Improving the gas sensing property of WO3 nanomaterials

The principle for gas sensing applications using metal oxide semiconductor based on the change in resistance of the sensitive layer in presence of gases.. One of the metal oxide material

MINISTRY OF EDUCATION AND TRAINING HANOI UNIVERSITY OF SCIENCE AND TECHNOLOGY

-

NGUYEN HOANG HUNG

IMPROVING THE GAS SENSING PROPERTY OF

MAJOR: ENGINEERING PHYSICS

MASTER OF SCIENCE THESIS ENGINEERING PHYSICS

SUPERVISOR: Dr DANG DUC VUONG

BỘ GIÁO DỤC VÀ ĐÀO TẠO TRƯỜNG ĐẠI HỌC BÁCH KHOA HÀ NỘI

-

NGUYỄN HOÀNG HƯNG

NGHIÊN CỨU CẢI THIỆN ĐẶC TÍNH NHẠY KHÍ

TABLE OF CONTENT

TABLE OF CONTENT i

LIST OF ABBRIVATION iii

LIST OF TABLE iv

LIST OF FIGURE v

PREFACE vii

ACKNOWLEDGMENT ix

CHAPTER 1: INTRODUCTION 1

I Chemical sensors and Gas sensors based on metal oxides 1

1 Chemical sensors 1

2 Gas sensors based on metal oxides 2

2.1 Characterized features of gas sensors based on metal oxides 4

2.1.1 Sensitivity 4

2.1.2 Response and recovery time 5

2.1.3 Selectivity 5

2.1.4 Stability 5

2.2 Basic scientist approach 6

2.2.1 The nature of gas sensitivity 6

2.2.2 Factors affecting the sensitivity 8

3 Approaches 12

II Motivation and objectives 22

1 Tungsten oxide among metal oxides for gas detection 22

2 Structural properties of tungsten oxide 25

3 Gas sensors based on tungsten oxide 27

CHAPTER 2: EXPERIMENTAL AND METHODOLOGY 33

I WO3 materials synthesis and doping 33

1 WO3 microsheets synthesis 33

2 WO3 nanoparticles synthesis 34

3 Doping 35

II Characterization 35

1 Electron microscopy 36

1.1 Interaction of high energy electrons with matter 36

1.2 SEM and FESEM 37

2 Chemical Analysis 38

2.1 Energy dispersive X-ray analysis 38

2.2 X-ray diffraction 38

3 Gas sensing properties 40

CHAPTER 3: RESULT AND DISCUSSION 43

I Materials synthesis 43

1 Tungsten trioxides microsheets 43

2 Tungsten trioxides nanoparticles and doping 45

II Gas sensing properties 46

1 Tungsten trioxides microsheets 46

2 Tungsten trioxides nanoparticles and doping 49

CONCLUSION 54

RELATED PRESENTATION 55

OUTLOOK 56

REFERENCE 57

LIST OF ABBRIVATION

 HS: Hydrothermal and Solvothermal

 XRD: X-ray Diffraction

 SEM: Scanning electron microscope

 FE-SEM: Field Emission Scanning Electron Microscopy

 EDX or EDS: Energy-dispersive X-ray spectroscopy

 TEM: Transmission electron microscopy

 CNTs: Carbon nanotubes

 VLS: Vapor-liquid-solid

 PVD: Physical vapor deposition

 CVD: Chemical vapor deposition

 LEIS: Low energy ion scattering

 DA: Depletion approximation

LIST OF TABLE

Table 1 Sign of resistance change to change in gas atmosphere [93] 4

Table 2 Typical deposition techniques 20

Table 3 Known polymorphs of tungsten trioxide 26

Table 4 Occupational Exposure Standards 2000 28

Table 5 Some properties of NH3 28

Table 6 Requirements for NH3 gas detection equipment 29

Table 7 Selected publications on NH3 gas sensors based on WO3 31

LIST OF FIGURE

Figure 1 Chemical sensors 1

Figure 2 Cross-section of a chemical sensor 2

Figure 3 An example of resistance change when a reduce gas was introduced 5

Figure 4 Response and recovery time 1

Figure 5 Schematic representation of barrier formation 8

Figure 6 Three mechanisms of conductance 10

Figure 7 Chemical (a) and electronic (b) sensitization schemes 11

Figure 8 Sol-gel processing options 13

Figure 9 Pressure/temperature map of material processing techniques 16

Figure 10 Particle processing by conventional and hydrothermal 16

Figure 11 General purpose pressure autoclave and white Teflon 1

Figure 12 VLS synthesis apparatus 18

Figure 13 Map of temperature variations in furnace 1

Figure 14 Schematic illustration of nucleation and growth of ZnO nanorods 1

Figure 15 SEM images of ZnO nanorods 1

Figure 16 Screen printing technique 21

Figure 17 Spin coating 22

Figure 18 Dip coating 1

Figure 19 Comparison of the papers published on gas sensors 23

Figure 20 Schematic model of crystalline WO3 in the undistorted cubic phase 1

Figure 21 Structural model of the WO3 grain surface 26

Figure 22 NH3’s structure and symmetry axis 1

Figure 23 Some types of ammonia detector 1

Figure 24 Some of commercialized gas sensors head 1

Figure 25 Schematic diagram of WO3 microsheets synthesis 34

Figure 26 Steps of WO3 nanoparticles synthesis 35

Figure 27 Diagram of heat treatment 1

Figure 28 Electron scattering and secondary signal generation 36

Figure 29 Schematic diagram of an SEM 37

Figure 30 Pt interdigitated electrodes and heater used in system 1

Figure 31 Static gas sensing system and principal circuit 41

Figure 32 Dynamic gas sensing system 41

Figure 33 SEM images of WO3 microsheets 43

Figure 34 The XRD pattern and EDX pattern of WO3 thin film 44

Figure 35 SEM images of WO3 microsheets 1

Figure 36 FESEM images of WO3 nanoparticles 45

Figure 37 SEM images of WO3 nanomaterials 1

Figure 38 EDX partner of Fe2O3 nanorods doped WO3 1% wt 46

Figure 39 Dependence of the electrical resistance on working temperature 1

Figure 40 Response to NH3 of WO3 nanosheets at 60oC and 255oC 47

Figure 41 Response to NH3 of WO3 microsheets at 160C 47

Figure 42 The sensor response of the materials 48

Figure 43 The sensor response as a function of gas concentration 49

Figure 44 Response to NH3 of WO3 nanoparticles at 55 C 50

Figure 45 Response to NH3 of WO3 nanoparticles at 95 C 50

Figure 46 The dependence of gas response on NH3 51

Figure 47 Response to NH3 of WO3 nanoparticles at 294 C 1

Figure 48 The dependence of the sensor response on operating temperature 1

PREFACE

Nowadays, the pollution level is increasing due to the misuse of chemicals in industry, agriculture as well as in life The presence of inflammable gases, toxic gases that have caused large damage to both people and their property Aims to minimize the risks as well as industrialization and modernization of industrial processes, it is necessary to fabricate a kind of environmentally benign devices capable of detecting gases Since then the term “gas sensor” was born

