oxide interfaces from the measurement. The only interface present in gas sensing films. In a closely related study, Park et al.
that would contribute to the careful resistance measurement is that found that multiple ZnO–CuO nanowire interfaces measured in par-
of the metallic contacts, which can be independently measured and allel across two electrodes showed asymmetric I–V character while
subtracted from the total measurement [120]. Additionally, a mea- the ZnO–ZnO and CuO–CuO nanowire interfaces had symmetric
surement could also be done where the current must move axially I–V character. This shows that at least some of the expected diode-
through each nanowire but must overcome one or more countable like behavior in retained in these low-quality or low-contact area
interfacial junctions in its path. This would allow one to find the junctions between nanowires. It is very possible that the bulk semi-
added effect of each interface on the sensor performance compared conductor heterojunction band models do not accurately describe
to the single nanowire. This experiment should then be expanded the behavior of the oxide interfaces in a mixed nanocrystalline pow-
to nanowires coated with p-type or n-type nanoparticles in both der film. The authors are currently undertaking much of the work
discrete and continuous fashion. A single-nanowire measurement described above, but would like to stress that a large effort across
would quantify the resistance change induced by the depletion many institutions will be needed in order for these experiments to
region at the p–n interface impinging upon the nanowire core with- develop a comprehensive model that will enable bottom-up design
out the current ever actually traversing that interface. The effects of of heterojunction-based sensors.
268 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

5.5. Useful analogues to catalysis sensor electrodes in a single device [145]. As the deposition region
is approximately 100 ␮m × 100 ␮m, only certain synthesis tech-
It is important to realize that the proposed experiments above niques that allow precise deposition can be integrated with such
are not likely to explain other cited effects such as synergistic a device. For this reason, any heterojunction nanocomposite in
mechanisms or targeted catalytic behavior. Using new materials powder form will not be easily integrated through screen-printing
as gas sensors will also make it difficult to study the mechanisms of thick films. Oxide nanoparticles can, however, be precisely
at work, as the electronic properties and surface reactions of these deposited as thin films by methods just as ink-jet printing, but any
materials are far less-studied. However, much photocatalysis and agglomerates larger than about 5 ␮m may not pass easily through
catalysis work related to the reaction mechanisms of gases and the print head. CVD, PLD, and sputtering are all compatible with
liquids at metal oxide surfaces will be helpful for understanding proper masking of the neighboring substrates.
the behavior of these sensors. An important point that warrants
inclusion here is the analogues between resistive gas sensing and 6. Future outlook
photocatalytic/catalytic materials. Many unique nanostructured
heterojunction materials have been developed specifically for pho- Recent work has greatly expanded our understanding of the role
tocatalytic and catalytic applications using constituents that are of nanostructured heterojunctions as gas sensing materials. Going
typically found in gas sensors such as SnO2 , ZnO, and TiO2 [109]. forward, new materials and heterostructure designs will require
Chalcogenides and halides specifically have been used as addi- further study of mechanisms and influencing factors on gas sensing
tives in catalysis and photocatalysis extensively with great success performance. Future directions in nanostructured heterojunction
[54,121–124]. Incorporating these materials into nanostructured gas sensors should have a dual focus on (1) new materials and new
heterojunction gas sensors has also been shown to improve gas- heterojunction interfaces, and (2) the mechanisms that govern the
sensing performance [125,126]. Many of these heterojunctions sensing performance.
are used for separation of photogenerated charge carriers in dye- The present review focuses mostly on combinations of binary
sensitized solar cells or photocatalysis to increase charge carrier compounds, but heterojunction interfaces are by no means limited
lifetime and current extraction [127,128]. Others separate these to these. Rare-earth perovskites [(RE)CoO3 , RE = La, Nd, Eu, Sm, Gd,
electrons/holes and funnel them into a material with a lower/higher etc.] have shown promise as stable p-type sensing materials with
conduction/valence band energy where they can more readily per- high activity toward the oxidation of CO and reduction of NO [28].
form the desired reduction/oxidation of the gas or liquid compound Nanostructured spinel ferrites have also shown good gas sensing
[129]. Synthesis of catalytic materials generally follows the same performance in the cases of Ce-doped CuFe2 O4 detecting LPG
goals of high surface area, small crystallite size, high porosity, long- [146], ZnFe2 O4 detecting formaldehyde [147], and NiFe2 O4 detec-
term stability, rate of reaction, and ease of synthesis that is directly ting chlorine [148]. Qi et al. recently showed that p-La0.7 Sr0.3 FeO3
applicable to gas sensing materials. Therefore, there is much to be nanoparticle-coated SnO2 nanofibers had a much higher sensing
gained in sensor research by following relevant literature in pho- response to ethanol than either material on its own [149]. Further
tocatalytic and catalytic materials, and vice versa. research should also not be limited to just oxides.
To facilitate this discussion, several excellent reviews using Organic and hybrid organic/inorganic sensors have also shown
oxides for photocatalysis and catalysis are listed here: recent great promise as gas sensors [150–152]. These generally can have
advances in heterogeneous photocatalysis [130], hierarchical better selectivity than metal oxides but often worse thermal and
oxide-based composite nanostructures [109], charge separation temporal stability of sensing performance [28]. The selectivity can
and transport in nanostructures [131], fundamentals of catalysis be enhanced by introducing a “lock-and-key” sensing mechanism,
in ceramics [132,133], and synthesis methods for nanoparticles in with little or no chance of false positive readings [153]. This is done
catalysis including microemulsion [134] and self-assembly [135]. by attaching functional group to the polymer or carbon nanotube
Examples of studies applicable to gas sensing include SnO2 -based surface with which only a single type of gas analyte will react and
[136,137], ZnO-based [138–141], and TiO2 -based [128,142–144] cause a measurable change. The drawbacks of these organic materi-
heterojunction materials. The catalysis literature usually presents als may be in part mitigated by using a metal oxide host structure to
very good analysis of measured reaction products and reaction create an organic/inorganic hybrid. This has been shown for exam-
pathways, which can further aid understanding of gas sensing ple in the case of an n–p TiO2 @polyaniline core–shell structure that
mechanisms in analogous oxide systems. had higher sensitivity than the native polyaniline nanofibers, with
detection capabilities to ammonia as low as 25 ppb [154].
