IOP PUBLISHING NANOTECHNOLOGY

Nanotechnology 23 (2012) 015705 (7pp) doi:10.1088/0957-4484/23/1/015705

Interaction of hydrogen with ZnOnanopowders—evidence of hydroxyl groupformationKester W J Wong1, Matthew R Field1, Jian Zhen Ou2,Kay Latham3, Michelle J S Spencer1,4, Irene Yarovsky1 andKourosh Kalantar-zadeh2

1 Applied Physics, School of Applied Sciences, RMIT University, Vic 3001, Australia2 School of Electrical and Computer Engineering, RMIT University, Vic 3001, Australia3 Applied Chemistry, School of Applied Sciences, RMIT University, Vic 3001, Australia4 Department of Chemistry, La Trobe Institute for Molecular Science, La Trobe University,Vic 3086, Australia

E-mail: kester.wong@rmit.edu.au and kourosh.kalantar@rmit.edu.au

Received 8 September 2011, in final form 31 October 2011Published 8 December 2011Online at stacks.iop.org/Nano/23/015705

AbstractThere have been many investigations to reveal the nature of the hydrogen gas and ZnOnanopowder interaction at elevated temperatures, while at present no conclusive description ofsuch an interaction has been confidently reported. In this work, we demonstrate that a hydroxylgroup is formed during this interaction, depending on size and relative crystallinity ofnanopowders. Our in situ Raman spectroscopy investigations show that the interaction directlyaffects the intensity of the Raman signal at 483 cm−1, relative to the peak at 519 cm−1. Ex situx-ray diffraction (XRD) and infrared spectroscopy also show extra peaks at 44◦ and 1618 cm−1,respectively, after hydrogenation. These peaks were all identified as surface hydroxyl groups,which can be related to the formation of water on the ZnO nanopowder surfaces.

S Online supplementary data available from stacks.iop.org/Nano/23/015705/mmedia

(Some figures may appear in colour only in the online journal)

1. Introduction

ZnO is a well-studied transparent conducting oxide materialwith a wide bandgap of 3.4 eV [9] and has mainly n-typeconductivity due to the large amount of oxygen vacancies andzinc interstitials in the surface [45]. It has been a subjectof intensive studies where it has shown promising results ina variety of applications such as dye-sensitized solar cells,piezoelectric, optoelectric and sensor devices [34, 28, 16, 26].ZnO is very versatile and can be fabricated into variousnanosized structures, and increasingly more novel andfunctional morphologies are being reported [41, 31, 10].

Hydrogen (H2) exists ubiquitously in our environment.It can interact with metal and metal oxides, where thematerials can be contaminated by molecular or dissociatedH2 which subsequently alters the physiochemical propertiesof the material [11, 37]. The investigation of hydrogen–

ZnO interaction is not uncommon and has been widelyreported [33, 2, 32, 14, 22].

Lavrov et al identified two shallow donors of H usingRaman spectroscopy, as bond-centred hydrogen (HBC) andhydrogen bound within the oxygen vacancy (HO) [15]. HBC

was found to be unstable at annealing temperatures above190 ◦C [15]. However, the HO donor was found to be stableat temperatures up to 500 ◦C [15]. It was also believed thatthese H shallow donors (typically found in hydrothermallygrown ZnO samples) are undetected in the sample usingIR spectroscopy, even though they have sufficient thermalstability to remain in ZnO for a month or longer at roomtemperature [33, 7]. Two O–H local vibrational modes(3326 and 3611 cm−1) were identified directly using IRabsorption spectroscopy when ZnO was annealed in H2 gasambient [23, 33]. Nonetheless, these two IR absorption linesare strongly dependent on the source of the ZnO material;

0957-4484/12/015705+07$33.00 © 2012 IOP Publishing Ltd Printed in the UK & the USA1

Nanotechnology 23 (2012) 015705 K W J Wong et al

therefore, shifted IR spectral lines are expected for differentZnO nanostructures. In order to completely remove hydrogenfrom ZnO, an annealing temperature of at least 500–700 ◦C isnecessary, depending on the ZnO source material [17, 33, 20].Atomic hydrogen present in ZnO can be reactive enoughto reduce the Zn polar surface of the ZnO(0001) materialinto metallic Zn clusters, yet can be almost inert on the O-polar ZnO(0001) surface [22]. When annealed at highertemperatures, the number of interstitials (Zni) accumulatingon the ZnO surfaces increases, leading to an abundant numberof oxygen vacancies (VO) in the oxide material. Zni, in thisinstance, act as a shallow donor on the surface [6].