During the last decades of the century, the kind of gas sensor, which was best known, was based on the metal oxide semiconductor In particular, materials such as TiO2, SnO2, WO3, are widely used in gas sensing applications to detect toxic gases The principle for gas sensing applications using metal oxide semiconductor based on the change in resistance of the sensitive layer in presence of gases One of the metal oxide material promising for semiconductor gas sensor applications was tungsten oxide With many advantages such as high sensitivity, low response time, low operating temperature, tungsten oxide material was gradually brought to second place

in the world of gas sensor based on metal oxides semiconductor (after SnO2)

One of the gases that was widely used and caused great impact on human health is ammonia Recently, ammonia (NH3) is used in many industries, the NH3 gas leak in the pipeline has caused serious consequences to health So, in the gases to be detected,

NH3 in one of the most concerned gas and sensitive material to detect this gas that was emphasized by scientists is WO3

Developing in parallel with nanotechnology, WO3 is a sensitive materials even at large sizes, but when the material reach to the size limit, the sensitivity was strongly improved and appear more interesting properties Currently, there are many routes to synthesis WO3 nanomaterials such as ball milling, thermal oxidation, chemical vapour deposition (CVD), physical deposition However, these methods require a rigorous technological processes and conditions It is difficult to obey in Vietnam science condition Recently, wet chemical method combined with hydrothermal technology emerged with many advantages as simple technology, inexpensive, not undemanding

on technological process as well as technical conditions Moreover this method allows mass production and variable morphologies could be synthesized The above advantages make wet chemical method has been studying and using more and more in all over the world

In this thesis, the WO3 materials are synthesized and measured in NH3 gas sensor application The morphological form of the material was deposited by wet chemical methods combining hydrothermal technology Gas sensing properties of the materials was improved by reducing in grain size and doping with Fe2O3 nanorods The thesis title: “Improving the gas sensing property of WO3 material” was selected and the results are presented in three main chapters:

Chapter I Introduction: A short introduction to chemical sensors based on

metal oxides, with a particular emphasis on WO3 This chapter also includes the motivation, targets and organization of this investigation

Chapter II Experimental and methodology: Illustrating the experimental

details used in this work, method to analyze the structural and morphological properties of material, a gas effective sensing system was also built in this chapter

Chapter II Experimental and methodology: Aiming at contributing to the

understanding of the whole gas sensing process

Final Conclusions and future Outlook are also proposed in this thesis

ACKNOWLEDGMENT

First of all, I would like to thank my advisor, Dr Dang Duc Vuong for the scientific support and the detailed, helpful discussions of the results He was not only dedicated to guide and assist me during the experimental process, but also give me knowledge and the most scientific approaches Besides, he has given me many valuable ideas as well as point out the shortcomings that helped me complete this thesis

I am especially grateful to my colleagues in Department of Electronic Materials – School of Engineering Physics, they have helped me a lots in experiment as well as always available to answer me any questions of the basic theory

Finally, I would like to add personal thanks to my family for their love and supporting to my work, although they have never really known what I was doing

Hanoi, 01st of March , 2012

CHAPTER 1: INTRODUCTION

The purpose of this chapter is to present the general framework where this investigation is placed Therefore, it will advance from general aspects of chemical sensors to the more specific details concerning gas sensors based on tungsten oxide

I Chemical sensors and Gas sensors based on metal oxides

1 Chemical sensors

There are many definitions on

chemical sensors, According to

Göpel and Schierbaum [32], the

most simple, understandable and

suitable definition on this is:

“Chemical sensors are devices

which convert a chemical state into

an electrical signal” By “chemical

state”, it must be understood different concentrations or partial pressures of molecules

or ions in a gas, liquid or solid phase If it is not specified, it is often assumed that these chemical sensors are just the primary link of the measuring chain, in other words, an interface between the chemical world and the electronics

Some typical properties associated with chemical sensors, according to Stetter and Penrose [81] are:

 a sensitive layer is in chemical contact with the analyte

 a change in the chemistry of the sensitive layer (a reaction) is produced after the exposure to the analyte

 the sensitive layer is on a platform that allows transduction of the change to electric signals

 they are physically “small”

 they operate in real time

 they do not necessarily measure a single or simple physical or chemical property

 they are typically less expensive and more convenient than an equivalent instrument for the same chemical measurements

As stated above, every chemical sensor is divided into two domains: the physical transducer and the chemical interface layer At the chemical interface, the analyte

Figure 1 Chemical sensors.

interacts chemically with a surface, producing a change in physical/chemical properties These changes are measured by the transducer domain, which monitors this change and generates a related electrical signal (Figure 2)

One way to classify chemical sensors is by the transduction mechanism As shown

in [81], the different transduction principles and the magnitudes that can be measured are:

 Electrochemical: voltages, currents, impedance

 Mechanical: weight, size, shape

 Thermal: heat flow, heat content

 Magnetic: field strength, field detection

 Radiant: frequency, intensity

Figure 2 Cross-section of a chemical sensor.

