CrystEngComm

Publ

ishe

d on

31

Mar

ch 2

014.

Dow

nloa

ded

by U

nive

rsity

of

Illin

ois

at C

hica

go o

n 19

/06/

2014

05:

22:0

4.

PAPER View Article OnlineView Journal | View Issue

6080 | CrystEngComm, 2014, 16, 6080–6088 This journal is © The R

a Thin Films Technology Laboratory, Department of Physics, COMSATS Institute of

Information Technology, Islamabad, Pakistan. E-mail: aminislamabad@gmail.com,

nabbasqureshi@yahoo.com; Tel: +92 321 5105363bCentre for Micro & Nano Devices, Department of Physics, COMSATS Institute of

Information Technology, Islamabad, Pakistan. E-mail: asbhatti@comsats.edu.pkc School of Materials, The University of Manchester, Manchester, UK.

E-mail: azad.malik@manchester,ac.uk

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ce00153b

Cite this: CrystEngComm, 2014, 16,

6080

Received 21st January 2014,Accepted 31st March 2014

DOI: 10.1039/c4ce00153b

www.rsc.org/crystengcomm

Effects of Mg doping on optical and CO gassensing properties of sensitive ZnO nanobelts†

Muhammad Amin,a Nazar Abbas Shah,*a Arshad Saleem Bhattib

and Mohammad Azad Malikc

We report the synthesis, optical characterization and enhanced carbon monoxide (CO) gas sensing

properties of magnesium (Mg) doped 1D zinc oxide (ZnO) nanobelts obtained via a vapor transport

method. The structural, morphological and compositional properties of the samples were investigated

by powder X-ray diffraction (p-XRD), field emission scanning electron microscopy (FESEM) and energy

dispersive X-ray (EDX) analysis. Optical characterization was carried out using Raman spectroscopy,

Photoluminescence (PL), diffuse reflectance spectroscopy (DRS), UV sensing and CO gas sensing.

Crystalline nanobelts were obtained with average thickness of about 34 nm, width of 290 nm, and length

of 3.25 μm. Significant changes in the bandgap energy due to Mg doping were observed. The gas sensing

properties of undoped and Mg doped ZnO nanostructures were tested based on the resistance change

upon exposure to air and CO gas. The ability of Mg doped ZnO nanobelts to sense 20 ppm of CO gas at

350 °C was enhanced fivefold, with good stability, indicating that Mg doping is very effective in improving

the CO sensing of ZnO nanobelts. In addition, a model that describes the CO gas sensing mechanism of

both undoped and Mg doped ZnO nanostructures is presented.

1 Introduction

1D oxide semiconductor nanocrystals are considered to behighly efficient materials for gas sensing, due to their highsensitivity to surface chemical reactions, rapid response andhigh surface to volume ratio in comparison to thin film gassensors. Consequently, the synthesis of 1D oxide semiconduc-tor nanostructures with different morphologies and sizes is ofsignificant importance for fundamental research and noveldevice development.1–3 Among several excellent metal oxidenanomaterials, ZnO is well suited for a number of applica-tions, owing to its characteristics, such as a direct and widebandgap (3.37 eV), high exciton binding energy (60 meV) atroom temperature, simple fabrication and good biocompati-bility.4,5 ZnO nanostructures are therefore considered as oneof the most promising gas sensing materials, due to their highsensitivity to toxic and combustible gases, carrier mobility,

and good chemical and thermal stability at moderately hightemperatures.6 Various techniques have been used to improvetheir response, reaction speed, and stability.7,8 However,increasing the sensing response and detection limit of 1DZnO nanobelts still remains a challenge. Many techniques,such as doping,9 fabrication of heterostructures,10 systematicmorphology control,11 and functionalization,12–14 have beenused to improve the stability, sensitivity response and recov-ery speed of 1D nanostructure based sensors. Doping is con-sidered to be an effective method for improving the gassensing properties of ZnO nanostructures at low temperature.Various researchers have used doped ZnO nanostructures forthe detection of CO, using different techniques. For example,Gaspera et al.15 showed that doping ZnO with transition metalions lowers the CO detection limit by 1–2 ppm at 300 °C. Mgdoped ZnO has been intensively studied due to its band gapengineering effect. One of the important features of ZnO isthat doping with Mg increases its bandgap.16 C. Jin et al.17

demonstrated that Mg doped ZnO nanowire sensors showed amaximum response of 1.04 (4.65%) for 100 ppm of CO gas at100 °C. Although Mg doped ZnO nanostructures have beenutilized for a number of applications and there are reports onthe synthesis of Mg doped ZnO nanostructures,18,19 to thebest of our knowledge, there are no reports on the synthesisof very thin transparent Mg doped ZnO nanobelts via a vaportransport method using magnesium acetate as the source

oyal Society of Chemistry 2014

CrystEngComm Paper

Publ

ishe

d on

31

Mar

ch 2

014.