5.6. Integration of nanostructured heterojunction materials into The architectures in which these materials are combined can
silicon microcircuit technology be just as important as the choice of materials themselves. This
was shown above, for example, by Huang et al. [85]. Core–shell
One of the less-considered challenges for the future of structures are very promising due to the maximization of interfacial
nanostructured heterojunction materials is integration with heterojunction area, which makes the electronic interactions of the
advanced silicon-based microelectronic devices [70]. Complex interface most dominant. The large surge in recent 1D nanostruc-
nano-heterostructures often have precise multi-step synthesis pro- ture research [2,22,155,156] has developed an excellent platform
cesses that can limit compatibility with architectures desired in for future researchers to create 1D core–shell structures simply
multi-purpose microelectronic devices. For example, a device may by applying deposition processes such as spin-coating, sputtering
serve several diagnostic purposes besides gas sensing, and the and ALD to existing methods. These processes could even allow
sensor platform will need to be integrated into a small section for multiple dissimilar shell layers of controlled thickness that
of the circuitry. In these applications, it is impractical to use a may give enhanced sensing responses. Several unique morpholo-
1 cm × 1 cm alumina platform with a several-␮m thick particulate gies known as hierarchical structures have also been synthesized
film that must be continually heated to 350 ◦ C. It is also increas- through self-assembly and vapor phase growth processes [24].
ingly desirable to use silicon substrates so both the sensing and Hierarchical structures are essentially anything made from sepa-
signal processing can occur on a single patterned substrate. Micro- rate “building blocks” of their own defined shape in a sequential
electromechanical systems (MEMS) microhotplate platforms have manner. Self-assembly of 0D, 1D, and 2D building blocks can give
been developed by the National Institute of Standards and Tech- unique shapes and morphologies while resisting agglomeration
nology (NIST) that have an array of up to 36 or more heated and densification that would result if the building blocks were
D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272 269

just deposited and dried [24]. Vapor phase growth of hierarchi- [10] X. Yu, G. Zhang, H. Cao, X. An, Y. Wang, Z. Shu, et al., ZnO@ZnS hollow
cal materials is most often used to grow 1D–1D type brush-like dumbbells–graphene composites as high-performance photocatalysts and
alcohol sensors, New J. Chem. 36 (2012) 2593.
structures, and lends itself more easily to heterojunction formation [11] S.-W. Choi, J.Y. Park, S.S. Kim, Synthesis of SnO2 –ZnO core–shell nanofibers via
[81]. These morphologies can include hierarchical nano-brushes, - a novel two-step process and their gas sensing properties, Nanotechnology
urchins, -flowers, -combs, and -dendrites [24]. Hollow nanospheres 20 (2009) 465603.
[12] L. Wang, Y. Kang, Y. Wang, B. Zhu, S. Zhang, W. Huang, et al., CuO nanoparticle
(Fig. 13) and nanoflowers have also recently shown much promise decorated ZnO nanorod sensor for low-temperature H2 S detection, Mater. Sci.
due to their high porosity, gas-permeable walls, and wall thick- Eng. C 32 (2012) 2079–2085.
nesses of less than two Debye lengths [26,110,136,157,158]. New [13] Y. Liu, G. Zhu, J. Chen, H. Xu, X. Shen, A. Yuan, Co3 O4 /ZnO nanocomposites for
gas-sensing applications, Appl. Surf. Sci. 265 (2013) 379–384.
synthesis routes still need to be developed in order to self-assemble
[14] W. Wang, Z. Li, W. Zheng, H. Huang, C. Wang, J. Sun, Cr2 O3 -sensitized ZnO
these types of structures using building blocks of different oxides. electrospun nanofibers based ethanol detectors, Sens. Actuators B: Chem. 143
The mechanisms involved in vapor phase growth are also not well (2010) 754–758.
[15] H. Gu, Z. Wang, Y. Hu, Hydrogen gas sensors based on semiconductor oxide
understood in some oxides, and warrant further investigation.
nanostructures, Sensors 12 (2012) 5517–5550.
As new processing and fabrication techniques become available [16] M. Rumyantseva, V. Kovalenko, a. Gaskov, E. Makshina, V. Yuschenko, I.
and allow for design of complex structures, it becomes exceed- Ivanova, et al., Nanocomposites SnO2 /Fe2 O3 : sensor and catalytic properties,
ingly essential to understand the fundamental sensing mechanisms Sens. Actuators B: Chem. 118 (2006) 208–214.
[17] B.P.J. de Lacy Costello, R.J. Ewen, N.M. Ratcliffe, P.S. Sivanand, Thick film
at work in order to properly choose the morphology and het- organic vapour sensors based on binary mixtures of metal oxides, Sens. Actu-
erojunction architecture for a given sensing application. There ators B: Chem. 92 (2003) 159–166.
are now simply too many possibilities to use a purely empiri- [18] A. Chen, X. Huang, Z. Tong, S. Bai, R. Luo, C.C. Liu, Preparation, characteriza-
tion and gas-sensing properties of SnO2 –In2 O3 nanocomposite oxides, Sens.
cal problem-solving approach. The authors stress a new focus on Actuators B: Chem. 115 (2006) 316–321.
single-nanowire heterostructure measurements to reduce variabil- [19] C. Liangyuan, B. Shouli, Z. Guojun, L. Dianqing, C. Aifan, C.C. Liu, Synthesis
ity and confounding factors in mechanistically explaining observed of ZnO–SnO2 nanocomposites by microemulsion and sensing properties for
NO2 , Sens. Actuators B: Chem. 134 (2008) 360–366.
sensing behavior. Many of these mechanisms in nanostructured [20] Y. Zeng, Y. Bing, C. Liu, W. Zheng, G. Zou, Self-assembly of hierarchical
heterojunctions are simultaneously being investigated by the catal- ZnSnO3 –SnO2 nanoflakes and their gas sensing properties, Trans. Nonferrous
ysis community, and interdisciplinary work should be encouraged Met. Soc. China 22 (2012) 2451–2458.
