current status of micro and nano structured optical fiber sensors

Accepted ManuscriptTitle: Transport and gas sensing properties of In2O3 nanocrystalline thick films: a Hall effect based approach Authors: A.. The effective values of the charge carrier

Accepted Manuscript

Title: Transport and gas sensing properties of In2O3

nanocrystalline thick films: a Hall effect based approach

Authors: A Oprea, A Gurlo, N Bˆarsan, U Weimar

Sensors and Actuators B: Chemical (2008), doi:10.1016/j.snb.2009.03.002

This is a PDF file of an unedited manuscript that has been accepted for publication

As a service to our customers we are providing this early version of the manuscript.The manuscript will undergo copyediting, typesetting, and review of the resulting proofbefore it is published in its final form Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers thatapply to the journal pertain

Accepted Manuscript

Transport and gas sensing properties of In 2 O 3 nanocrystalline thick films: a Hall effect

based approach

A.Oprea*1, A Gurlo2, N Bârsan1, U Weimar1

1 Institute of Physical and Theoretical Chemistry, University of Tuebingen, Auf der

Morgenstelle 8, 72076 Tuebingen, Germany

2Fachbereich Material- und Geowissenschaften, Technische Universitaet Darmstadt,

Petersenstr 23, 64287 Darmstadt, Germany

Abstract

Undoped nanosized In2O3 with n-type conduction was produced in both polymorphic forms (cubic and rhombohedral) and deposited by screen printing as thick films These films

show high sensitivity to low O3 concentration levels They have been investigated by four

point conductance and Hall Effect measurements under sensor operating conditions (elevated

temperature and ozone exposure) The effective values of the charge carrier concentration and

mobility have been calculated from the experimental records using the recipe for the single

crystals The response to O3 is discussed in the frame of the standard models for gas sensors

The observed deviations from the model are explained in connection with the film crystalline

structure and microscopic parameters spread

Keywords: In2O3, mobility; gas sensitivity

1 Introduction

In2O3 is investigated since decades At the beginning the scientific interest was, more

or less, the reason and the stimulus of investigations However, remarkable optical

transmission and the metallic like conduction (when suitably impurified or having

* Corresponding author: alexandru.oprea@ipc.uni-tuebingen.de

revised Manuscript

Accepted Manuscript

stoichiometric deficiencies) of indium oxide led soon to practical application Either alone or,

most frequently, in combination with other oxides of transition metals, it became the most

utilised transparent conducting oxide (TCO) in optoelectronics and related fields For such

purposes the material is produced as thin and compact layers with low structural defect

concentration The literature abounds in studies performed on films of this type, deposited by

different means on a large variety of substrates Some comprehensive overviews are also

available [1] After observing the high sensitivity of In2O3 towards ozone [2-4] (see also

survey in [5]) or other oxidising gases [6-13] and in direct connection with the rising concern

about the ozone negative effects on ambient quality and human health the interest for gas

sensors based on this compound begun to increase Since then several pertinent investigations

[5, 14] on the O3-sensing properties of In2O3 have been performed resulting in laboratory

versions of chemoresistive gas sensors with detection limits in few parts per billion (ppb)

range

The material utilised in gas sensing should have a very large specific area and therefore it

is typically prepared as porous thin/thick layer The most employed manufacturing paths are

making use of powder technologies, at least in their last step Morphologies with

grain / crystallite dimensions spread over many order of magnitude (few nanometer to

micrometer) [6, 10, 15, 16] or nanostructures [17] have been reported A good understanding

of electrical conduction in films with such structures is indispensable for the optimal design of

the sensing devices In the same time, there is a scientific interest on this topic as well, due to

the strong connection between the combined electrical transport mechanisms, taking place in

and across the material grains and the surface interactions with the gaseous ambient, mainly at

elevated temperatures (250 - 450°C) where the films have to operate as gas sensing elements