The effects of hydrogen on ZnO, with differentmorphologies and stoichiometry, have been investigatedusing different methods [44, 36, 17]. However, conclusiveexperimental studies of the interaction of hydrogen with ZnOnanostructures are far fewer in number. It is suggestedthat hydrogen is much more susceptible to being trappedbetween, or adsorbed onto, ZnO nanoparticles due to theincreased surface area and larger amount of interstitialspacing between individual nanostructures, as opposed tomicrostructures [34]. Using theoretical density functionaltheory (DFT) calculations, we have recently demonstratedthat hydrogen can be molecularly and dissociatively adsorbedbetween closely spaced ZnO nanorods, which provided anexplanation as to why the nanostructures perform better ingas sensing applications [34]. It has been shown that thebehaviour of macroscopic hexagonal ZnO prisms [15] wasquite different to that of ZnO nanoparticles in studying aparticular gas reaction. Additionally, there are different viewson what the optimum temperature should be for the gas–ZnOinteraction [4], as well as what dopants and concentrationsshow best [27, 25]. Understanding the detailed properties ofH interacting with ZnO nanostructures obtained from differentsources also remains limited.

To date, there have been very few, if any, reportson the in situ analysis of the gas–surface interaction usingRaman spectroscopy, which are also confirmed by othercharacterization techniques such as XRD and FTIR. IRabsorption studies of hydrogen in ZnO at a high temperaturehave been performed, but IR absorption data at a lowertemperature has not been reported [13]. It has been suggestedthat the direct IR measurement of H-related shallow donors inZnO is hardly possible [13]. In several investigations, therehave been reports on the existence of IR absorption lines, andRaman and photoconductivity signals that are being detectedbut they have not been clearly assigned to hydrogen-relateddefects, as indicated in the publications of Lavrov [15, 14].

Very few experimental investigations reported real-timemonitoring of the interaction of hydrogen with ZnO surfaces.For ZnO powders, in particular, Windisch et al [42] suggestthat further testing is needed in order to make directcomparisons between XRD and Raman spectra results in orderto determine the presence of hydrogen. Also, the work byDu and Biswas [2] advises that there appears to be a hiddenhydrogen reservoir in ZnO that is invisible to IR spectroscopy,which can be associated with an interstitial H2 species [2].

In light of current uncertainties in the identification ofthe presence of hydrogen in ZnO nanomaterials, we conduct

in situ Raman spectroscopy, and ex situ XRD and IRspectroscopy characterization experiments, to investigate H2–ZnO interactions within nanopowders of various dimensions.The primary focus of this paper is to determine the effect of H2

on the IR, Raman spectra and XRD peak positions and relativeintensities obtained from the H2-exposed ZnO nanopowders.

2. Methods

Two different sizes of ZnO nanopowders used in this study(samples A and B) were sourced from Sigma Aldrich (<50 nmparticle size) and CRM (China Rare Metal Material Co.),respectively, with 99.99% purity. The polycrystalline powderswere drop-cast deposited onto quartz substrates after beingdispersed in ethanol solution and then heat treated (80 ◦C) on ahot plate to remove the volatile compound.

The morphology of the samples was characterizedby scanning electron microscopy (SEM) and transmissionelectron microscopy (TEM). SEM was performed using anFEI Nova NanoSEM operating at 15 kV in immersion mode.ZnO powders were deposited onto silicon substrates to preventelectron charging, using the same procedure as the depositionof ZnO on quartz substrates. TEM was performed using aJEOL 2010 TEM equipped with an LaB6 filament operatingat 200 kV. TEM samples were prepared by dropping thenanoparticles onto a copper grid (Lacey GSCu300FL).

The customized Raman spectrometer for this in situ studyconsists of a sample chamber (Linkham HFS91-PB4) that hasa quartz glass cover to allow the penetration of the Ramansystem laser. It was equipped with a built-in sample heaterand thermostat. The chamber outlet connects to a mass flowcontroller, regulating the flow of gas at 200 sccm. In situRaman spectroscopy was carried out with the sample chamberpositioned under a standard Olympus microscope (Olympusobjective lens Mplan 10× or SLMplan 50×) equipped witha 532 nm green laser (Shanghai Dream SD2-532-050T). TheRaman spectra were collected and analysed using Spectrasuite2009 software.

The crystal structure of the ZnO nanoparticles wasstudied by XRD using a Bruker D8 ADVANCE diffractometerfitted with a scintillation counter detector, using graphite-monochromated Cu Kα (λ = 1.5406 A) radiation. Data werecollected at room temperature with a scan range of 2θ = 5◦–70◦, a step size of 0.02◦ and a count rate of 2 s per step.

Diffuse reflectance IR spectroscopy was employed tostudy the ZnO surface vibrational frequency using a PerkinElmer Spotlight 400 device. The samples were then mixedwith potassium bromide (KBr powder), which is transparentto the incident IR radiation, and pressed into pellets for IRabsorbance analysis. The KBr powder must be of spectroscopygrade and spectroscopically dry [35]. As a result, the powderwas heated at 100 ◦C for a few minutes to remove residualwater.