[21] G. Eranna, B.C. Joshi, D.P. Runthala, R.P. Gupta, Oxide materials for develop-
to this end. Several ways of enhancing the response, sensitivity and
ment of integrated gas sensors—a comprehensive review, Crit. Rev. Solid State
selectivity in nanostructured metal-oxide based heterojunction Mater. Sci 29 (2004) 111–188.
materials were elucidated above. Those most prominent include [22] M.M. Arafat, B. Dinan, S.A. Akbar, A.S.M.A. Haseeb, Gas sensors based on
one dimensional nanostructured metal-oxides: a review, Sensors 12 (2012)
engineering of the potential energy barrier at the interface by
7207–7258.
using a second constituent, introduction of selective p–n and n–p–n [23] Y.-F. Sun, S.-B. Liu, F.-L. Meng, J.-Y. Liu, Z. Jin, L.-T. Kong, et al., Metal oxide
response inversions, selective synergistic surface reactions and nanostructures and their gas sensing properties: a review, Sensors (Basel) 12
grain refinement. This is by no means an exhaustive list, but serves (2012) 2610–2631.
[24] J.-H. Lee, Gas sensors using hierarchical and hollow oxide nanostructures:
to report those most cited as well as a few unique explanations that overview, Sens. Actuators B: Chem. 140 (2009) 319–336.
hold promise for future development. These are likely just a fraction [25] E. Comini, Metal oxide nano-crystals for gas sensing, Anal. Chim. Acta 568
of the possibilities and it is the hope of the authors that this review (2006) 28–40.
[26] H.-J. Kim, J.-H. Lee, Highly sensitive and selective gas sensors using p-
will stimulate the reader to additional discoveries in the future. type oxide semiconductors: overview, Sens. Actuators B: Chem. 192 (2014)
607–627.
[27] C. Wang, L. Yin, L. Zhang, D. Xiang, R. Gao, Metal oxide gas sensors: sensitivity
Acknowledgements
and influencing factors, Sensors (Basel) 10 (2010) 2088–2106.
[28] G. Korotcenkov, Metal oxides for solid-state gas sensors: What determines
Throughout the writing of this manuscript, the first author was our choice? Mater. Sci. Eng. B 139 (2007) 1–23.
supported by a NASA Space Technology Research Fellowship. The [29] N. Yamazoe, G. Sakai, K. Shimanoe, Oxide semiconductor gas sensors, Catal.
Surv. Asia 7 (2003) 63–75.
authors would also like to acknowledge helpful discussions with Dr. [30] D. Shaposhnik, R. Pavelko, E. Llobet, F. Gispert-Guirado, X. Vilanova, Hydro-
Roberto Myers and Dr. Suliman Dregia of The Ohio State University. gen sensors on the basis of SnO2 –TiO2 systems, Proc. Eng. 25 (2011)
1133–1136.
[31] B.P. de Lacy Costello, R. Ewen, P.R. Jones, N. Ratcliffe, R.K. Wat, A study of
References the catalytic and vapour-sensing properties of zinc oxide and tin dioxide in
relation to 1-butanol and dimethyldisulphide, Sens. Actuators B: Chem. 61
[1] K.D. Benkstein, C.J. Martinez, G. Li, D.C. Meier, C.B. Montgomery, S. Semancik, (1999) 199–207.
Integration of nanostructured materials with MEMS microhotplate plat- [32] B.P. de Lacy Costello, R. Ewen, N. Guernion, N. Ratcliffe, Highly sensitive mixed
forms to enhance chemical sensor performance, J. Nanoparticle Res. 8 (2006) oxide sensors for the detection of ethanol, Sens. Actuators B: Chem. 87 (2002)
809–822. 207–210.
[2] A. Kolmakov, M. Moskovits, Chemical sensing and catalysis by one- [33] B. Lyson-Sypien, a. Czapla, M. Lubecka, E. Kusior, K. Zakrzewska, M. Radecka,
dimensional metal-oxide nanostructures, Annu. Rev. Mater. Res. 34 (2004) et al., Gas sensing properties of TiO2 –SnO2 nanomaterials, Sens. Actuators B:
151–180. Chem. (2013) 1–10.
[3] D. Barreca, E. Comini, A.P. Ferrucci, A. Gasparotto, C. Maccato, C. Maragno, [34] M.L. Zhang, J.P. Song, Z.H. Yuan, C. Zheng, Response improvement for
et al., First example of ZnO–TiO2 nanocomposites by chemical vapor depo- In2 O3 –TiO2 thick film gas sensors, Curr. Appl. Phys. 12 (2012) 678–683.
sition: structure, morphology, composition, and gas sensing performances, [35] S. Bai, D. Li, D. Han, R. Luo, A. Chen, C.L. Chung, Preparation, characterization
Chem. Mater. 19 (2007) 5642–5649. of WO3 –SnO2 nanocomposites and their sensing properties for NO2 , Sens.
[4] J. Zhang, X. Liu, L. Wang, T. Yang, X. Guo, S. Wu, et al., Synthesis and gas Actuators B: Chem. 150 (2010) 749–755.
sensing properties of ␣-Fe2 O3 @ZnO core–shell nanospindles, Nanotechnol- [36] A. Trinichi, Y.X. Li, W. Wlodarski, S. Kaciulis, L. Pandolfi, S. Viticoli, et al., Inves-
ogy 22 (2011) 185501. tigation of sol–gel prepared CeO2 –TiO2 thin films for oxygen gas sensing, Sens.
[5] C.W. Na, H.-S. Woo, I.-D. Kim, J.-H. Lee, Selective detection of NO2 and C2 H5 OH Actuators B: Chem. 95 (2003) 145–150.
using a Co3 O4 -decorated ZnO nanowire network sensor, Chem. Commun. 47 [37] G. Fang, Z. Liu, C. Liu, K. Yao, Room temperature H2 S sensing properties and
(2011) 5148–5150. mechanism of CeO2 –SnO2 sol–gel thin films, Sens. Actuators B: Chem. 66
[6] W.J. Moon, J.H. Yu, G.M. Choi, The CO and H2 gas selectivity of CuO-doped (2000) 46–48.
SnO2 –ZnO composite gas sensor, Sens. Actuators B 87 (2002) 464–470. [38] G. Korotcenkov, I. Boris, a. Cornet, J. Rodriguez, a. Cirera, V. Golovanov, et al.,
[7] D.R. Patil, L.a Patil, Cr2 O3 -modified ZnO thick film resistors as LPG sensors, The influence of additives on gas sensing and structural properties of In2O3-
Talanta 77 (2009) 1409–1414. based ceramics, Sens. Actuators B: Chem. 120 (2007) 657–664.