Dow

nloa

ded

by U

nive

rsity

of

Illin

ois

at C

hica

go o

n 19

/06/

2014

05:

22:0

4.

View Article Online

material and their application as CO gas sensors with enhancedsensing response.

In this research paper, we report the successful synthesisof very thin transparent Mg doped ZnO nanobelts with anaverage thickness of 34 nm using a simple vapor transportmethod, vapor liquid solid (VLS), for enhanced CO gas sens-ing. The effects of Mg doping on the structural, optical, mor-phological and gas sensing properties of the ZnO nanobeltswere also investigated. Toxic gases can affect human life andhealth, even at levels of a few parts per million, and there-fore, highly sensitive gas sensors are required. CO gas istasteless, odorless and toxic, and is not easily detected byhumans. Therefore, it is essential to develop a sensitive COgas sensor to protect humans from exposure to CO gas.

2 Experimental details2.1 Synthesis of Mg doped ZnO nanobelts

A vapor transport method was used for the synthesis ofundoped and Mg doped ZnO nanostructures, as shown inFig. 1(a). Two experiments were performed under the sameconditions for undoped and Mg doped ZnO nanostructures.The ion sputtering technique was used for deposition of athin catalyst layer (1 nm) of gold (Au) on Si(100) substrates.Source materials, containing a mixture of 99.99% pure ZnO(0.5 g) and 99.9% graphite powder (0.5 g) for undoped ZnO,and a mixture of 99.99% pure ZnO (0.5 g), 99.9% graphite pow-der (0.5 g) and 0.1 g magnesium acetate [Mg(CH3COO)2·4H2O]for Mg doped ZnO, were placed in a ceramic boat at the centreof a quartz tube (length 100 cm and diameter 3.5 cm). Thesubstrates were placed on another ceramic boat downstreamof the source material. The quartz tube was then mounted

This journal is © The Royal Society of Chemistry 2014

Fig. 1 Schematic diagram of the experimental setup. (a) Synthesis of Zngas sensing setup. Inset shows the sensor and holder. (d) Optical fluorescenAu electrodes.

inside a horizontal tube furnace. The temperature of the fur-nace was maintained at 900 °C for 60 min under an argon (Ar)atmosphere with a constant flow rate of 50 sccm for the twoexperiments. The other end of the quartz tube was connectedto a flexible tube (diameter 1.0 cm and length 100 cm) and waskept open. Graphite powder was used to lower the evaporationtemperature of ZnO (1975 °C). On introducing graphite powderinto the ZnO precursor, a carbothermal reaction results in theformation of Zn and ZnOx (sub-oxides) vapors at 900–1000 °C,as shown in the equations.20

ZnO + C → Zn + CO (1)

ZnO + CO → Zn + CO2 (2)

ZnO + (1 − x)C → ZnOx + (1 − x)CO (x < 1) (3)

ZnO + (1 − x)CO → ZnOx + (1 − x)CO2 (4)

After the reaction, the furnace was cooled down to roomtemperature. The collected nanostructure samples were charac-terized by FESEM, EDX and X-ray diffraction (XRD, X'Pert PRODiffractometer, PANalytical with Cu Kα radiation, λ = 1.5418 Å) inorder to investigate the morphology and crystal structure. Ramanspectroscopy (Renishaw, 514.5 nm laser), photoluminescence(PL, DONGWOO Optron, 488 nm Ar laser) and EDX were used tostudy the Mg doping and the chemical composition of the ZnOnanobelts. UV-vis diffuse reflectance spectroscopy (DRS, lambda950 UV-vis spectrophotometer) was utilized to measure the bandgapenergy of the nanostructures. CO gas sensing experimentswere carried out by measuring the respective resistances witha two probe method using a multimeter (Keithly 2100).

CrystEngComm, 2014, 16, 6080–6088 | 6081

O nanostructures. (b) CO gas sensing setup. (c) Actual image of thet image of the nanobelt sensor. (e) Microscopic image of interdigitated

CrystEngCommPaper

Publ

ishe

d on

31

Mar

ch 2

014.

Dow

nloa

ded

by U

nive

rsity

of

Illin

ois

at C

hica

go o

n 19

/06/

2014

05:

22:0

4.

View Article Online

2.2 Gas sensor fabrication and characterization

The CO gas sensors were fabricated by thermally depositing a200 nm thick layer of SiO2 onto single crystalline Si(100) wafers.Slurry droplets containing Mg doped and undoped ZnO nano-belts (20 μL) were dropped onto SiO2 coated Si substratesequipped with a pair of interdigitated Au (200 nm) electrodeswith a gap of 50 μm, using the photolithography technique.These sensors were heat treated in air for 4 hours at 400 °Cbefore the gas sensing experiments were performed. The sens-ing experiments were performed at 350 °C with 5 min cyclesof dry air and 20 ppm CO gas. The sensing response (S = Ra/Rg)of the device was measured in a self-designed gas chamberconnected to a Keithly multimeter, gas flow meters and thetube furnace, using the resistance changes upon exposure toair (Ra) and CO gas (Rg). The schematic design and the gassensing characterization setup are shown in Fig. 1(b) & (c).The inset of Fig. 1(c) shows the sensor holder with pressurecontacts. An optical fluorescent image of the nanobelt sensorand a microscopic image of the interdigitated Au electrodesare shown in Fig. 1(e) & (d), respectively.