3. Procedures

The ZnO samples (samples A and B) were annealed beforebeing exposed to target gases. Sample A consists of ZnO

2

Nanotechnology 23 (2012) 015705 K W J Wong et al

Figure 1. SEM micrographs of the prepared ZnO nanopowder samples obtained from (a) Sigma Aldrich (sample A) and (b) CRM (sample B).

grains of the order of less than 50 nm and sample B consistsof grains of the order of several hundreds of nanometres.The ZnO samples were annealed at 300 ◦C, using a rampingrate of 20 ◦C min−1. It is suggested that such an annealingtemperature can only eliminate ∼80% of the interstitialhydrogen from the ZnO interstitial site (H+

i ), leaving the morethermally stable hydrogen in the defect site (shallow donor,HO

+) in ZnO [33]. Nevertheless, the majority of the H donorswere removed from the surface, as will be shown later in thecharacterization results.

In situ Raman spectrum measurements were carried outwhen the ZnO samples were exposed to the following gasmixtures: N2, 3% H2 in N2 balance and zero air. N2 gasserved as an inert gas carrier for the H2, while zero airreplenished the O atom lost during the H2/ZnO interaction. Inthe experiment, the ZnO samples were placed in the chamberat room temperature and a Raman spectrum was taken as abackground comparison. Following that, Raman spectra weretaken throughout three different durations (30 min for N2,10 min 3% H2 in N2 balance and 20 min in zero air). Themeasurements were conducted at 300 ◦C.

For the XRD measurements, the two ZnO nanopowdersamples were exposed to 3% H2 in N2 balance for 30 min at300 ◦C. After that, one sample was examined immediately,while the other was stored for five days at room temperaturebefore being examined.

For the FTIR measurements, the two ZnO nanopowdersamples were annealed at 300 ◦C for 30 min, one in zeroair and the other in 3% H2 in N2 balance also at 300 ◦Cfor 30 min. These samples were then mixed with IR gradepotassium bromide (KBr powder) and pressed into pellets forIR absorbance analysis. A DRIFT experiment was carried outto study the change in frequency of the ZnO local vibrationalmodes induced by the presence of H2 using FTIR spectroscopy(Perkin Elmer Spotlight 400).

4. Results and discussion

Here we present and discuss the characterization experimentsconducted to assess the effect of H2 gas exposure on ZnOnanopowder samples, which leads to identification of theresulting gas–surface interactions.

SEM images of the two different types of ZnOnanopowders are shown in figure 1. As can be seen there isa clear distinction between the two types of nanopowders interms of their structural dimensions and homogeneity. Thegrain sizes of the annealed ZnO nanopowder samples fromCRM (sample B) are much larger and different than those fromSigma Aldrich (sample A). For sample A (figure 1(a)), ZnOnanograins of varying sizes were identified on the substrate,being typically as small as 10 nm to as large as several tens ofnm in length. However, the average grain size was overall lessthan 50 nm. The dimensions of the annealed ZnO nanopowdersourced from sample B were found to be of the order of severalhundreds of nanometres.

TEM images for samples A and B are shown in figure 2.ZnO nanoparticles with varying grain sizes were depicted infigure 2(a), whereas it was apparent that sample B is composedof larger particles with spherical morphology as shown infigure 2(b), which is in good agreement with the SEM imagesobtained. It can be seen that the nanoparticles are crystallinewith lattice spacings of 2.81 and 2.71 A for samples A and B,respectively. The lattice spacing in sample A can be correlatedto the (1010) planes in the ZnO crystal structure, whereas thecalculated lattice spacing in sample B may be identified as the(0002) lattice plane of the ZnO nanopowder [12].

Raman spectroscopy was conducted during the exposureof both samples to a selected mixture of gases. Ramanspectroscopy was utilized as it can be the best tool to revealany surface interactions with the target gas due to its sensitivityto the type of functional groups that can be formed as aresult of such interactions. Our equipment and the customizedgas chamber could allow in situ observations of the surfacereaction of hydrogen with both ZnO nanopowder samples inreal time.

The Raman spectra of sample A exposed to N2, 3%H2 in N2 and zero air at 300 ◦C are shown in figure 3; theRaman spectra of sample B for the same conditions are shownin figure 4. The Raman band modes identified are the E2

(high) at 435 cm−1, A1 longitudinal optical (LO) at 543 cm−1,E1 longitudinal optical (LO) at 566 cm−1 and a secondaryA1 longitudinal optical (2LO) at 1105 cm−1 [21]. Anotherlow frequency feature involving bands associated with thedifference between the E2 (high) and E2 (low) modes was alsoidentified at 333 cm−1 [42]. There exists a peak at 979 cm−1

3

Nanotechnology 23 (2012) 015705 K W J Wong et al

Figure 2. TEM micrographs of the non-annealed samples (a) A and(b) B. The inset shows a high resolution image for each sample inwhich lattice spacings of 2.81 and 2.71 A are seen.

that has been assigned as the A1(LO)+ E2 (high) mode, whichis known to correspond to the vibrational mode caused by thetwo-phonon combination [21].