Besides, chemical sensors can also be arranged by the chemical reaction that occurs at the interface This approach is very interesting since chemical parameters, such as the type of chemical reaction, equilibrium constant, kinetic parameters, will

determine the sensor performance, including sensor sensitivity and selectivity

2 Gas sensors based on metal oxides

Sensors using DC resistance of heated metal oxide semiconductors are members

of the Electrochemical class of chemical sensors, subclass of impedance sensors Gas sensors based on different metal oxides (SnO2, TiO2, In2O3, WO3) should be identified

as different types within the “class” of electrochemical-impedance sensors and are often called MOX sensors (from Metal Oxide sensors) In their simplest configuration, MOX sensors consist of a substrate with a heater, electrodes and a sensitive layer in contact with the electrodes

A gas sensor based on metal oxide has some advantages that make MOX sensors appealing for gas sensor users Although highly debatable, some of them are [37]:

 Low cost, small size and easy to handle (compared to other gas sensors)

 Fast sensor response and recovery

 Robust construction and good mechanical strength

 Long operating life

On the other hand, they have some disadvantages that are still a matter of research for scientists (some of them being common to most of gas sensors):

 Poor selectivity

 Strong influence of humidity

 High power consumption (except micromachined-supported)

From the point of view of a user, the sensor response is usually based on the variation of the DC sensor resistance with gas pressure or concentration, although AC resistance or time-derivative of the resistance can be also measured The typical empirical formula describing the variation of conductance (G) with partial pressure of the gas (p) for MOX sensors is [68]:

(1) Where, G0 and G are conductance before and after expose the target gas, A is a constant, is a factor that was affected by semiconductor type

The relationship between sensor resistance and the concentration of deoxidizing gas can be expressed by the following equation over a certain range of gas concentration [68]:

(2) where: Rs is electrical resistance of the sensor, [C] is gas concentration

Many other empirical formulae have been proposed for the detection of certain gases, to avoid humidity interference or to compensate the drift of the sensor response [6],[8],[10]

The effect related to changing of electrical resistance of a semiconductor in presence of impurities in its volume or at the surface was demonstrated for Ge [14] for the first time in 1953 Later, it was shown that the conductivity of ZnO thin films heated to ~300°C was sensitive to the presence of traces of reactive gases in the air [74] Similar properties were reported for SnO2, with higher stability [101] These results initiated further development of commercial gas sensors

The early metal oxide-based sensor materials possessed a number of unpleasant characteristics, such as high cross-sensitivity, sensitivity to humidity, long-term signal drift and slow sensor response In order to improve sensor performance, a series of various metal-oxide semiconductors have been tested [60] At first, the poor

understanding of sensor response mechanisms caused the use of trial and error strategy in the search of an appropriate material The most successful investigations were connected with SnO2, ZnO, and TiO2 Parallel to this approach, the basic research of metal-oxide materials was carried out in scientific laboratories

2.1 Characterized features of gas sensors based on metal oxides

2.1.1 Sensitivity

Sensitivity is the detection ability of related gas at certain concentration (called the

gas responsibility) On the other hand, Sensitivity is a change of measured signal per analyte concentration unit, the slope of a calibration graph, i.e

In this report, the sensor response was defined as a ratio Rair/Rgas

(3) Where: Rair is the resistance in air

Rgas is the resistance in presence of related gas

In fact, This formula is relative and changeable dependent on target gas The target gas interacts with the surface of the metal oxide film (generally through surface adsorbed oxygen ions), which results in a change in charge carrier concentration of the material This change in charge carrier concentration serves to alter the conductivity (or resistivity,) of the material An n-type semiconductor is one where the majority charge carriers are electrons, and upon interaction with a reducing gas an increase in conductivity occurs Conversely, an oxidizing gas serves to deplete the sensing layer of charge carrying electrons, resulting in a decrease in conductivity A p-type semiconductor is a material that conducts with positive holes being the majority charge carriers; hence, the opposite effects are observed with the material and showing an increase in conductivity in the presence of an oxidizing gas (where the gas has increased the number of positive holes) A resistance increase with a reducing gas is observed, where the negative charge introduced in to the material reduces the positive (hole) charge carrier concentration A summary of the response is provided in Table 1 and an example was showed in figure 3

Table 1 Sign of resistance change to change in gas atmosphere [93]

Classification Oxidizing Gases Reducing Gases

n-type Resistance increase Resistance decrease

p-type Resistance decrease Resistance increase

Figure 3 An example of resistance change when a reduce gas was introduced

a n-type semiconductor b p-type semiconductor.

2.1.2 Response and recovery time

Response time is the time required for sensor to respond to a step concentration

change from initial value to a certain

concentration value

Recovery time is the time it takes for

the sensor signal to return to its initial

value after a step concentration change

from a certain value to initial value

In the gas sensor field, the smaller

response and recovery time the higher

efficiency Generally, the response and

recovery time could be calculated as the time in that the sensor resistance changes about 90% of stable value

2.1.3 Selectivity

Selectivity refers to characteristics that determine whether a sensor can respond

selectively to a group of analyte (gas) or even specifically to a single analyte The appearance of another gas in the environment is not affected to the sensor response The sensor selectivity is depended on some factors such as: materials, doping, materials doped concentration and sensor working temperature

For a general MOX sensor, the selectivity is one of their disadvantages because most of them respond to many analyte The best resolution for this problem concerned that is doping and fabricating different morphologies of sensitive materials

2.1.4 Stability

Stability is the ability of a sensor to provide reproducible results for a certain

period of time This includes retaining the sensitivity, selectivity, response, and recovery time

Figure 4 Response and recovery time

Fabrication methods, heat treatment technology (increasing temperature rate and stability of temperature) are the main factors that affect to the MOX sensors stability Besides, it is also strongly depend on the humidity (The concentration of vapor in air) After a long time working in Vietnam climate condition, a MOX sensors generally take several days to get their initial stable value

In addition to above feature, there are some important parameters [34], such as:

 Detection limit is the lowest concentration of the analyte that can be detected

by the sensor under given conditions, particularly at a given temperature

 Dynamic range is the analyte concentration range between the detection limit

and the highest limiting concentration

 Linearity is the relative deviation of an experimentally determined calibration

graph from an ideal straight line

 Resolution is the lowest concentration difference that can be distinguished by

sensor

 Working temperature is usually the temperature that corresponds to maximum

sensitivity

 Hysteresis is the maximum difference in output when the value is approached

with an increasing and a decreasing analyte concentration range

 Life cycle is the period of time over which the sensor will continuously operate

All of these parameters are used to characterize the properties of a particular material or device An ideal chemical sensor would possess high sensitivity, dynamic range, selectivity and stability; low detection limit; good linearity; small hysteresis and response time; and long life cycle Investigators usually make efforts to approach only some of these ideal characteristics, disregarding the others On one hand, this is because the task of creating an ideal sensor for some gases is extremely difficult, if at all possible On the other hand, real applications usually do not require sensors with all perfect characteristics at once For example, a sensor device monitoring the concentration of a component in industrial process does not need a detection limit at the ppb level, though the response time at range of seconds or less would be desirable

In case of environmental monitoring applications, when the concentrations of pollutants normally change slowly such as ammonia, the detection limit requirements can be much higher, but response time of a few minutes can be acceptable

2.2 Basic scientist approach

2.2.1 The nature of gas sensitivity in semiconductor metal oxide nanomaterials

Basically, the actual gas sensing process consists of three different parts: receptor,

transducer and operation mode [79] The receptor is the surface of the metal oxide, where chemical species undergo adsorption, reaction and desorption Traditionally, the adsorption of a gas species on a solid has been divided into physic-sorption and chemisorption Although arguably, a molecule is considered to be chemisorbed if there is an electronic transfer between the gas and the solid, whereas there is no transfer in the case of physic-sorption Ideally, the interaction of the gaseous molecules will induce a change in the depletion layer of the metal oxide grain (see next subsection) These changes are transduced into an electrical signal depending on the microstructure of the sensitive film (the transducer) The porosity of the film, the grain size and the different grain intersections will determine the output signal, which takes into account the whole sensitive layer This output signal is usually electric, although the measurement of the thermo-voltage or of the changes in the sensor temperature is also possible Excellent reviews providing more details can be found in [63],[33], Hereafter the attention will be focused on the role of oxygen surface species, the role played by catalytic additives and the microstructure of the sensitive film

According to Williams and Moseley [13],[31],[63] most target gases are detected due to their influence on the oxygen stoichiometry of the surface Many studies have revealed that the key reaction involves modulation of the concentration of surface oxygen ions The reactions involved in generating conductivity changes are reported

to be confined to the first monolayer

The change of electrical properties of the metal-oxide semiconductor due to adsorption of gas molecules is primarily connected with the chemisorption of oxygen Molecular oxygen adsorbs on the surface by trapping an electron from the conduction band of the semiconductor At temperatures between 100 and 500°C the ionized molecular and atomic forms can be present at the surface [43] The molecular form dominates below 150°C, whereas above this temperature, ionic species prevail [7] The general reaction equation can be written as:

(4) Here is an oxygen molecule in the ambient atmosphere and e is an electron that can reach the surface, overcoming the electric field resulting from negative charging of the surface Their concentration is ns S denotes unoccupied chemisorption sites for oxygen, is the chemisorbed oxygen with α= 1 or 2 for singly or doubly ionized form and β = 1 or 2 for atomic or molecular form, respectively The presence

of charged species on the surface of a semiconductor induces band bending and

formation of a depletion layer [56] Depending on the type of semiconductor the concentration of charge carriers in the surface layer can be either increased or decreased The space charge layer is described by the thickness Ls and surface potential (Vs) [47],[11]

For granular metal oxides, the formation of a depletion layer at the surface of grains and grain boundaries leads to the formation of Schottky barriers between the oxide crystallites, as depicted in Figure 5 The density of surface oxygen ions and the height and width of Schottky barriers depend on the oxygen partial pressure in the surrounding atmosphere The electronic theory of adsorption [95] is in quantitative agreement with the experimentally observed conductance dependencies of semiconductor layers on oxygen partial pressure [72]

Figure 5 Schematic representation of barrier formation at the grain boundaries due

to the space charge layer

In the figure 5, The shaded part denotes the space charge region (high resistivity); the un-shaded part denotes the core region(low resistivity) CB and VB are the lowest edge of the conduction band and the highest edge of the valence band, respectively

Depending on the content of the atmosphere, the concentration of the surface oxygen ions and therefore the occupation of the surface states can be changed, leading

to the change in conductivity As a measure of gas sensitivity one can use either the conductivity change of the sample, exposed to the analyte-containing atmosphere in relation to its conductivity in the reference gas, or the slope of the dependence of conductivity on analyte concentration [92]

2.2.2 Factors affecting the sensitivity of metal-oxide gas sensor materials

As mentioned above, the requirements for each gas sensor depend on the particular application It is not necessary to have material with a detection limit of one molecule if the sensor is designed to work in the 1 – 10% concentration range Nonetheless, materials with high sensitivity and low detection limit always attract the attention of scientists and engineers In this section, the main approaches for

increasing the gas sensitivity of metal-oxide sensor materials are listed, namely those utilizing the size effects and doping by metal or other metal oxides

a Size and shape effects

Since during the formation of a space charge layer the carrier concentration in volume is decreased only in thickness Ls, three types of conductance mechanisms, as illustrated in Figure 6 can be realized For large crystallites the grain size D >> 2L, and the conductance of the film is limited by Schottky barriers at grain boundaries In this case, the sensitivity is practically independent of D When grain size is comparable to 2L (D = 2L) every conducting channel in the necks between grains becomes small enough to influence the total conductivity Since the number of necks

is much larger the grain contacts, they govern the conductivity of the material and define the size-dependence of gas sensitivity If D < 2L, every grain is fully involved

in the space charge layer, and the electron transport is affected by the charge on the particles surfaces