3 Results and discussion3.1 Morphology and microstructure analysis

The general morphology of the synthesized undoped and Mgdoped ZnO nanostructures was investigated using FESEM, TEMand high resolution TEM. Fig. 2(a) shows undoped ZnO nano-rods with uniform diameter and length, having an averagediameter of 168 ± 30 nm and an average length of 2.4 ± 1.2 μm.The inset of Fig. 2(a) shows a magnified image of the undopedZnO nanorods. Fig. 2(e) shows the corresponding EDX analysisof the undoped ZnO nanorods, clearly showing the Zn andO peaks. The approximate atomic ratio was found to be66.61 : 33.39. Fig. 2(b) shows an FESEM image of Mg dopedZnO nanobelts. The inset of Fig. 2(b) shows a magnifiedimage of the Mg doped ZnO nanobelts. The general morphol-ogy for all the nanobelts is the same, i.e. transparent, smoothedges, uniform width along the entire length, and rectangularcross section. The average thickness, width and length of thenanobelts are 34 ± 5 nm, 290 ± 160 nm and 3.25 ± 1.51 μm,respectively. The corresponding EDX analysis of the Mg dopedZnO sample (Fig. 2(f)) shows the presence of oxygen, mag-nesium and zinc, and the O :Mg : Zn ratio was found to be30.79 (O) : 3.43 (Mg) : 65.78 (Zn). The aspect ratios of theundoped and Mg doped ZnO nanostructures were found to be14.3 and 8.5 (W/T), respectively. The FESEM micrographs clearlyshow that the morphology changes from nanorods to nano-belts on Mg doping. A possible reason for the morphologytransition of ZnO from nanorods to nanobelts on Mg doping isthe fact that mixing/doping/alloying of specific elements playsa major role in modifying the dimensions of nanostructures.21,22

ZnO nanobelts are formed in a continuous ‘1D branching andsubsequent 2D interspace filling’ process.23 For group II–VIsemiconductors with a wurtzite crystal structure, the characteristic

6082 | CrystEngComm, 2014, 16, 6080–6088

polar surfaces can induce asymmetric growth, leading to theformation of unique nanostructures.24

Fig. 2(c) & (d) show high resolution TEM (HRTEM) imagesof undoped and Mg doped ZnO nanostructures, respectively.The inset TEM image of Fig. 2(c) clearly shows the catalyst par-ticles (Au) on the tips of the nanorods, suggesting VLS growth.The HRTEM image (Fig. 2(c)) shows the lattice fringes of theundoped crystalline nanostructures, whereas these fringes areblurred in the Mg doped nanobelts (Fig. 2(d)). The correspond-ing selected area electron diffraction (SAED) patterns shown inthe insets of Fig. 2(c) & (d) are consistent with the HRTEMobservations. These results confirm the defect free growth ofundoped ZnO nanorods. Similar results were reported byJ. Singh et al. for magnesium doped ZnO nanowires.18

3.2 XRD and Raman spectroscopic analyses of Mg doped ZnOnanostructures

Fig. 3(a) shows the XRD patterns of undoped and Mg dopedZnO nanostructures. All the peaks match the hexagonal phaseof ZnO (ICDD PDF-2 entry 01-079-0207). The calculated latticeparameters (a = 3.2650 Å and c = 5.2278 Å) are comparable tothose of pure ZnO (a = 3.2533 Å and c = 5.2072 Å). No diffrac-tion peaks due to impurities were observed in the XRD pat-terns. Slight shifts of the peak positions were observed for thedoped samples, as compared to the undoped samples, asshown in the inset of Fig. 3(a), which indicates that Mg dop-ing induces lattice strain in the ZnO nanobelts. The c latticeparameters calculated from the (002) plane were found to be5.2278 Å and 5.1840 Å for the doped and undoped samples,respectively. The decrease (ca. 0.0438 Å/0.83%) in this latticeparameter indicates lattice compression along the c-axis25

caused by the replacement of Zn by Mg, which has a smalleratomic radius (0.57 Å) as compared to Zn (0.60 Å).26,27