[8] A. Kusior, M. Radecka, M. Rekas, M. Lubecka, K. Zakrzewska, A. Reszka, et al., [39] K. Galatsis, Y.X. Li, W. Wlodarski, E. Comini, G. Sberveglieri, Comparison of sin-
Sensitization of gas sensing properties in TiO2 /SnO2 nanocomposites, Proc. gle and binary oxide MoO3 , TiO2 and WO3 sol–gel gas sensors, Sens. Actuators
Eng. 47 (2012) 1073–1076. B: Chem. 83 (2002) 276–280.
[9] A. Chen, S. Bai, B. Shi, Z. Liu, D. Li, C.C. Liu, Methane gas-sensing and catalytic [40] A. Gurlo, N. Barsan, M. Ivanovskaya, U. Weimar, W. Gopel, In2 O3 and
oxidation activity of SnO2 –In2 O3 nanocomposites incorporating TiO2 , Sens. MoO3 –In2 O3 thin film semiconductor sensors: interaction with NO2 and O3,
Actuators B: Chem. 135 (2008) 7–12. Sens. Actuators B: Chem. 47 (1998) 92–99.
270 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

[41] Y. Pimtong-Ngam, S. Jiemsirilers, S. Supothina, Preparation of tungsten nanoscale metal oxide semiconductors for gas sensing, Sensors (Basel) 9
oxide–tin oxide nanocomposites and their ethylene sensing characteristics, (2009) 7866–7902.
Sens. Actuators A: Phys. 139 (2007) 7–11. [71] X.-J. Yue, T.-S. Hong, X. Xu, Z. Li, High-performance humidity sensors based
[42] U.-S. Choi, G. Sakai, K. Shimanoe, N. Yamazoe, Sensing properties of Au-loaded on double-layer ZnO–TiO2 nanofibers via electrospinning, Chin. Phys. Lett. 28
SnO2 -Co3 O4 composites to CO and H2 , Sens. Actuators B: Chem. 107 (2005) (2011) 090701.
397–401. [72] X.-J. Zhang, G.-J. Qiao, High performance ethanol sensing films fabricated from
[43] S. Somacescu, A. Dinescu, A. Stanoiu, C.E. Simion, J.M. Calderon Moreno, ZnO and In2 O3 nanofibers with a double-layer structure, Appl. Surf. Sci. 258
Hydrothermal synthesis of ZnO–Eu2 O3 binary oxide with straight strips mor- (2012) 6643–6647.
phology and sensitivity to NO2 gas, Mater. Lett. 89 (2012) 219–222. [73] H.A. Khorami, M. Keyanpour-Rad, M.R. Vaezi, Synthesis of SnO2 /ZnO com-
[44] E. Comini, M. Ferroni, V. Guidi, G. Faglia, G. Martinelli, G. Sberveglieri, posite nanofibers by electrospinning method and study of its ethanol sensing
Nanostructured mixed oxides compounds for gas sensing applications, Sens. properties, Appl. Surf. Sci. 257 (2011) 7988–7992.
Actuators B: Chem. 84 (2002) 26–32. [74] X. Song, Y. Liu, Characterization of electrospun ZnO–SnO2 nanofibers for
[45] M.K. Verma, V. Gupta, SnO2 –CuO nanocomposite thin film sensor for fast ethanol sensor, Sens. Actuators A: Phys. 154 (2009) 175–179.
detection of H2 S gas, J. Exp. Nanosci. 8 (2013) 326–331. [75] Z. Zhang, C. Shao, X. Li, L. Zhang, Electrospun nanofibers of ZnO–SnO2 het-
[46] C.M. Carney, S. Yoo, S.A. Akbar, TiO2 –SnO2 nanostructures and their H2 erojunction with high photocatalytic activity, J. Phys. Chem. C 114 (2010)
sensing behavior, Sens. Actuators B: Chem. 108 (2005) 29–33. 7920–7925.
[47] R. Vasiliev, M. Rumyantseva, Effect of interdiffusion on electrical and [76] Z. Zhang, C. Shao, X. Li, C. Wang, M. Zhang, Y. Liu, Electrospun nanofibers of p-
gas sensor properties of CuO/SnO2 heterostructure, Mater. Sci. 57 (1999) type NiO/n-type ZnO heterojunctions with enhanced photocatalytic activity,
241–246. ACS Appl. Mater. Interfaces 2 (2010) 2915–2923.
[48] K.-W. Kim, P.-S. Cho, S.-J. Kim, J.-H. Lee, C.-Y. Kang, J.-S. Kim, et al., The selec- [77] Z. Liu, D.D. Sun, P. Guo, J.O. Leckie, An efficient bicomponent TiO2 /SnO2
tive detection of C2 H5 OH using SnO2 –ZnO thin film gas sensors prepared nanofiber photocatalyst fabricated by electrospinning with a side-by-side
by combinatorial solution deposition, Sens. Actuators B: Chem. 123 (2007) dual spinneret method, Nano Lett. 7 (2007) 1081–1085.
318–324. [78] C. Zhu, W. Deng, J. Pan, B. Lu, J. Zhang, Q. Su, et al., Structure effect
[49] M. Ivanovskaya, D. Kotsikau, G. Faglia, P. Nelli, Influence of chemical compo- of dual-spinneret on the preparation of electrospun composite nanofibers
sition and structural factors of Fe2 O3 /In2 O3 sensors on their selectivity and with side-by-side heterojunctions, J. Mater. Sci. Mater. Electron. 24 (2013)
sensitivity to ethanol, Sens. Actuators B: Chem. 96 (2003) 498–503. 2287–2291.
[50] C.S. Dandeneau, Y.-H. Jeon, C.T. Shelton, T.K. Plant, D.P. Cann, B.J. Gibbons, [79] Y.-C. Her, C.-K. Chiang, S.-T. Jean, S.-L. Huang, Self-catalytic growth of hier-
Thin film chemical sensors based on p-CuO/n-ZnO heterocontacts, Thin Solid archical In2 O3 nanostructures on SnO2 nanowires and their CO sensing
Films 517 (2009) 4448–4454. properties, CrystEngComm 14 (2012) 1296.
[51] I. Kosc, I. Hotovy, V. Rehacek, R. Griesseler, M. Predanocy, M. Wilke, et al., [80] S. Sen, P. Kanitkar, A. Sharma, K.P. Muthe, A. Rath, S.K. Deshpande, et al.,
Sputtered TiO2 thin films with NiO additives for hydrogen detection, Appl. Growth of SnO2 /W18 O49 nanowire hierarchical heterostructure and their
Surf. Sci. 269 (2013) 110–115. application as chemical sensor, Sens. Actuators B: Chem. 147 (2010) 453–460.