The considered Schottky barrier formation model was developed for the infinite planar geometry system It can be safely used to describe the barrier formation

semi-in case of large metal-oxide grasemi-ins However, for materials with the grasemi-in sizes comparable to the length of the depletion region, the effect of the curvature cannot be neglected, since the density of surface states depends on the grain radius [17],[59]

The shape of a bottom of conductive band for the grains of different size was studied theoretically Application of the depletion approximation (DA) under spherical symmetry allowed the calculation of an analytical solution for the potential

, where is the energy of the bottom of of the conductive band at a distance r from the center of a grain For a more general case, where DA is not applicable, the Poisson equation using a complete expression for the charge density can be solved only numerically The calculated potential shape inside the grains agrees with the experimentally observed flattening of the band bending for films in air [58],[59], Thus, by reducing the particle size the conduction of the sample may be controlled by the grain boundaries, necks, or grains The latter case is the most desirable, since it allows achieving the highest resistance change For different semiconductor oxides the length of depletion layer may vary in the range of 1-100

nm Numerous experimental investigations of nanostructured metal-oxide films revealed a strong increase in sensitivity when the average grain size was reduced to several nanometers [43],[1],[36],[85],[62] Systematic analysis of size-dependence of SnO2 sensitivity was presented recently [44],[22]

Figure 6 Three mechanisms of conductance in metal-oxide gas-sensitive materials

Another prospective approach is to affect the sensitivity by changing the microstructure and porosity For this purpose the low-temperature vapor co-deposition

of metal and inert gas can be used After removing the gas by annealing, the highly porous metal structure can be formed Then, metal can be oxidized by reaction with oxygen This approach was used for preparation of porous Pb/PbO nanostructures by co-deposition of Pb vapors with CO2 at 80 K followed by annealing [29] SnO2 and TiO2 mesoporous powders fabricated using a self-assembly of a surfactant followed

by treatment with phosphoric acid as well as conventional tin oxide powders with surfaces modified by mesoporous SnO2 show higher sensor performance than corresponding metal oxide powder materials, which have lower specific surface area [77],[27] Other porous metal oxides also exhibit increased gas sensitivity [75], [78],[100]

In recent years, a definite trend in using quasi one-dimensional (1D) nano-objects for gas sensor applications has been observed [19],[20],[23],[42],[48],[70] This is due partly to expanding opportunities for synthesis and characterization of such structures [35] Besides, the application of nanowires, nanorods, nanobelts, and nanotubes for gas sensors can significantly lower the detection limit, since the conductance of 1D objects is affected by lower amounts of adsorbed analyte than is the case for thin granular films It was found that SnO2 nanowires are sensitive to low CO concentrations, so the gas sensitivity of SnO2 nanobelts (the quasi-1D materials with defined crystal structure) to polluting gases like CO, NO2, and ethanol was tested [19]

An additional increase of sensitivity can be achieved by creating 1D objects with necks that define the conductivity of the whole nano-object The comparative study of the sensor response to 0.4 ppm of hydrogen of straight SnO2 nanowires with diameter

of ~100 nm and segmented nanowires consisting of thick parts 500 nm in diameter connected by thin parts 10 nm in diameter [26] was carried out It was found that response is larger for segmented nanowires, despite the fact that their mean radius is almost three times larger than that of straight nanowires

The chemical scheme is usually considered in the case of catalytic additives, for example Pt or Pd doped MOX material exposed to hydrogen, hydrocarbons, or carbon monoxide [28],[87],[16] It is supposed that the reduction of gas molecules is first activated by the metal surface, forming the active surface species that then react via a spillover process with the charged oxygen molecules, adsorbed on tin oxide This reaction leads to the re-injection of the localized electrons back to the bulk, thus increasing the conductivity of the material For instance, the sensing mechanism proposed for the Pt/MOX + CO system involves two main processes [28] First, at elevated temperature, Pt is oxidized by the chemisorbed oxygen:

(5)

Second, exposure to CO leads to a reduction of platinum oxide:

(6)

3 Approaches

In recent years, there are many research group who have tried completing their research into gas sensors by minimize error or by a systematic research approach The goal is to determine a parameter set so that optimized sensors can be manufactured and applied in practice for fulfilling certain tasks such as monitoring different substances in different environmental conditions [9] Two of these parameters will be briefly described here: Methods for synthesis the sensitive materials and deposition-preparation the layer on substrate

3.1 Methods for synthesis the sensitive materials

Up to now, there are many successful methods for synthesis of MOX materials with different morphologies Some of them are simple, very useful and suitable in Vietnam condition, that make strong attention for scientists of gas sensor field, such

as Sol-gel, Vapor – Solid – Liquid and hydrothermal

3.1.1 Sol – gel

The sol-gel process is a wet-chemical technique widely used in the fields of materials science and ceramic engineering Such methods are used primarily for the fabrication of materials (typically metal oxides) starting from a colloidal solution (sol) that acts as the precursor for an integrated network (or gel) of either discrete particles

or network polymers Typical precursors are metal alkoxides and metal salts (such as chlorides, nitrates and acetates), which undergo various forms of hydrolysis and polycondensation reactions.[40]

In this chemical procedure, the “sol” (or solution) A sol is a stable dispersion of colloidal particles or polymers in a solvent The particles may be amorphous or crystalline An aerosol is particles in a gas phase, while a sol is particles in a liquid A gel consists of a three dimensional continuous network, which encloses a liquid phase,

In a colloidal gel, the network is built from agglomeration of colloidal particles In a polymer gel the particles have a polymeric sub-structure made by aggregates of sub-colloidal particles Generally, the sol particles may interact by van der Waals forces or hydrogen bonds A gel may also be formed from linking polymer chains In most gel systems used for materials synthesis, the interactions are of a covalent nature and the gel process is irreversible The gelation process may be reversible if other interactions are involved The method has some advantages, such as:

 The idea behind sol-gel synthesis is to “dissolve” the compound in a liquid in

order to bring it back as a solid in a controlled manner

 Multi component compounds may be prepared with a controlled stoichiometry

by mixing sols of different compounds

 The sol-gel method prevents the problems with co-precipitation, which may be inhomogeneous, be a gelation reaction