A Raman spectroscopic analysis at 514 nm excitationwavelength was performed in order to investigate the vibra-tion properties of the undoped and Mg doped ZnO nano-structures, as shown in Fig. 3(b). There is a significantdifference between the two spectra. Peaks at 338, 436 and579 cm−1 were observed for undoped ZnO nanorods. Thepeak at 338 cm−1 is attributed to the A1 symmetry mode, thepeak at 436 cm−1 corresponds to the ZnO non-polar opticalphonon E2 (high) mode, and the peak at 579 cm−1 is assignedto the E1 (LO) mode, which is due to the formation of defects,such as oxygen vacancies.28 The Mg doped ZnO nanobeltsshowed a shift of the signals to 335 cm−1 and 438 cm−1, ascompared to the pure ZnO. The peak at 438 cm−1 is blueshifted with weak intensity, and is broadened for the Mgdoped ZnO. Usually, Raman peak shifts occur for three rea-sons: phonon confinement effects,29 lattice strain,30 and oxy-gen vacancies.31 The broadening of the E2 (high) mode peakcan be attributed to the size effect and residual stress in theMg doped nanobelts.32 Therefore, in our case, the red shift ofthe Raman active E1 (LO) and E2 (high) modes of the Mgdoped ZnO nanobelts are strong evidence for successful Mgdoping in the ZnO nanobelts.

This journal is © The Royal Society of Chemistry 2014

Fig. 2 (a) FESEM image of undoped ZnO nanorods; inset shows a magnified image. (b) FESEM image of Mg doped transparent ZnO nanobelts;inset shows a magnified image. (c) and (d) show HRTEM images of undoped nanorods and Mg doped nanobelts, respectively; insets of (c) and(d) show magnified TEM images and the corresponding SAED images. (e) and (f) show EDX analyses of the undoped ZnO nanorods and theMg doped ZnO nanobelts, respectively.

CrystEngComm Paper

Publ

ishe

d on

31

Mar

ch 2

014.

Dow

nloa

ded

by U

nive

rsity

of

Illin

ois

at C

hica

go o

n 19

/06/

2014

05:

22:0

4.

View Article Online

3.3 Photoluminescence spectroscopy

Fig. 4(a) & (b) show room temperature PL spectra with He–Cdlaser excitation at 488 nm. Both undoped and Mg doped ZnOshow broad green emission bands, centered at 527 nm and530 nm, respectively. The PL results clearly show shifts in the

This journal is © The Royal Society of Chemistry 2014

Gaussian fittings of the respective bands, and new bands at598 nm and 647 nm are present for Mg doped ZnO nanobelts.Green emission is generally attributed to various intrinsicdefects produced during the synthesis of ZnO nanostruc-tures.33 The presence of surface defects, such as oxygen vacan-cies (Vo), Zn vacancies (VZn), interstitial zinc (Zni) and antisite

CrystEngComm, 2014, 16, 6080–6088 | 6083

Fig. 3 (a) X-ray diffraction analysis of undoped and Mg doped ZnO nanostructures. The inset figure clearly shows slight shifts of the peak positionsfor the doped samples. (b) Raman spectra of undoped ZnO nanorods and Mg doped ZnO nanobelts.

Fig. 4 Room temperature PL spectra of (a) undoped ZnO nanorods deconvoluted into three Gaussian peaks, and (b) Mg doped ZnO nanobeltsfitted with four Gaussian peaks.

CrystEngCommPaper

Publ

ishe

d on

31

Mar

ch 2

014.

Dow

nloa

ded

by U

nive

rsity

of

Illin

ois

at C

hica

go o

n 19

/06/

2014

05:

22:0

4.

View Article Online

defects (OZn), have been reported by various researchers, andcontribute to the green emission of ZnO.34–36 In our experi-ment, an interesting phenomenon was observed, in that theintensity of the visible emission band of the Mg doped ZnOnanobelts is much stronger than that of the undoped ZnOnanorods. This can be attributed to the increase in the con-centration of oxygen vacancies and surface recombinationeffects in the thin nanobelt system.37,38

3.4 Diffuse reflectance spectroscopy

The optical properties of undoped ZnO and Mg doped ZnOnanobelts were investigated using UV-visible diffuse reflectancespectroscopy. Fig. 5(a) & (b) show plots of F2(R) vs. photon energy

6084 | CrystEngComm, 2014, 16, 6080–6088

for the synthesized nanostructures. F(R) is the Kubelka–Munkfunction, which is given by the relationship:39

F(R) = ((1 − R)2/2R) = (α/S) (5)

where R, α, and S are the diffuse reflection, absorption andscattering coefficients, respectively. The bandgap can bedefined by extrapolating the linear parts of the plots to thephoton energy axes.40 The optical bandgap was calculated,and was found to increase from 3.18 eV for undoped ZnOnanorods to 3.32 eV for Mg doped ZnO nanobelts, becauseMgO has a wider bandgap than ZnO.41 Hsu et al. reportedthat doping with Mg increases the band gap of ZnO.16,42 Thisincrease in bandgap might be due to an increase in carrier

This journal is © The Royal Society of Chemistry 2014

Fig. 5 Plots of F2(R) vs. energy (eV) for (a) undoped ZnO nanorods and (b) Mg doped ZnO nanobelts. The inset figures show the respectivereflectance spectra. (c) Schematic representation of the energy transition mechanism in the undoped and Mg doped ZnO nanostructures. (d) UVsensing responses of Mg doped ZnO nanobelts and undoped ZnO nanorods.