In spite of this principle scientific interest we did not find any articles dealing with the

electrical transport in granular In2O3layers for gas sensing and, to the best of our knowledge,

Accepted Manuscript

there are no references addressing the concentration and the mobility of the charge carriers in

such layers

The present paper deal with the electrical conduction of O3 sensing films deposited by

screen printing from undoped nanosized In2O3powders Both In2O3polymorphs, i.e

bixbyite-type c-In2O3 (cubic, C-type structure of rare-earth oxides, space group Ia , No 204) and 3

corundum-type rh-In2O3(rhombohedral, space group R3 , No 167) were studied To the best c

of our knowledge the O3 sensing properties of the undoped corundum-type rh-In2O3 and the

electrical parameters under operation conditions of O3 sensing layers from both polymorphs

have not been evaluated until now Only two articles report the gas sensing investigations of

corundum-type ITO [18] and rh-In2O3[19]

Therefore the investigations, based on Hall Effect [20-22] and four point conductance

measurements, have been performed in synthetic atmospheres with controlled composition

and temperature They aim to provide the missing information concerning the charge carrier

concentration and mobility in In2O3 thick porous layers for gas sensing and, by using it, to

sketch some features of the interplay between the electrical transport and sensing properties in

such films

2 Experimental

As just stated above, the experimental basis of the investigations consists of Hall

Effect and four point conductance measurements on nano-granular screen-printed In2O3 thick

films The samples, heated at temperatures appropriate for gas sensing, have been exposed

during the electrical tests to synthetic atmospheres containing nitrogen, oxygen, ozone and

water vapour (humidity) prepared with a dynamic gas mixing station [23] In this way it is

possible to either reproduce the condition in which a sensor made from the same material is

usually operated (that is, to make operando investigations) [24], or to create a completely

unusual atmosphere, relevant for gas sensing mechanisms and their relation with the electrical

transport [25]

Accepted Manuscript

2.1 Synthesis, structural characterisation of the materials and sample preparation

Synthesis was performed by the sol-gel method based on the ammonia-induced

hydrolysis of indium nitrate in methanol with acetylacetone as a complexing agent

(acetylacetone route hereafter) and without acetylacetone (hydrolysis route hereafter) [26]

The processed powders (calcined in air at 500°C for 1 h) were used for the structural

characterization and for the screen-printing For the present approach it is important to point

out that both In2O3polymorphs, have been obtained through the above referred sol-gel routes

The acetylacetone route resulted in the bixbyite-type c-In2O3, while the hydrolysis route leads

to the corundum-type rh-In2O3 The rh-In2O3 reveals nanorhombohedra terminated by planes

with a size ranging between 50 – 100 nm; the c-In2O3 possess much smaller, highly

agglomerated, crystallites with a size below 20 nm

The structural and morphological characterisation of c-In2O3 is reported in Ref [14],

those of rh-In2O3, in Ref [27, 28]; the detailed characterisation of the sensing properties of

the cordundum-type rh-In2O3will be presented in detail elsewhere [29]

In the last technological step thick (~ 20 µm) sensing films have been screen printed

[14, 30] on substrates suitable for Hall Effect and electrical measurements They are provided

with platinum electrodes on the layer side and with a platinum heating meander on the

opposite one The description of the sample geometry and electrode functionality can be

found in [31, 32]

2.2 Measuring system

The main part of the measuring system has already been described elsewhere [31, 32]

It consists in a gas mixing station which supplies with gaseous test mixture a flat measuring

chamber placed between the polar pieces of a Brucker electromagnet A computer driven

power electronics ensure magnetic field with the strength up to 1 T and required time

dependencies For the present study some supplementary facilities have been added to the

above referred experimental setup They are related to the ozone generation and humidity