[52] G. Cui, M. Zhang, G. Zou, Resonant tunneling modulation in quasi-2D [81] J.Y. Lao, J.G. Wen, Z.F. Ren, Hierarchical ZnO nanostructures, Nano Lett. 2
Cu2 O/SnO2 p–n horizontal-multi-layer-heterostructure for room tempera- (2002) 1287–1291.
ture H2 S sensor application, Sci. Rep. 3 (2013) 1–8. [82] Y. Chen, C. Zhu, X. Shi, M. Cao, H. Jin, The synthesis and selective gas sensing
[53] G. Cui, L. Gao, B. Yao, S. Wang, P. Zhang, M. Zhang, Electrochemistry of characteristics of SnO2 /␣-Fe2 O3 hierarchical nanostructures, Nanotechnol-
CuO/In2 O3 p–n heterojunction nano/microstructure array with sensitivity to ogy 19 (2008) 205603.
H2 at and below room-temperature, Electrochem. Commun. 30 (2013) 42–45. [83] A. Maiti, J. Rodriguez, M. Law, SnO2 nanoribbons as NO2 sensors: insights from
[54] U. Shaislamov, B.L. Yang, CdS-sensitized single-crystalline TiO2 nanorods and first principles calculations, Nano Lett. 3 (2003) 1025–1028.
polycrystalline nanotubes for solar hydrogen generation, J. Mater. Res. 28 [84] L. Mazeina, Y.N. Picard, S.M. Prokes, Controlled growth of parallel oriented
(2012) 418–423. ZnO nanostructural arrays on Ga2 O3 nanowires, Cryst. Growth Des. 9 (2009)
[55] T. Rakshit, S.P. Mondal, I. Manna, S.K. Ray, CdS-decorated ZnO nanorod het- 1164–1169.
erostructures for improved hybrid photovoltaic devices, ACS Appl. Mater. [85] H. Huang, H. Gong, C.L. Chow, J. Guo, T.J. White, M.S. Tse, et al., Low-
Interfaces 4 (2012) 6085–6095. temperature growth of SnO2 nanorod arrays and tunable n–p–n sensing
[56] L. Xing, S. Yuan, Z. Chen, Y. Chen, X. Xue, Enhanced gas sensing performance response of a ZnO/SnO2 heterojunction for exclusive hydrogen sensors, Adv.
of SnO2 /␣-MoO3 heterostructure nanobelts, Nanotechnology 22 (2011) 1–6. Funct. Mater. 21 (2011) 2680–2686.
[57] J. Chen, K. Wang, W. Zhou, Vertically aligned ZnO nanorod arrays coated [86] M. Wagh, L. Patil, T. Seth, D. Amalnerkar, Surface cupricated SnO2 –ZnO thick
with SnO2 /noble metal-nanoparticles for highly sensitive and selective gas films as a H2 S gas sensor, Mater. Chem. Phys. 84 (2004) 228–233.
detection, IEEE Trans. Nanotechnol. 10 (2011) 968–974. [87] Q. Kuang, C. Lao, Z. Li, Y. Liu, Enhancing the photon-and gas-sensing prop-
[58] A. Chowdhuri, P. Sharma, V. Gupta, K. Sreenivas, K.V. Rao, H2 S gas sensing erties of a single SnO2 nanowire based nanodevice by nanoparticle surface
mechanism of SnO2 films with ultrathin CuO dotted islands, J. Appl. Phys. 92 functionalization, J. Phys. Chem. B 112 (2008) 11539–11544.
(2002) 2172. [88] C.L. Zhu, Y.J. Chen, R.X. Wang, L.J. Wang, M.S. Cao, X.L. Shi, Synthesis and
[59] A. Chowdhuri, V. Gupta, K. Sreenivas, R. Kumar, S. Mozumdar, P.K. Patanjali, enhanced ethanol sensing properties of ␣-Fe2 O3 /ZnO heteronanostructures,
Response speed of SnO2 -based H2 S gas sensors with CuO nanoparticles, Appl. Sens. Actuators B: Chem. 140 (2009) 185–189.
Phys. Lett. 84 (2004) 1180. [89] P. Hidalgo, R. Castro, Surface segregation and consequent SO2 sensor response
[60] H.-R. Kim, A. Haensch, I.-D. Kim, N. Barsan, U. Weimar, J.-H. Lee, The role of NiO in SnO2 –NiO, Chem. Mater. 17 (2005) 4149–4153.
doping in reducing the impact of humidity on the performance of SnO2 -based [90] T. Jinkawa, G. Sakai, J. Tamaki, N. Miura, N. Yamazoe, Relationship between
gas sensors: synthesis strategies, and phenomenological and spectroscopic ethanol gas sensitivity and surface catalytic property of tin oxide sensors
studies, Adv. Funct. Mater. 21 (2011) 4456–4463. modified with acidic or basic oxides, J. Mol. Catal. A: Chem. 155 (2000)
[61] Z. Lou, J. Deng, L. Wang, L. Wang, T. Fei, T. Zhang, Toluene and ethanol sensing 193–200.
performances of pristine and PdO-decorated flower-like ZnO structures, Sens. [91] Z.A. Ansari, S.G. Ansari, T. Ko, J. Oh, Effect of MoO3 doping and grain size on
Actuators B: Chem. 176 (2013) 323–329. SnO2 -enhancement of sensitivity and selectivity for CO and H2 gas sensing,
[62] H.-S. Woo, C.W. Na, I.-D. Kim, J.-H. Lee, Highly sensitive and selec- Sens. Actuators B: Chem. 87 (2002) 105–114.
tive trimethylamine sensor using one-dimensional ZnO–Cr2 O3 hetero- [92] W.-H. Zhang, W.-D. Zhang, Fabrication of SnO2 –ZnO nanocomposite sensor
nanostructures, Nanotechnology 23 (2012) 245501. for selective sensing of trimethylamine and the freshness of fishes, Sens.