CrystEngComm Paper

Publ

ishe

d on

31

Mar

ch 2

014.

Dow

nloa

ded

by U

nive

rsity

of

Illin

ois

at C

hica

go o

n 19

/06/

2014

05:

22:0

4.

View Article Online

concentration, which blocks the lowest states in the conduc-tion band; this effect is known as the Burstein–Moss effect.40

A schematic diagram of the energy transition mechanismfrom the excitonic and defects states in the undoped and Mgdoped ZnO nanostructures is shown in Fig. 5(c). The exci-tonic transition energy was blue shifted to a high energyvalue due to Mg doping in ZnO. During the formation of theMg : ZnO structure, Mg2+ ions were substituted for the Zn2+

ions without changing the ZnO structure. However, thedoped Mg in the ZnO acts as donor, and the bandgap energyof the structure increases.43 The optical properties of the syn-thesized ZnO nanostructures are in agreement with the XRDand EDX results.

3.5 UV sensing

ZnO nanostructures are good materials for ultraviolet lightsensors due to the wide bandgap (UV range for ZnO is about368–390 nm).44 Fig. 5(d) shows the room temperature UVsensing results for the Mg doped and undoped ZnO nano-structures. The sensing responses of the Mg doped ZnOnanobelts and undoped nanorods were found to be 2.32 and

This journal is © The Royal Society of Chemistry 2014

1.13, respectively. A twofold enhanced sensing response wasobserved for the Mg doped ZnO nanobelts. The resistancedecreases with UV light (wavelength ranges from 315 to365 nm with an 18 W UV Phillips Lamp) illumination andincreases again when the UV light is switched off. When theZnO nanostructure sensor is exposed to air, a negative spacecharge layer is created on the surface, and the adsorbed oxy-gen molecule captures an electron from the conduction band(sensor exhibits higher resistivity). When the photon energyis greater than the bandgap energy, Eg, radiation is absorbedby the nanostructure sensor, creating an electron–hole pair.The photo-generated, positively-charged hole neutralizes thechemisorbed oxygen responsible for the higher resistance,increasing the conductivity of the device. As a consequence,the conductivity of the material increases, giving rise to aphotocurrent. This process continues in a cyclic manner onswitching the UV light on and off. The enhanced response inthe case of the Mg doped ZnO nanobelts was attributed tothe energy levels introduced by the dopant in the bandgapand in the conduction band of ZnO. These states acted as“hopping” states, and increased the probability of an electronbeing excited to the conduction band.45

CrystEngComm, 2014, 16, 6080–6088 | 6085

CrystEngCommPaper

Publ

ishe

d on

31

Mar

ch 2

014.

Dow

nloa

ded

by U

nive

rsity

of

Illin

ois

at C

hica

go o

n 19

/06/

2014

05:

22:0

4.

View Article Online

3.6 Characterization of CO gas sensors

Fig. 6(a) & (b) show the sensing response signals of the dopedand undoped ZnO nanostructures. Previous reports showedthat the response of resistive sensors is highly affected by theoperating temperature. Therefore, in order to optimize theoperating temperature of Mg doped ZnO nanobelts, the sensorswere tested at different temperatures, ranging from 200 °C to400 °C for 20 ppm of CO gas, as shown in Fig. 6(d). It wasobserved that the sensing response increases with increasingworking temperature, reaches a maximum at about 350 °C,and then starts decreasing. At low temperature, the CO mole-cules are not activated enough to react with the surfaceadsorbed oxygen species. Above 350 °C, the decrease in COgas adsorption is not adequately compensated by the increasein surface reactions, and the sensor response decreases. Thisbehavior may be interpreted on the basis of adsorption/desorption and reaction processes taking place on the surfaceof the sensing layer.46 The sensing experiment was performedat 350 °C with 5 min cycles of dry air and 20 ppm CO gas. Theresistance decreased upon exposure to CO and completelyrecovered to the initial value upon removal of CO. The sens-ing responses of the undoped nanorods and Mg doped ZnOnanobelts were found to be 1.05 and 5.5, respectively. A

6086 | CrystEngComm, 2014, 16, 6080–6088

Fig. 6 Signal response and recovery profiles of (a) undoped ZnO nanomechanism of how the Mg doped ZnO nanobelt sensor responds to COresponse increases until 350 °C, and then decreases.

fivefold enhanced sensing response was observed for the Mgdoped ZnO nanobelts. The response and recovery times of thetwo sensors were nearly equal, as shown in Fig. 6(a) & (b). Mgdoped ZnO nanobelts are capable of adsorbing large amountsof oxygen due to their larger surface area (greater numberof defects), which increases the chance of ZnO interactingwith CO gas and results in fast chemisorption/desorption ofCO gas at 350 °C, as compared to undoped ZnO nanostruc-tures.47,48 N. Hongsith et al. reported that the sensor responseis proportional to the reaction rate constants kCO (T) andkoxy (T), through the oxygen density, which is given by the fol-lowing equation:49

S = Ra/Rg = (ΓtkCO(T)koxy(T)[Oionads]

b[CO]b)/n0 + 1, (6)

where Ra = resistance in air, Rg = resistance in gas, Γt = timeconstant, [Oion

ads] = chemisorbed oxygen concentration, [CO] =CO concentration, b = charge parameter and n0 = carrier con-centration in air.