Accepted Manuscript

level control O3, the main target gas in the investigations was produced with an Anseros

ozone test system SIM 6000 with an integrated ozone generator and MP UV ozone analyzer;

its concentration in the carrier gas was determined before and after sample exposure by using

the Anseros MP UV ozone analyzer and Environics Series 300 computerized UV ozone

analyzers, respectively Because the wide pulse modulation (WPM) regime of the O3

generator was factory-set in the low frequency range, it was necessary to smear off the O3

concentration variations with a 1 l buffer The time constants of the buffer itself, of about

10 min at 100 sccm carrier gas flow, are strongly reflected by the sample response, but did not

affect the present investigations, intended for near equilibrium conditions

For reasons not clarified yet, in the O3 delivered by the generator a non negligible

level of humidity was present, mainly at high O3generation rates The exact values could not

be determined because of the incompatibility between the existing psychometers and the O3

containing atmosphere In many experiments this parasitic humidity and the humidity

background accidentally present in the gas circuitry, with important consequences in the

sample response towards strong oxidising gases, was reduced below 30 ppm (parts per

million) with a cryogenic N2 vapour trap (to avoid the oxygen condensation) However,

controlled amounts of humidity have been provided by a dedicated channel of the gas

manifold, when required

The O3 loss by adsorption and reaction in the pipe lines and measuring chambers,

which would randomly modify the test mixture composition, was prevented by using adequate

materials as perfluoroalkoxy (PFA) and polytetrafluoroethylene (PTFE, Teflon®)

2.3 Measurement procedure and acquired data

In order to gather data relevant for the influence of the surrounding atmosphere on the

conduction mechanisms in the investigated material three main types of operando

measurements have been carried out: I-V characteristics, four point conductance and dc Hall

voltage In all of them four point sample geometry and electrical set-ups have been utilised

Accepted Manuscript

During the investigations the samples have been exposed to N2 / O2 mixtures with a

mixing ratio varying from 100 ppm O2/ N2 to 100% O2 and, additionally, to O3 dosed in

concentrations from 10 ppb to 2 ppm in some O2/ N2 selected combinations Due to set-up

functional limitations the mixtures containing 50% r.h (relative humidity) contained up to

1ppm O3 only The experimental data have been usually recorded very near to the

thermodynamic equilibrium of the gaseous and solid interacting phases Waiting times of 12 –

24 h between the gas exposure steps have been typically used In the routine tests, intended to

check the material suitability for low level O3sensing, significantly shorter intervals of 2 – 4 h

appeared to be sufficient The parasitic thermoelectric and thermomagnetic effects occurring

together with the Hall Effect, especially at elevated operation temperature, have been

eliminated by making use of reversing magnetic fields and polarisation voltages The time

dependency of the magnetic field, a trapezoidal one, avoids, on one hand, large transient Eddy

currents in the electromagnet coils, and, on the other hand, allows the direct visualisation of

the sample response linearity Therefore the electrical and magneto-electrical measurements

extended over both stationary and transitory regions

3 Results and discussions

The outputs of the performed measurements are three types of raw data: the I-V

characteristics, four point conductance and trapezoidally shaped Hall voltages, all of them

depending on the ambient composition and working temperature Once the linearity of the

responses confirmed by the transitory regions of the records, only the steady state values of

the electrical parameters and the specific sensitivity curves have been considered in the

further analysis of the experimental results

Thus, in the first step of data evaluation, the effective charge carrier concentration and

mobility have been obtained by using the recipe for single crystals [21, 22, 33-35] They give

an intuitive picture of the material behaviour and provide the basis for subsequent analysis

and discussions

Accepted Manuscript

The discussions on the results are starting from the general accepted models for gas

sensing and conduction in MOX semiconductors (as presented and commented in [32, 36,

37])