 Enables mixing at an atomic level

 Results in small particles, which are easily sinterable

Sol-gel synthesis may be used to prepare materials with a variety of shapes, such as porous structures, thin fibers, dense powders and thin films

Figure 8 Sol-gel processing options

A sol-gel process could be came from different precursor, in common We can divided it into three main route, such as:

- Colloid route

In this route, we use metal salts in aqueous solution, pH and temperature control The salt when dissolved in water will dissociate into ions and the phenomenon of combination of the ions with water molecules to form complexes will occurs This hydrolysis process form the single-complexes, After that, the single-complexes condense with each other to form complex multi-core, also known as colloidal particles (Equation 7 and 8)

Hydrolysis

(7) Condensation-polymerization

(8) Salt that was used for this method is usually nitrate (NO3), chloride (Cl) and sulfate (SO4) based salt For synthesizing SnO2 nanoparticles, It often goes from SnCl4.5H2O solution reacts with NH4OH, (NH4) 2SO4 or NH4HCO3

[24] For example Group G Saikai, Kyushu University, Japan have used sol-gel method combining hydrothermal technology from SnCl4 in NH4HCO3 solution to form SnO2 particles of 6 nm uniform [5] On the other hand, in order to synthesis of α-

Fe2O3 nanorods, the salt is used as Fe(NO3)3 under the assistance of Na2SO4 The obtained result is α-Fe2O3 nanorods like sea-urchin [66] In this report, WO3 materials, the precursor is sodium tungstate (Na2WO4)

- Metal-organic route

The combination of metal cations with organic ligands is used as precursor in this route The organic ligand consisting of citric acid, carboxylic acid, oleic acid, Phthalic acid The bonding between the ligands in the complexes is the coordinator bonding

so the binding energy should be smaller than the binding energy of the ion thus the polarity characteristic decreases It causes the reaction can easily occurs to make high uniformity and small particle size

- Metal alkoxides route

In addition to above routes, metal alkoxides M(OR)x in alcoholic solution and water addition are also used for sol-gel method Where, M is a metal; (OR) is alkoxides group and R is usually alkyl groups (R = CH3, C2H5 ) Depending on the purpose, we choose the different alkoxides based metals The synthesizing of SnO2

materials by this method may come from tin tetra-isopropoxide alkoxide (TTIP) hydrolyzing in isopropanol, ethanol, methanol, hydroxypropyl cellulose [65] In addition, we also use the catalyst, nitric acid (HNO3), hydrochloric acid (HCl) for example, to control the hydrolysis and condensation through the adjustment of pH

For an easily description, for example SnO2 materials, sol-gel process occurs under the following reactions:

This process is repeated, the compound -Sn-O-Sn- linked together to form colloidal particles During heating and constantly stirring process, the solvent was evaporated, the colloidal particles will bind to each other When the viscosity of the sol solution is reducing the gel structure gradually is taken form In this method, the properties of SnO2 alkoxides largely depends on the alkoxides concentration in solution, the pH of the solution, water amount, temperature, [24] In general, To improve stability and uniformity of the material, we conducted an extra step called hydrothermal

3.1.2 Hydrothermal synthesis

The term ‘hydrothermal’ is of geological origin and has undergone several changes from the original Greek words ‘hydros’ and ‘thermos’ meaning water and heat, respectively It was first used by the British geologist, Sir Roderick Murchison (1792 – 1871), to describe the action of water at elevated temperature and pressure, leading to the formation of various rocks and minerals in the earth’s crust [15]

In chemical nano-world, hydrothermal processing can be defined as any heterogeneous reaction in an aqueous solvent (or non-aqueous solvent for solvothermal processing) under high pressure and temperature conditions, which induces the dissolution and recrystallization of materials that are relatively insoluble under ordinary conditions Figure 9 shows a pressure/temperature map of HS in relation to other material processing techniques In comparative terms, the hydrothermal processing of materials is considered environmentally benign Further, the hydrothermal technique offers the highly controlled diffusivity of strong solvent media in a closed system In the context of nanotechnology, the hydrothermal technique provides an ideal method for producing ‘designer particulates’, i.e mono-dispersed particles with high purity, high crystallinity and controlled physicochemical characteristics Such particles are in great demand by industry

Figure 10 shows the major differences in particle products obtained by ball milling, sintering/firing and hydrothermal methods For example, ball milling involves breaking down bulk material into small irregular shaped particles, and hence is considered a crude fabrication method in comparison to the controlled growth provide

by HS The hydrothermal product particle size can range from a few nanometers up to several microns, depending on temperature, nucleation seed content, pH and solvent concentration

The behavior of solvents under hydrothermal conditions allows the development

of crystal structures under sub- and supercritical states (along with pH variations, viscosity, coefficient of expansion and density, etc.) to be understood in terms of varying pressure and temperature Similarly, thermodynamic studies provide valuable

information on the behavior of solutions with respect to varying pressure and temperature conditions Some commonly studied aspects are solubility, stability, yield and dissolution / precipitation reactions, etc However, fundamental understanding of the kinetics during hydrothermal crystallization is limited This is due to an absence of data relating to the formation of intermediate phases and the inaccessibility of direct

in situ investigation techniques under conditions of high pressure and temperature

Figure 9 Pressure/temperature map of material processing techniques[15]

Figure 10 Particle processing by conventional and hydrothermal techniques,

producing irregular shaped particles and “designer particulates”

Hydrothermal materials processing requires a vessel capable of containing a highly corrosive solvent, operating under extreme pressure and temperature conditions The hydrothermal apparatus, commonly known as an autoclave, reactor, pressure vessel or high pressure bomb, must meet a variety of objectives, processing conditions and tolerances A generic hydrothermal autoclave should be:

 Leak-proof under high pressure/temperature conditions.[15]

 Easily assembled/disassembled

 Inert to acids, bases and oxidizing agents

 Resilient to high pressure and temperature experiments, so that no machining or treatment is needed after each experimental run