[63] C.M. Chang, M.H. Hon, I.C. Leu, Influence of size and density of Au nanopar- Actuators B: Chem. 134 (2008) 403–408.
ticles on ZnO nanorod arrays for sensing reducing gases, J. Electrochem. Soc. [93] Q.-H. Xu, D.-M. Xu, M.-Y. Guan, Y. Guo, Q. Qi, G.-D. Li, ZnO/Al2 O3 /CeO2 com-
160 (2013) B170–B176. posite with enhanced gas sensing performance, Sens. Actuators B: Chem. 177
[64] S.D. Bakrania, M.S. Wooldridge, The effects of the location of Au additives on (2013) 1134–1141.
combustion-generated SnO2 nanopowders for CO gas sensing, Sensors (Basel) [94] Y. Hu, X. Zhou, Q. Han, Q. Cao, Y. Huang, Sensing properties of CuO–ZnO
10 (2010) 7002–7017. heterojunction gas sensors, Mater. Sci. Eng. B 99 (2003) 2002–2004.
[65] T.-J. Hsueh, S.-J. Chang, C.-L. Hsu, Y.-R. Lin, I.-C. Chen, Highly sensitive ZnO [95] L.C. Tien, D.P. Norton, B.P. Gila, S.J. Pearton, H.-T. Wang, B.S. Kang, et al., Detec-
nanowire ethanol sensor with Pd adsorption, Appl. Phys. Lett. 91 (2007) tion of hydrogen with SnO2 -coated ZnO nanorods, Appl. Surf. Sci. 253 (2007)
053111. 4748–4752.
[66] X. Liu, J. Zhang, X. Guo, S. Wang, S. Wu, Core–shell ␣-Fe2 O3 @SnO2 /Au hybrid [96] H. Tang, M. Yan, H. Zhang, S. Li, X. Ma, M. Wang, et al., A selective NH3
structures and their enhanced gas sensing properties, RSC Adv. 2 (2012) 1650. gas sensor based on Fe2 O3 –ZnO nanocomposites at room temperature, Sens.
[67] S. Basu, P.K. Basu, Nanocrystalline metal oxides for methane sensors: role of Actuators B: Chem. 114 (2006) 910–915.
noble metals, J. Sens. 2009 (2009) 1–20. [97] D.R. Patil, L.a. Patil, P.P. Patil, Cr2 O3 -activated ZnO thick film resistors for
[68] G. Korotcenkov, Gas response control through structural and chemical modi- ammonia gas sensing operable at room temperature, Sens. Actuators B: Chem.
fication of metal oxide films: state of the art and approaches, Sens. Actuators 126 (2007) 368–374.
B: Chem. 107 (2005) 209–232. [98] W. Zeng, T. Liu, Z. Wang, Sensitivity improvement of TiO2 -doped SnO2 to
[69] C. Liewhiran, S. Phanichphant, Effects of palladium loading on the response of volatile organic compounds, Phys. E Low-Dimensional Syst. Nanostruct. 43
thick film flame-made ZnO gas sensor for detection of ethanol vapor, Sensors (2010) 633–638.
7 (2007) 1159–1184. [99] N. Singh, A. Ponzoni, R.K. Gupta, P.S. Lee, E. Comini, Synthesis of In2 O3 –ZnO
[70] R.L. Vander Wal, G.M. Berger, M.J. Kulis, G.W. Hunter, J.C. Xu, L. Evans, core–shell nanowires and their application in gas sensing, Sens. Actuators B:
Synthesis methods, microscopy characterization and device integration of Chem. 160 (2011) 1346–1351.
D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272 271

[100] M. Ivanovskaya, D. Kotsikau, G. Faglia, P. Nelli, S. Irkaev, Gas-sensitive [131] P.V. Kamat, Meeting the clean energy demand: nanostructure architectures
properties of thin film heterojunction structures based on Fe2 O3 –In2 O3 for solar energy conversion, J. Phys. Chem. C 111 (2007) 2834–2860.
nanocomposites, Sens. Actuators B: Chem. 93 (2003) 422–430. [132] I.E. Wachs, K. Routray, Catalysis science of bulk mixed oxides, ACS Catal. 2
[101] D. Xu, M. Guan, Q. Xu, Y. Guo, Y. Wang, Ethanol sensor of CdO/Al2 O3 /CeO2 (2012) 1235–1246.
obtained from Ce-doped layered double hydroxides with high response and [133] M. Keane, Ceramics for catalysis, J. Mater. Sci. 38 (2003) 4661–4675.
selectivity, Funct. Mater. Lett. 06 (2013) 1350035. [134] M. Boutonnet, S. Lögdberg, E. Elm Svensson, Recent developments in the
[102] M.-R. Yu, R.-J. Wu, M. Chavali, Effect of “Pt” loading in ZnO–CuO hetero- application of nanoparticles prepared from w/o microemulsions in hetero-
junction material sensing carbon monoxide at room temperature, Sens. geneous catalysis, Curr. Opin. Colloid Interface Sci. 13 (2008) 270–286.
Actuators B: Chem. 153 (2011) 321–328. [135] K. Zhu, D. Wang, J. Liu, Self-assembled materials for catalysis, Nano Res. 2
[103] N. Bahlawane, E. Fischer Rivera, K. Kohse-Höinghaus, A. Brechling, U. (2009) 1–29.
Kleineberg, Characterization and tests of planar Co3 O4 model catalysts pre- [136] W.-W. Wang, Y.-J. Zhu, L.-X. Yang, ZnO–SnO2 hollow spheres and hierarchical
pared by chemical vapor deposition, Appl. Catal. B: Environ. 53 (2004) nanosheets: hydrothermal preparation, formation mechanism, and photocat-
245–255. alytic properties, Adv. Funct. Mater. 17 (2007) 59–64.
[104] M. Mashock, K. Yu, S. Cui, S. Mao, G. Lu, J. Chen, Modulating gas sensing prop- [137] L. Zheng, Y. Zheng, C. Chen, Y. Zhan, X. Lin, Q. Zheng, et al., Network structured
erties of CuO nanowires through creation of discrete nanosized p–n junctions SnO2 /ZnO heterojunction nanocatalyst with high photocatalytic activity,
on their surfaces, ACS Appl. Mater. Interfaces 4 (2012) 4192–4199. Inorg. Chem. 48 (2009) 1819–1825.