3.1 I-V characteristics

For each material and measuring condition an I-V characteristic was determined All

of them are very linear over more than 4 orders of magnitude, proving, by that, the pure

ohmic character of the samples, at least under all particular condition occurring during the

investigations Fig 1 shows typical I-V plots in both standard linear scales (left panel) and

logarithmic scales (right panel) for rh-In2O3 samples heated at 200°C One has to briefly

remark the wide range of slopes encountered in the graph with linear scales reflecting the

strong dependence of the material resistance on the O3 concentration In the graph with

logarithmic scales the slope is always the same, and expresses the linearity of the response;

the intercept on a vertical axis is decreasing with the resistance increase The linear behaviour

results from the low voltage drop on each grain, less than 3 mV, which, at the working

temperature of 150°C - 270°C, is well below the thermal voltage  3550

e

T k

(where:T , e , k B denote respectively the absolute temperature, elementary charge and

Boltzmann constant) The evaluation has been done in the most unfavourable conditions, that

is, maximal applied voltage of 100 V, and maximal crystallite diameter of 100 nm, by using

the actual electrode spacing of the samples of 3 mm At the above specified polarisation the

double barriers associated to each grain to grain contact never reach the Schottky diode

operation region (due to polarisation) and the ohmic response is the “normal” one Therefore,

for each measuring conditions the sample resistance / conductance is the only one significant

parameter resulting from an I-V characteristic (actually due to its high linearity) The

experimental values of the conductance determined during the investigation will be given

later on in the §3.4

Accepted Manuscript

3.2 Four point resistance

Self-standing four point resistance measurements have been performed only to prove

the sensitivity of the sensing layers and to provide a “sensor fingerprint” familiar in the field

of gas sensing Fig 2 is well stressing the general trend of the studied materials, namely a

high response to reduced concentrations of O3 The saturation trend of the curves is enhanced

to some extent by the presence of the residual humidity, not removed for these routine tests

We have to point out that we have recently reported unusual O3-sensing properties for

c-In2O3, i.e we observed that the screen-printed c-In2O3 sensors showed saturation in the

two-point conductance measurements even at low O3 concentrations [14] This effect was

explained by possible influence of adsorbed oxygen, similar effects were also observed for

-Fe2O3 [38, 39]

An exhaustive discussion concerning the different sensor parameter (gas response,

sensitivity, selectivity, reproducibility, time constants) will be provided in a dedicated paper

(including also film preparation and morphology), currently under preparation [29]

The results from four point resistance / conductance measurements acquired together

with the Hall voltages during the Hall Effect measurements are included in common graphs

(Fig 3 and Fig 4)

3.3 Hall voltage

The steady state values of the Hall voltage have themselves no direct meaning if taken

alone, but they can provide a rough estimate of the effective concentration of the majority

charge carrier, when using the standard single crystal recipe [21, 22, 33-35] in the first data

evaluation step In the same frame one can also calculate the effective Hall mobility of the

majority charge carriers from the conductance records in a subsequent step The mobility

extraction procedure is susceptible of some errors as long as one utilises the ratio of two

quantities extending over many order of magnitude (with a roughly exponential dependency

on the O3 concentration) and therefore the rapid variations of the obtained values have to be

Accepted Manuscript

considered with care Independent of the mobility value, the Hall measurements confirmed

the negative sign of the free charge carriers participating to the electrical transport in our

In2O3 samples; this feature was already known from the decrease of the material resistivity

under reducing gases exposure [40]

The effective parameters give a more “friendly” description of the transport properties

presented by the investigated materials at macroscopic scale, close to the classical Drude

model [41-44] To, however, relate them to the processes taking place at microscopic level

and in relation with the sample structure / morphology is a difficult - if still solvable - task

[31, 32, 35]

3.4 The influence of the ambient atmosphere on the effective electrical parameters

In the following some significant experimental dependencies of the effective electrical

parameters (single crystal recipe) are presented (Fig 3 and Fig 4) and shortly commented

The trends in the behaviour of the samples are better visible on Fig 5, where the normalised