In view of the above requirements, autoclaves are generally fabricated from thick glass or quartz cylinders and high strength alloys, such as austenitic stainless steel, iron, nickel, cobalt-based super alloys or titanium and its alloys The primary parameters to be considered in the selection of a suitable reactor are the experimental temperature and pressure conditions,

including corrosion resistance in the

pressure/temperature range for a given

solvent Materials processing from

aqueous phosphoric acid media or other

highly corrosive media, i.e extreme pH

conditions, require the use of an

un-reactive Teflon lining, as shown in Figure

11 or inert tubes (platinum, gold or silver)

to protect the autoclave body from

corrosion

3.1.3 Vapor – Liquid – Solid

The vapor–liquid–solid method (VLS) is a mechanism for the growth of dimensional structures, such as nanowires, from chemical vapor deposition The growth of a crystal through direct adsorption of a gas phase on to a solid surface is generally very slow The VLS mechanism circumvents this by introducing a catalytic liquid alloy phase which can rapidly adsorb a vapor to super-saturation levels, and from which crystal growth can subsequently occur from nucleated seeds at the liquid–solid interface The physical characteristics of nanowires grown in this manner depend, in a controllable way, upon the size and physical properties of the liquid alloy [41]

one-Figure 11 General purpose pressure

autoclave and white Teflon lining used

for HS[15]

For example, to fabricate ZnO nanowires, nanorods, A thin film of gold was sputter deposited onto Si/SiO2 substrates under with the thickness about 10nm The

Au islands were formed on the substrates by heating the coated samples in a rapid thermal annealing furnace at 700 C for 90 min under Ar atmosphere ZnO powder

Figure 12 VLS synthesis apparatus

(99.99 %) and graphite ZnO powder (99.99 %)

was chosen to be source materials for this

example Equal amounts of ZnO powder and

graphite powder (99.99 %) were mixed well

and placed into a small alumina boat The boat

and Au-coated substrate were placed into a

small quartz tube The substrate was placed

about 2 – 10 cm from the source along the

direction of Ar flow Equal amounts of ZnO

powder (99.99 %) and graphite powder (99.99

%) were mixed well and placed into a small

alumina boat as the source The boat and

Au-coated substrate were placed into a small quartz tube The substrate was placed about

2 – 10 cm from the source along the direction of Ar flow Map of temperature variations between the boat and substrate are shown in Figure 13 This unit was then loaded into the greater horizontal quartz tube (4 cm diameter and 110 cm length) inside the central hot zone of a tube furnace The complete unit was heated at 950 –

1150 C for 30 – 90 min under constant flow of argon gas (∼ 20 – 80 sccm) [38]

Figure 13 Map of temperature

variations in furnace

Zn, CO, and CO2 gases are produced by following chemical reactions (12 and 13), and transferred to the Au catalyst surface by Ar flow as shown in Figure 14

(12) (13)

Zn atoms preferentially adsorb on Au droplet surface due to higher sticking coefficient of Zn on liquid versus solid Based on low energy ion scattering (LEIS) measurements, it has been shown that little or no CO, CO2 and O2 adsorb on Au clusters On the figure 14, Zn atoms condense and attach to the edges of nuclei and then oxidized by CO/CO2; lateral growth of ZnO nuclei causes the completion of one ZnO monolayer

Figure 14 Schematic illustration of nucleation and growth of

ZnO nanorods by the VLS mechanism [38]

As observed in figure 14, we can see the nanorods are grown on Au/SiO2

substrates The range of nanorods diameter are 40–120 nm In the images, an Au–Zn alloyed droplets are present at the tips of nanorods This could be strong evidence that ZnO nanorods have been grown by the VLS mechanism

This work is one of the bottom up nanometers approaches By changing synthesis conditions and sourced materials, it is possible to obtain the similar results for other materials, SnO2 for example

3.2 Methods for coating sensitive materials onto the substrates

Table 2 Typical deposition techniques used for the preparation of gas-sensitive

materials

PVD

Sputtering Evaporation

The synthesis and deposition of the sensing layer is obviously the most crucial part in the preparation of gas sensors Three main groups can be distinguished: powder/slurry deposition, chemical vapour deposition (CVD) and physical vapour deposition (PVD) [9](Table 2) The main difference between powder/slurry based films and CVD or PVD has traditionally been attributed to their different film thickness While the former lead to sensitive layers of several microns of thickness (thick films), the layer thickness of the latter varies between 20 and 1000 nm Beyond this classification, there is a fundamental difference in the microstructure of these thick and thin films Thin films are usually very “compact” (not porous), so the interaction with gas is limited to the external surface of the sensitive layer On the other hand, gas can penetrate through most of thick films and so the interaction can occur throughout the whole layer This has led to some authors argue that thick film must be more sensitive than thin films [79], since the change of conductivity is not limited to the outermost zone of the sensitive layer but to the whole layer

Nevertheless, this classification must be carefully taken For instance, it is well known that spin-coating techniques, which are actually using a slurry, are able to obtain “thin films” in the sense of thickness (and the slurry can be obtained by a sol-

Figure 15 SEM images of ZnO nanorods

gel process, for example), being actually “thick films” in terms of porosity [71] As regards screen-printing, it must be understood that this technique is a two-step process: firstly the powder is obtained (by sol-gel, precipitation or any other method) and then a slurry based on this powder is screen printed

Regarding substrates, thick films have been typically deposited on alumina substrates provided with electrodes (usually interdigitated) and a heater Thin films are of course deposited on flatter surfaces, i.e silicon, what allows the use of micromachined gas sensors However, the compatibility between powder technology and micromachined substrates have been also presented, what opens a new line of low power-consumption gas sensors with thick sensitive films [79]

3.2.1 Screen printing

Figure 16 Screen printing technique

Screen printing is a very popular technique for the fabrication of thick film Three steps of this technique are described in Figure 16 The starting materials are often in the form of fine powder that was mixed with a suitable solvent to form colloids The masks had been designed for opening space where need to be covered the materials The materials were spreads on the surface of the grid by the lever system, which is then compressed through the chinks of the mask and to be pasted onto the substrate Finally, the entire substrate is heat treated to stabilize the membrane and remove the previous organic solvent This technique has the advantage of making the film with uniform thickness of the membrane, that is exactly the same as thickness of mask Membrane thickness is usually made from several μm to hundreds of μm