[105] K. Jain, R.P. Pant, S.T. Lakshmikumar, Effect of Ni doping on thick film SnO2 [138] Y. Jiang, M. Wu, X. Wu, Y. Sun, H. Yin, Low-temperature hydrothermal syn-
gas sensor, Sens. Actuators B: Chem. 113 (2006) 823–829. thesis of flower-like ZnO microstructure and nanorod array on nanoporous
[106] K. Zakrzewska, M. Radecka, TiO2 –SnO2 system for gas sensing—photo- TiO2 film, Mater. Lett. 63 (2009) 275–278.
degradation of organic contaminants, Thin Solid Films 515 (2007) 8332–8338. [139] S. Li, K. Yu, Y. Wang, Z. Zhang, C. Song, H. Yin, et al., Cu2 S@ZnO hetero-
[107] T. Maekawa, J. Tamaki, N. Miura, N. Yamazoe, Development of SnO2 -based nanostructures: facile synthesis, morphology-evolution and enhanced
ethanol gas sensor, Sens. Actuators B 9 (1992) 63–69. photocatalysis and field emission properties, CrystEngComm 15 (2013)
[108] A. Maldotti, A. Molinari, Design of heterogeneous photocatalysts based on 1753.
metal oxides to control the selectivity of chemical reactions, Photocatalysis [140] C. Borgohain, K.K. Senapati, K.C. Sarma, P. Phukan, A facile synthesis of
303 (2011) 185–216. nanocrystalline CoFe2 O4 embedded one-dimensional ZnO hetero-structure
[109] P.-X. Gao, P. Shimpi, H. Gao, C. Liu, Y. Guo, W. Cai, et al., Hierarchical assem- and its use in photocatalysis, J. Mol. Catal. A: Chem. 363–364 (2012) 495–500.
bly of multifunctional oxide-based composite nanostructures for energy and [141] L. Lin, Y. Yang, L. Men, X. Wang, D. He, Y. Chai, et al., A highly efficient TiO2 @ZnO
environmental applications, Int. J. Mol. Sci. 13 (2012) 7393–7423. n–p–n heterojunction nanorod photocatalyst, Nanoscale 5 (2013) 588–593.
[110] S.-J. Kim, C.W. Na, I.-S. Hwang, J.-H. Lee, One-pot hydrothermal synthesis of [142] J. Yang, X. Zhang, H. Liu, C. Wang, S. Liu, P. Sun, et al., Heterostructured
CuO–ZnO composite hollow spheres for selective H2 S detection, Sens. Actu- TiO2 /WO3 porous microspheres: preparation, characterization and photocat-
ators B: Chem. 168 (2012) 83–89. alytic properties, Catal. Today 201 (2013) 195–202.
[111] S.S. Kim, H.G. Na, S.-W. Choi, D.S. Kwak, H.W. Kim, Novel growth of CuO- [143] X. Zou, X. Li, Q. Zhao, S. Liu, Synthesis of LaVO4 /TiO2 heterojunction nano-
functionalized, branched SnO2 nanowires and their application to H2 S tubes by sol–gel coupled with hydrothermal method for photocatalytic air
sensors, J. Phys. D: Appl. Phys. 45 (2012) 205301. purification, J. Colloid Interface Sci. 383 (2012) 13–18.
[112] F. Shao, M.W.G. Hoffmann, J.D. Prades, R. Zamani, J. Arbiol, J.R. Morante, et al., [144] M. Ye, D. Vennerberg, C. Lin, Z. Lin, Nanostructured TiO2 architectures
Heterostructured p-CuO (nanoparticle)/n-SnO2 (nanowire) devices for selec- for environmental and energy applications, J. Nanosci. Lett. 2 (2) (2012)
tive H2 S detection, Sens. Actuators B: Chem. 181 (2013) 130–135. http://nanofm.mse.gatech.edu/Papers/Ye%20MD%20et%20al%20J%20Nanosci
[113] S. Ayg, D. Cann, Hydrogen sensitivity of doped CuO/ZnO heterocontact sen- %20Lett%202%201%202012.pdf
sors, Sens. Actuators B: Chem. 106 (2005) 837–842. [145] S. Semancik, R.E. Cavicchi, M.C. Wheeler, J.E. Tiffany, G.E. Poirier, R.M. Walton,
[114] J.H. Yu, G.M. Choi, Electrical and CO gas sensing properties of ZnO–SnO2 com- et al., Microhotplate platforms for chemical sensor research, Sens. Actuators
posites, Sens. Actuators B: Chem. 52 (1998) 251–256. B: Chem. 77 (2001) 579–591.
[115] Z. Jiao, S. Wang, L. Bian, J. Liu, Stability of SnO2 /Fe2 O3 multilayer thin film gas [146] M.S. Khandekar, N.L. Tarwal, J.Y. Patil, F.I. Shaikh, I.S. Mulla, S.S. Suryavanshi,
sensor, Mater. Res. Bull. 35 (2000) 741–745. Liquefied petroleum gas sensing performance of cerium doped copper ferrite,
[116] S. Phadke, J.-Y. Lee, J. West, P. Peumans, A. Salleo, Using align- Ceram. Int. 39 (2013) 5901–5907.
ment and 2D network simulations to study charge transport through [147] M.M. Rahman, S.B. Khan, M. Faisal, A.M. Asiri, K.A. Alamry, Highly sensi-
doped ZnO nanowire thin film electrodes, Adv. Funct. Mater. 21 (2011) tive formaldehyde chemical sensor based on hydrothermally prepared spinel
4691–4697. ZnFe2 O4 nanorods, Sens. Actuators B: Chem. 171–172 (2012) 932–937.
[117] R. Khan, H.-W. Ra, J.T. Kim, W.S. Jang, D. Sharma, Y.H. Im, Nanojunction effects [148] C.V.G. Reddy, S.V. Manorama, V.J. Rao, Semiconducting gas sensor for chlorine
in multiple ZnO nanowire gas sensor, Sens. Actuators B: Chem. 150 (2010) based on inverse spinel nickel ferrite, Sens. Actuators B 55 (1999) 90–95.
389–393. [149] Q. Qi, J. Zhao, R.-F. Xuan, P.-P. Wang, L.-L. Feng, L.-J. Zhou, et al., Sensitive
[118] A. Kar, M. a Stroscio, M. Meyyappan, D.J. Gosztola, G.P. Wiederrecht, M. Dutta, ethanol sensors fabricated from p-type La0.7 Sr0.3 FeO3 nanoparticles and n-
Tailoring the surface properties and carrier dynamics in SnO2 nanowires, type SnO2 nanofibers, Sens. Actuators B: Chem 191 (2014) 659–665.