(relative) values of the considered parameter are included

Before addressing the original results of the investigation it is important to shortly

comment on the experimental dependency of the conductance (and implicitly electron

concentration) on temperature As Fig 4 shows the material under consideration follows some

general trends of n type MOX In a first stage, with the increasing of the temperature over the

room value, the shallow donor bulk levels ionise more and more towards complete ionisation,

if possible Deeper donor levels will follow at higher temperatures In parallel with these

electronic processes other physical and chemical processes are activated by the increase of the

temperature So, the ambient oxygen adsorbs at the semiconductor surface, at the beginning

(below 150°C), as molecular ions and then (above 150°C), when O2 molecule dissociate, as

atomic ions by trapping conduction electrons In this way occupied surface levels appear that

are not available in the absence of the adsorbat The release of the electrons and the

desorption processes are necessarily occurring together (for a thorough description in relation

Accepted Manuscript

with sensing mechanisms [36, 37] and in relation with Hall Effect see [32]) Capturing free

charge carriers at the surface / interface of the grains results also in surface / interface

barriers / double barriers They significantly obstruct the electron drift in the material,

between the grains, being the reason for the observed conductance decrease The temperature

range addressed in Fig 4, that is the one interesting for In2O3as O3sensing material, displays

exactly the functional region where the trapping processes are prevailing over the donor

ionisation ones, with the total effect of increasing the MOX resistance (see also the

comments in §3.5 in connection with Fig.6)

At this point one comes back to the original results The first important output of the

experimental data is itself the value of effective electron mobility in undoped In2O3 gas

sensing films under operation conditions, not reported until now in the literature for none of

the known In2O3polymorphs (cubic and rhombohedral) Here, one has to additionally remark

that unexpected high values have been obtained For porous and granular layers deposited

through powder techniques such large values are not encountered in the literature [32] The

electron mobility of rh - In2O3is 2 - 5 times that of c - In2O3but one can not assign this to the

structural difference between both polymorphs Though the samples have been prepared and

deposited in similar conditions the few differences in the layer morphology underlined in §2.1

could also influence the ratio referred above The slowly increased porosity of the c - In2O3,

due to the formation of nano / micro agglomerates together with the reduced grain size are

strong reasons for decreasing the mobility of the charge carriers (see the discussions in the

next paragraph, §3.5)

Another significant result is the reduced spread (less than a factor 2) of the effective

mobility values in one gas exposure sequence (better seen on Fig 5) This means that the

investigated samples are well fitting the standard modelling procedures for gas sensors, which

ascribe the target gas effects to the charge carrier concentration only, considering the mobility

as a constant material parameter In these conditions one expects power law dependencies

Accepted Manuscript

[36] of the conductance on the analyte concentration They are evident in Fig 4 for the 150°C

and 200°C plots The results obtained with both In2O3 polymorphs operated at high

temperatures (270°C and 300°C) are deviating from this tendency showing some limiting

trends, to be explained below

As the experimental data are suggesting, there is a strong dependence of the recorded

baseline resistance / conductance on the oxygen content (obvious if comparing the

conductance plots in Fig 3) An enhanced oxygen concentration, like that existing in the

natural atmosphere, could drastically reduce the concentration of the free charge in the

conduction band of the metal oxide semiconductors limiting their sensitivity to oxidising

gases (the case in Fig 3, left panel) If the sensing material enters into accentuated depletion

then the chemisorption of oxidising gases at the semiconductor surface, requiring conduction

electrons, is hindered Indeed, the charge transfer from conduction band to the surface states

of the analyte is restricted and, in consequence, the sensitivity is either decreasing or

saturating This behaviour is obvious in all electron concentration and conductance graphs for

high temperature (270°C, 300°C) and O3concentration (see Fig 3 and Fig 4.)