3.2.2 Spin coating

Spin coating has been used for several decades for the application of thin films A typical process involves depositing a small puddle of a fluid resin onto the center of a substrate and then spinning the substrate at high speed (typically around 3000 rpm) Centripetal acceleration will cause the resin to spread to, and eventually off, the edge

of the substrate leaving a thin film of resin on the surface Final film thickness and other properties will depend on the nature of the resin (viscosity, drying rate, percent solids, surface tension, etc.) and the parameters chosen for the spin process Factors

such as final rotational speed, acceleration, and fume exhaust contribute to how the properties of coated films are defined The substrate should be placed on a supporting plate and is fixed by fitting plate or vacuum system For more detailed, the technique was shown in figure 17

Figure 17 Spin coating

One of the most important factors in spin coating is repeatability Subtle variations

in the parameters that define the spin process can result in drastic variations in the coated film

3.2.3 Dip coating

Dip coating is an effective technology of making thin film The process could be described as follows (figure 18): The

substrate should be fixed to a motor The

motor rotation can be controlled at

different speed The substrate were

embedded in the membrane solution The

motor in on low speed model and pull

back the substrate slowly (Figure 18b)

When the substrate go out from the

solution, it will come with a thin layer of

material on the surface (Figure 18c) The

viscosity of the membrane liquid, drag

and rotation speed are adjusted to get the desired film thickness

II Motivation and objectives

1 Tungsten oxide among metal oxides for gas detection

Since Seiyama and Taguchi used the dependence of the conductivity of ZnO on the gas present on the atmosphere for gas sensing applications [74], [83], many different metal oxides have been proposed for gas detection Generally speaking,

Figure 18 Dip coating

these oxides can be divided into binary oxides and more complex oxides, being the former much more common in gas sensing applications

Among binary metal oxides, tin dioxide (SnO2) is the one that has received by far more attention since Taguchi built the first tin oxide sensor for Figaro Sensors in 1970 [84] This is probably due to its high reactivity to many gaseous species However, this characteristic has also revealed as a lack of selectivity, and thus investigation on other metal oxides has been considered necessary

Figure 19 Comparison of the papers published on gas sensors based on ZnO, Fe 2 O 3 ,

TiO 2 , SnO 2 , In 2 O 3 , and WO 3

Besides, developers of electronic noses have experimented with arrays of different sizes that may include around ten MOX sensors, apart from other types of chemical sensors The use of different MOX sensors is highly recommended in order to increase the amount of information Figure 19 displays the number of published papers belonging to different metal oxides for gas sensing applications It is evident that tin oxide receives clearly more attention than the rest However, the number of papers where tungsten oxide is used for gas sensing applications has been increasing during recent years, leading this material to be the second MOX most studied for gas

sensing applications (The number of papers has been evaluated using the database of

Elsevier Publishing house with a typical search of (WO 3 OR (tungsten oxide) OR (tungsten trioxide)) in topic of gas sensor)

2 Structural properties of tungsten oxide

Tungsten trioxide exhibits a cubic perovskite-like structure based on the corner sharing of WO6 regular octahedra, with

the O atoms (W atoms) at the corner

(center) of each octahedron [21] (Figure

20) The crystal network can also be

viewed as the results of alternating

disposition of O and WO2 planes, placed

normally to each main crystallographic

direction This structure is also found in

rhenium trioxide structure (ReO3), from

which takes its common name (ReO3

-structure) This structure is in itself

rather uncommon However, since it

forms the base of perovskite (one of the most important ternaries), it has in fact chief importance

Actually, the symmetry of tungsten oxide is lowered from the ideal ReO3 structure by two distortions: tilting of WO6 octahedra and displacement of tungsten the center of its octahedron [97] Variations in the details of these distortions give rise

to several phase transitions In fact tungsten trioxide adopts at least five distinct crystallographic modifications between absolute zero and its melting point at 1700 ºK When the temperature is decreased from the melting point, the crystallographic symmetry for WO3 changes in the sequence: tetragonal – orthorhombic – monoclinic – triclinic – monoclinic Most of the transitions appear to be first order, and they often display large hysteresis in the transition temperatures A summary of these transitions

is given in Table 3 [29],[52] It is interesting to notice that, as suggested by Table 3 and confirmed experimentally in [52], the coexistence of triclinic and monoclinic phases in WO3 at room temperature is common

Another point worth noting is that the tungsten trioxide structure is likely to host several kinds of defects One of the most elementary defects, as in most metal oxides,

is the lattice oxygen vacancy, where an oxygen atom is absent from a normal lattice site In many d0 oxides of Ti, V, Nb, Mo and W this sort of point defects are largely eliminated by the formation of crystallographic shear phases In the case of WO3, the removal of oxygen causes the appearance of these crystallographic shear planes into

Figure 20 Schematic model of crystalline

WO 3 in the undistorted cubic phase

the crystal along the [1m0] direction [49] This leads to the formation of a family of

WO3-x compounds From an electronic point of view, an oxygen vacancy causes the increase of the electronic density on the metallic (W) adjacent cations, leading to the formation of donor-like states slightly below the edge of the conduction band of the oxide, which acquires semiconducting properties [29] Finally, it is important to point out that in this work, superficial properties of tungsten oxide are of paramount importance, since that is where gas interaction occurs This important point is sometimes overlooked in many papers concerning MOX gas sensors, where bulk properties are extensively reported and little attention is paid at the surface of the material

Table 3 Known polymorphs of tungsten trioxide (Adapted from [52])

Phase Symmetry Temperature range (ºK)

Figure 21 Structural model of the WO 3 grain surface Left panel: idealized WO 3

structure with the (100) fracture planes shown Right panel: two possible states of the grains surface: in both cases the formation of the reduced tungsten ions W 5+ is

required by the neutrality condition (Adapted from [50])

The way to progress from the above-explained crystalline structure and the surface has been proposed by Kuzmin et al [50] If a crack along the (100)