3.7 Sensing mechanism: band theory model

In order to explain the sensing mechanism (shown in Fig. 6(c)),the band theory was applied to gas sensors, which have beenthe subject of intense study for a number of years.50,51 The

This journal is © The Royal Society of Chemistry 2014

rods and (b) Mg doped ZnO nanobelts. (c) Schematic of a possiblein air and in gas. (d) The response–temperature curve shows that the

Table 1 Summary of the results

Morphology ofnanostructures

Synthesistemp. (°C)

Band gap byUV-vis (eV)

Nanostructuresize

Aspectratio

Response to 20 ppm ofCO gas at 350 °C

PL peakpositions (nm)

Raman peakpositions (cm−1)

Nanorods (undoped) 900 3.20 Dav = 168 nm,Lav = 2.4 μm

14.3 1.05 527.4, 558.1 338.8, 435.5,576.5

Nanobelts (Mg doped) 900 3.33 Tav = 34 nm,Lav = 3.25 μm,Wav = 290 nm

8.5 (W/T) 5.41 530.2, 567.2,598.2, 647.7

335.6, 437.7,553.2

CrystEngComm Paper

Publ

ishe

d on

31

Mar

ch 2

014.

Dow

nloa

ded

by U

nive

rsity

of

Illin

ois

at C

hica

go o

n 19

/06/

2014

05:

22:0

4.

View Article Online

sensing mechanism is based on the change in electrical resis-tivity/conductivity as a result of a chemical reaction betweengas molecules and the reactive oxygen ions on the surface ofthe ZnO nanostructures. The change in electrical conductanceis given by eqn (7):52

G

Re rl

1 0

2( | | )n (7)

where l is the length of the nanostructure channel, r is theradius, μ is the electron mobility, e is the electron charge, andΔn0 represents the change in carrier concentration. When thesurface of the undoped and Mg doped ZnO nanostructures isexposed to the air, the oxygen that takes part in detection isconsequently ionized into dynamic oxygen species (O2−, O2

−

and O−), while moving from site to site by capturing electronsfrom the active surface sites of undoped and Mg doped ZnOnanobelts.53,54 At high temperature, reactive oxygen speciesare chemisorbed by the ZnO nanostructures, and electrontransfer takes place as a result. When the adsorption reachesa certain level, a thick depletion layer is formed, which causesa decrease in the carrier concentration, resulting in increasedresistance of the material. In contrast, when undoped and Mgdoped ZnO are exposed to CO gas, a chemical reaction takesplace, as the adsorbed oxygen ions produce CO2, which leadsto an increase in the carrier concentration and a decrease inthe resistance.55 The mechanism can be summarized by thefollowing chemical reactions.

O2(gas) → O2(ad) (8)

O2(ad) + e− → O2(ad)− (9)

O2(ad)− + e− → 2O(ad)

− (10)

2CO + O2(ad)− + e− → 2CO2(gas) + e− (11)

CO + 2O− → CO32− → CO2 + 1/2O2 + 2e− (12)

Table 1 shows a systematic summary of the results. Themorphology of the synthesized ZnO nanostructures changeson magnesium doping. The PL and Raman peak positionsare shifted, and new peaks are observed, which confirmsthat Mg has been doped into ZnO. The bandgap of theZnO nanostructures increases due to defects states resultingfrom Mg doping, and hence, the CO gas sensing response

This journal is © The Royal Society of Chemistry 2014

increases about fivefold, as compared to the undoped ZnOnanostructures.

Conclusion

Magnesium doped ZnO nanobelts were successfully synthe-sized via a vapor transport method using magnesium acetateas the source material. A clear change in the morphology wasobserved, from ZnO (nanorods) to Mg doped ZnO (nanobelts).Shifts of the XRD peaks of the Mg doped samples, as com-pared to ZnO, indicated that Mg has substituted Zn in theZnO lattice. The doping was also confirmed by EDX, photo-luminescence and Raman spectroscopy. The gas sensor fabri-cated from multiple networks of the Mg doped ZnO nanobeltsshowed an enhanced signal response to 20 ppm of CO gas at350 °C. The gas sensing response of the Mg doped ZnO nano-belt sensor was improved fivefold in comparison to theundoped ZnO nanorod sensor. The enhanced signal responsewas attributed to the catalytic effect of the dopant, defectssuch as oxygen vacancies (potential barrier modification),and fast chemisorption and desorption of CO gas. The bandtheory model was adopted to explain the possible sensingmechanism of the Mg doped ZnO nanostructures.