The reversed mechanism holds for the influence of the humidity According to the

experimental evidence the water vapour induces the enhancement of the sample conductance

and charge carrier concentration, acting against the oxygen effect This could simply be a

competition with the oxygen for the same adsorption sites resulting in a decrease in the O2

coverage degree of the sensing layer surface or a more complicated process (spectroscopic

information, not available yet, is required to discern the nature of the adsorbed species [24,

25]) Independent of the interaction details the result is the same: fewer trapped electrons at

the surface and lower depletion level of the grains In consequence the power law dependence

is recovered (conductance panel in Fig 4, plot for 50% r.h.)

The dependency of the electro-kinetic parameters (normalised or not) on the O3

concentration follows the trends for the oxygen, both O2 and O3 being oxidising gases One

Accepted Manuscript

easily observes that there is a strong charge carrier concentration decrease with the increase of

the ozone content in the test mixture This is due to the acceptor character of the surface state

induced by the O3 chemisorption, states which are able to trap significant amounts of

electrons from the conduction band [37] increasing , in this way, the heights of the

intergranular barriers The process is similar to the oxygen adsorption referred before As

Fig 5 indicates, the normalised conductance and normalised electron concentration (differing

up to a factor 1.5 introduced by the normalised mobility), are evolving over 2 order of

magnitude in dry air and over one order in humid air (for the O3 concentration range of 0 –

1000 ppb) The mobility however increases with a factor of 1.5 or less Such a small variation

does not allow deciding the cause without doubt Two processes could be at the origin of this

behaviour: the decrease in the rates and strengths of the electron scattering at the grain

interfaces, which are less crowded on the relatively small contact areas when the electron

concentration diminishes and / or the reduction of the bulk electron - electron scattering due to

the same decrease in the electron number participating at the conduction

In any case, the deconvolution procedure of the conductivity in two disproportioned

terms, as effective electron concentration and effective electron mobility are, can result itself

in slightly altered results as already accounted for in § 3.3

The humidity influence on the mobility seems to stem, through the mechanisms

addressed above, from its influence on the electron concentration (increase with mobility)

and, in turn, from the influence of the electron concentration on mobility leading to the

observed decrease

3.5 Consequences of the microscopic peculiarities on the electrical parameter values / dependencies and gas sensing properties

The use of nanosized materials in the sample fabrication has positive influence on its

sensitivity because of increased free surface / accessible interface areas The nano-granular

morphology, however, brings complications in the understanding and modelling of the

Accepted Manuscript

sensing layer response towards the target gases As already shown in literature [31, 32, 35,

36], the interplay of the grain size (l grain ), mean free path (l), and Debye length ( L ) controls D

the amount of the free charge still available in each grain, the depletion level of the grains and

the height of the associated Schottky barriers For the studied films these parameter have been

estimated by using the relations (2), (3) of Ref [31] supposing no initial (room temperature)

trapped charge at the grain surface and associated band banding [32] In the case of c - In2O3

one has: l grain: 20 – 50 nm; : 0.5 – 1 nm; L : 10 – 20 nm while for rh - In D 2O3: l grain: 50 –

100 nm; : 1-5 nm; L : 20-40 nm D

From this data it results that there always are more than 10 collisions of the charge

carriers inside each grain (l grain) and therefore a drift mobility, directly proportional to

the collision relaxation time, can be defined for both materials In such conditions, with strong

electron scattering in the bulk, the surface influence on the mobility should be relatively

reduced, as actually observed Not equally simple and likely in all investigated cases is the

situation of the Debye length, where values exceeding the half of the grain size are possible

for both polymorphs (small grains in combination with large Debye lengths) This means that

some grains almost reach the full depletion (flat band condition) Indeed, in the Schottky

approximation [45, 46] of full ionised donors, the width (w ) of the depleted region is given

by [46]:

 e V B k B T  L D

where the notation are the before explained ones and V , the barrier height B

The considered approximation actually means that the shallow donors are already all

ionised at the lowest working temperature and that the further increase of the temperature will

only favour the O2dissociation and Oxygen adsorption at the grain surface / interface with the

consequences described above: electron trapping on surface states, enlargement of the