Nanotechnology 22 (2011) 285709. [150] S. Wang, Y. Kang, L. Wang, H. Zhang, Y. Wang, Y. Wang, Organic/inorganic
[119] M. Iwamoto, Y. Yoda, N. Yamazoe, T. Seiyama, Study of metal oxide catalysts hybrid sensors: a review, Sens. Actuators B: Chem. 182 (2013) 467–481.
by temperature programmed desorption. 4. Oxygen adsorption on various [151] M.E. Franke, T.J. Koplin, U. Simon, Metal and metal oxide nanoparticles in
metal oxides, J. Phys. Chem. 82 (1978) 2564–2570. chemiresistors: does the nanoscale matter? Small 2 (2006) 36–50.
[120] J. He, P. Chang, C. Chen, K. Tsai, Focused-ion-beam-deposited Pt contacts on [152] C. Murugan, E. Subramanian, D.P. Padiyan, p–n Heterojunction formation in
ZnO nanowires, ECS Trans. 16 (2009) 13–20. polyaniline–SnO2 organic–inorganic hybrid composite materials leading to
[121] S. Liu, N. Zhang, Z.-R. Tang, Y.-J. Xu, Synthesis of one-dimensional CdS@TiO2 enhancement in sensor functionality toward benzene and toluene vapors at
core–shell nanocomposites photocatalyst for selective redox: the dual role of room temperature, Synth. Met. 192 (2014) 106–112.
TiO2 shell, ACS Appl. Mater. Interfaces 4 (2012) 6378–6385. [153] U. Tisch, H. Haick, Nanomaterials for cross-reactive sensor arrays, MRS Bull.
[122] N.G. Patel, C.J. Panchal, K.K. Makhija, Use of cadmium selenide thin films as a 35 (2010) 797–804.
carbon dioxide gas sensor, Cryst. Res. Technol. 29 (1994) 1013–1020. [154] Y. Li, J. Gong, G. He, Y. Deng, Fabrication of polyaniline/titanium dioxide com-
[123] B. Tian, J. Zhang, Morphology-controlled synthesis and applications of silver posite nanofibers for gas sensing application, Mater. Chem. Phys. 129 (2011)
halide photocatalytic materials, Catal. Surv. Asia 16 (2012) 210–230. 477–482.
[124] Y. Tang, Z. Jiang, J. Deng, D. Gong, Y. Lai, H.T. Tay, et al., Synthesis of [155] J. Huang, Q. Wan, Gas sensors based on semiconducting metal oxide one-
nanostructured silver/silver halides on titanate surfaces and their visible- dimensional nanostructures, Sensors (Basel) 9 (2009) 9903–9924.
light photocatalytic performance, ACS Appl. Mater. Interfaces 4 (2012) [156] K.J. Choi, H.W. Jang, One-dimensional oxide nanostructures as gas-sensing
438–446. materials: review and issues, Sensors (Basel) 10 (2010) 4083–4099.
[125] R. Xing, Y. Xue, X. Liu, B. Liu, B. Miao, W. Kang, et al., Mesoporous ZnS hier- [157] S.-J. Kim, I.-S. Hwang, Y.C. Kang, J.-H. Lee, Design of selective gas sensors using
archical nanostructures: facile synthesis, growth mechanism and application additive-loaded In2 O3 hollow spheres prepared by combinatorial hydro-
in gas sensing, CrystEngComm 14 (2012) 8044. thermal reactions, Sensors (Basel) 11 (2011) 10603–10614.
[126] T. Fu, Sensing behavior of CdS nanoparticles to SO2 , H2 S and NH3 at room [158] H. Zhang, Q. Zhu, Y. Zhang, Y. Wang, L. Zhao, B. Yu, One-pot synthesis and hier-
temperature, Mater. Res. Bull. 48 (2013) 1784–1790. archical assembly of hollow Cu2 O microspheres with nanocrystals-composed
[127] M. Nolan, First-principles prediction of new photocatalyst materials with porous multishell and their gas-sensing properties, Adv. Funct. Mater. 17
visible-light absorption and improved charge separation: surface modifica- (2007) 2766–2771.
tion of rutile TiO2 with nanoclusters of MgO and Ga2 O3 , ACS Appl. Mater
Interfaces 4 (2012) 5863–5871.
[128] C. Huang, F. Pan, I. Chang, Enhanced photocatalytic decomposition of methy- Biographies
lene blue by the heterostructure of PdO nanoflakes and TiO2 nanoparticles,
Appl. Surf. Sci. 263 (2012) 345–351.
[129] D.Y.C. Leung, X. Fu, C. Wang, M. Ni, M.K.H. Leung, X. Wang, et al., Hydro- Derek R. Miller is a Ph.D. candidate at The Ohio State University. He began his
gen production over titania-based photocatalysts, ChemSusChem 3 (2010) graduate career in 2011 on an NSF GRFP Fellowship and in 2013 was awarded a NASA
681–694. NSTRF Fellowship. His research focuses on growing and synthesizing new oxide
[130] A. Ibhadon, P. Fitzpatrick, Heterogeneous photocatalysis: recent advances and nanostructures for applications in gas sensing, photovoltaics, and biocompatible
applications, Catalysts 3 (2013) 189–218. surfaces. His recent focus on fabrication of oxide nano-heterostructures to enhance
272 D.R. Miller et al. / Sensors and Actuators B 204 (2014) 250–272

performance in these applications has generated important questions that led to the and his main research interests include thin film sensors, metal oxide sensors, and
primary content of this review. Derek was also the 2013 Chair of the ACerS’ PCSA, a nano-textured biomedical coatings.
national ceramics-focused student organization.
Patricia A. Morris is an Associate Professor in Materials Science and Engineering
Sheikh A. Akbar is a Professor in Materials Science and Engineering at The Ohio at The Ohio State University. She received her B.S. from The Ohio State University
State University. He received his Ph.D. in 1985 from Purdue University. He teaches in 1980 and Ph.D. in 1986 from MIT. Dr. Morris was a Technical Leader working on
courses on Sensor Materials, Electronic Ceramics, and Thermodynamics. He is the chemical sensors at Dupont from 1999 to 2005. Her current research interests focus
Founding Director of the NSF Center for Industrial Sensors and Measurements (CISM) on oxide nanoparticles deposited on miniaturized sensors using ink-jet printing.

You might also like