Acknowledgements

The authors would like to thank the Higher EducationCommission (HEC) of Pakistan for the financial supportthrough the “National Research Program for Universities”and for the award of Indigenous 5000 PhD scholarship pro-gram batch-IV. We would also like to thank the COMSATSInstitute of Information Technology, Islamabad for providingexcellent research facilities, and the Institute of IndustrialControl Systems. The authors are grateful to Dr Ahmer Naveed(Centre for Micro & Nano Devices, Department of Physics,COMSATS Institute of Information Technology, Islamabad,Pakistan) and Rizwan-ur-Rehman Sagar (School of MaterialsScience and Engineering, Tsinghua University, Haidian Beijing,China) for useful discussions.

References

1 C. Liangyuan, L. Zhiyong, B. Shouli, Z. Kewei, L. Dianqing,

C. Aifan and L. C. Chiun, Sens. Actuators, B, 2010, 143,620–628.

2 A. Kolmakov, Y. Zhang, G. Cheng and M. Moskovits, Adv. Mater.,

2003, 15, 997–1000.

CrystEngComm, 2014, 16, 6080–6088 | 6087

CrystEngCommPaper

Publ

ishe

d on

31

Mar

ch 2

014.

Dow

nloa

ded

by U

nive

rsity

of

Illin

ois

at C

hica

go o

n 19

/06/

2014

05:

22:0

4.

View Article Online

3 Y. Liu, E. Koep and M. Liu, Chem. Mater., 2005, 17, 3997–4000.

4 X. J. Huang and Y. K. Choi, Sens. Actuators, B, 2007, 122,

659–671.5 Y. Qiu and S. Yang, Adv. Funct. Mater., 2007, 17, 1345–1352.

6 T. Gao and T. H. Wang, Appl. Phys. A: Mater. Sci. Process.,

2005, 80, 1451–1454.7 R. C. Wang and H. Y. Lin, Sens. Actuators, B, 2010, 149, 94–97.

8 V. Kobrinsky, E. Fradkin, V. Lumelsky, A. Rothschild, Y. Komem

and Y. Lifshitz, Sens. Actuators, B, 2010, 148, 379–387.9 Q. Wan and T. H. Wang, Chem. Commun., 2005, 3841–3843.

10 X. Xue, L. Xing, Y. Chen, S. Shi, Y. Wang and T. Wang,

J. Phys. Chem. C, 2008, 112, 12157–12160.

11 M. Amin, U. Manzoor, M. Islam, A. S. Bhatti and N. A. Shah,

Sensors, 2012, 12, 13842–13851.

12 A. Kolmakov, D. O. Klenov, Y. Lilach, S. Stemmer and

M. Moskovits, Nano Lett., 2005, 5, 667–673.

13 Q. Kuang, C. S. Lao, Z. Li, Y. Z. Liu, Z. X. Xie, L. S. Zheng

and Z. L. Wang, J. Phys. Chem. C, 2008, 112, 11539–11544.

14 A. Wei, L. Pan and W. Huang, Mater. Sci. Eng., B, 2011, 176,

1409–1421.

15 E. D. Gaspera, M. Guglielmi, G. Perotto, S. Agnoli, G. Granozzi,

M. L. Post and A. Martucci, Sens. Actuators, B, 2012, 161,675–683.

16 H. C. Hsu, C. Y. Wu, H. M. Cheng, Z. Y. Wang, J. W. Zhao

and L. D. Zhang, Appl. Phys. Lett., 2006, 89, 013101–013103.

17 C. Jin, S. Park, H. Kim, S. An and C. Lee, Bull. Korean Chem.

Soc., 2012, 33, 1993–1997.

18 J. Singh, P. Kumar, K. S. Hui, K. N. Hui, K. Ramam,

R. S. Tiwaria and O. N. Srivastava, CrystEngComm, 2012, 14,5898–5904.

19 L. Hang, Z. Hui-zhao, X. Cheng-shan, W. Jie and L. Jun-lin,

Semicond. Photonics Technol., 2009, 15, 225–229.

20 N. Wang, Y. Cai and R. Q. Zhang, Mater. Sci. Eng., R, 2008,

60, 1–51.

21 M. Hafeez, U. Manzoor, A. S. Bhatti, B. Karnayar and S. I. Shah,

J. Appl. Phys., 2012, 111, 024313–024320.

22 H. J. Fan, B. Fuhrmann, R. Scholz, C. Himcinschi, A. Berger

and H. Leipner, Nanotechnology, 2006, 17, 231–239.

23 J. H. Park, H. J. Choi, Y. J. Choi, S. H. Sohn and J. G. Park,

J. Mater. Chem., 2004, 14, 35–36.

24 Z. L. Wang, X. Y. Kong and J. M. Zuo, Phys. Rev. Lett., 2003,

91, 185502.

25 F. Wang, C. Zhao, B. Liu and S. Yuan, J. Phys. D: Appl. Phys.,

2009, 42, 115411–115414.

26 H. Ryoken, N. Ohasahi, I. Sakaguchi, Y. Adachi, S. Hishita

and H. Haneda, J. Cryst. Growth, 2006, 287, 134–138.

27 U. Ozgur, Y. I. Alivov, C. Liu, A. Teke, M. A. Reshchikov,

S. Doan, V. Avrutin, S. J. Cho and H. Morkoc, J. Appl. Phys.,2005, 98, 041301–041403.

28 A. Umar, S. H. Kim, Y. S. Lee, K. S. Nahm and Y. B. Hahn,

J. Cryst. Growth, 2005, 282, 131–136.

29 K. F. Lin, H. M. Cheng, H. C. Hsu and W. F. Hsieh, Appl.

Phys. Lett., 2006, 88, 263117–263120.

6088 | CrystEngComm, 2014, 16, 6080–6088

30 A. Tiwari, M. Park, C. Jin, H. Wang, D. Kumar and J. Narayan,

J. Mater. Res., 2002, 17, 2480–2487.

31 P. Jiang, J. J. Zhou, H. F. Fang, C. Y. Wang, Z. L. Wang and

S. S. Xie, Adv. Funct. Mater., 2007, 17, 1303–1310.

32 C. Geng, Y. Jiang, Y. Yao, X. Meng, J. A. Zapien, C. S. Lee,

Y. Lifshitz and S. T. Lee, Adv. Funct. Mater., 2004, 14,589–594.

33 Y. Zhang, X. Song, J. Zheng, H. Liu, X. Li and L. You,

Nanotechnology, 2006, 17, 1916–1921.

34 K. Vanheusden, C. H. Seager, W. L. Warren, D. R. Tallant

and J. A. Voigt, Appl. Phys. Lett., 1996, 68, 403–405.

35 Y. W. Heo, D. P. Norton and S. J. Pearton, J. Appl. Phys.,

2005, 98, 073502–073505.

36 B. Lin, Z. Fu and Y. Jia, Appl. Phys. Lett., 2001, 79, 943–946.

37 I. Shalish, H. Temkin and V. Narayanamurti, Phys. Rev. B:

Condens. Matter Mater. Phys., 2004, 69, 245401–245404.38 S. A. Studenikin and M. Cocivera, J. Appl. Phys., 2002, 91,

5060–5065.39 M. Nagasawa and S. Shionoya, J. Phys. Soc. Jpn., 1971, 30,

158–167.40 E. Burstein, Phys. Rev., 1954, 93, 632–633.

41 D.-J. Qiu, H.-Z. Wu, N.-B. Chen and T.-N. Xu, Chin. Phys.

Lett., 2003, 20, 582–584.42 K. S. Ahn, Y. Yan, S. Shet, T. Deutsch, J. Turner and

M. Al-Jassim, Appl. Phys. Lett., 2007, 91, 231909–231912.43 S. C. Saime, U. Ibrahim, A. Aytimur and S. Ozcelik, Ceram. Int.,

2012, 38, 4201–4208.44 S. Hullavarad, N. Hullavarad, D. Look and B. Claflin,

Nanoscale Res. Lett., 2009, 4, 1421–1427.45 C. S. Lao, M. C. Park, Q. Kuang, Y. L. Deng, A. K. Sood,

D. L. Polla and Z. L. Wang, J. Am. Chem. Soc., 2007, 129,12096–12097.

46 T. Krishnakumar, R. Jayaprakash, N. Pinna, F. N. Donato,

A. Bonavita, G. Micali and G. Neri, Sens. Actuators, B, 2009,143, 198–204.

47 Y. Zhang, J. Q. Xu, Q. Xiang, H. Li, Q. Y. Pan and P. C. Xu,

J. Phys. Chem. C, 2009, 113, 3430–3435.

48 A. A. Ibrahim, G. N. Dar, S. A. Zaidi, A. Umar, M. Abaker,

H. Bouzid and S. Baskoutas, Talanta, 2012, 93, 257–263.

49 N. Hongsith, E. Wongrat, T. Kerdcharoen and S. Choopun,

Sens. Actuators, B, 2010, 144, 67–72.

50 F. George, L. M. Fine, A. A. Cavanagh and B. Russell,

Sensors, 2010, 10, 5469–5502.

51 S. R. Morrison, Sens. Actuators, 1981, 2, 329–341.

52 S. Osswald, M. Havel and G. Yury, J. Raman Spectrosc., 2007,

38, 728–736.53 G. N. Ara, A. Umarb, S. A. Zaidib, A. A. Ibrahimd, M. Abakera,

S. Baskoutasc and M. S. Al-Assiri, Sens. Actuators, B, 2012,173, 72–78.

54 N. Han, L. Chai, Q. Wang, Y. Tian, P. Deng and Y. Chen,

Sens. Actuators, B, 2010, 147, 525–530.

55 J. Li, H. Q. Fan and X. H. Jia, J. Phys. Chem. C, 2010, 114,

14684–14691.

This journal is © The Royal Society of Chemistry 2014