The most intense mode identified in the Raman spectraof the ZnO nanopowder samples is in good agreement withpreviously reported values and corresponds to the oxygenvibration mode E2 (high) of ZnO [19, 30]. A strong peak atthe E2 (high) mode correlates to the oxygen vibration in theZnO crystal lattice, hence good crystallinity [30, 21]. However,our relatively weak signal at the E2 (high) mode suggests poorcrystallinity of the ZnO nanostructures.

In our case, the mode indicated non-homogeneouscrystallinity of sample A (smaller sized ZnO nanopowder),with a higher degree of crystallinity for sample B, as indicatedby the more intense peak [21]. It should be noted thatthe A1(LO) and the E1(LO) modes at 543 and 566 cm−1

wavenumbers, respectively, are fairly close to each other. Thevery weak A1(LO) mode has been reported [29, 42, 1] butrarely assigned a band mode in the literature [42]. However, adetailed relationship between the A1(LO) and E1(LO) modeshave been reported by Cusco et al, where the E1(LO) mode

Figure 3. Raman spectra of sample A annealed at 300 ◦Csequentially in (a) N2, (b) 3% H2 in N2 and (c) zero air.

Figure 4. Raman spectra of sample B annealed and recorded at300 ◦C sequentially in (a) zero air, (b) 3% H2 in N2, (c) zero air and(d) 3% H2 in N2.

is more strongly affected by impurity and/or defect than theA1(LO) mode [1]. The E1(LO) mode is correlated to surfacedefects such as oxygen vacancy or Zn interstitials in thenanopowder. The fact that the E1(LO) mode, which is presentin our spectra, indicates the presence of defects in the ZnOsamples [29]. The presence of impurities or surface defects canalter these two modes, especially with the E1(LO) mode beingmore strongly affected by the presence of oxygen vacanciesand Zn interstitials [30]. The phonon modes, we observe, havebeen previously reported in Raman spectra [1, 19].

For the interaction of hydrogen with sample B (largersized ZnO nanopowder), a similar pattern of Raman signalswas identified. Two measurements in zero air and 3%H2 in N2 were carried out to depict the similar trend ofgas/ZnO interaction and repeatability. The relatively sharperE2 (high) peak indicates a more homogeneous distribution ofthe nanopowder grain size and dimension. Comparing therelative ratio of the A1(LO) and E1(LO) modes to the E2 (high)mode, where the latter peak is much more dominant, it can beimplied that the ZnO nanopowder sample contains a greateramount of surface defects due to the annealing effect.

For sample A, the intensity of the Raman peak at 483 cm−1

increased more significantly after exposure to 3% H2 ina balance of N2 gas, relative to the other two exposures(figure 3). This is clearly indicated by the different baseline

4

Nanotechnology 23 (2012) 015705 K W J Wong et al

Figure 5. (a) XRD patterns of sample A: (i) taken immediately afterannealing (300 ◦C) with 3% H2 in N2 balance, (ii) annealed at 300 ◦Cin atmosphere and (iii) before annealing (hydrated). (b) XRDpatterns of sample A annealed at 300 ◦C with 3% H2 in N2 balance:(i) XRD measurement performed immediately after the annealingand (ii) XRD measurement performed five days after the annealing.

in the 483 cm−1 region. The intensity of this particularband has been observed to be dependent on the presence ofH2 [42]. While Windisch et al reported an additional peakat ∼570 cm−1, and regarded it as an anomalous mode arisingfrom oxygen vacancies in ZnO powders, Wang et al assignedthis peak to the E1—LO mode of oxygen deficiency in ZnOnanowires [39, 42]. Xu et al have also reported a strongintensity of the A1(LO) band with a sharp peak at 556 cm−1,indicating their thin film was severely oxygen-deficient [43].We believe that inconsistency in the reported oxygen-deficientband assignments between 540 and 581 cm−1 can possibly becorrelated to the different ZnO nanomaterials studied.

Despite these studies, none of the authors have madea correlation between the presence of hydrogen and theformation of hydroxyl groups on the ZnO surface, inconjunction with the assigned Zn interstitials and oxygen-deficient band. In such interactions, the formation ofsurface hydroxyls has been reported, mostly by FTIR orphotoluminescence methods [5, 8]. However, to the bestof our knowledge, the identification of the Raman peakcorrelated with a surface hydroxyl species has not yet been

Figure 6. XRD patterns of: (a) sample A and (b) sample B, bothannealed at 300 ◦C in atmosphere.

presented. Here, we conducted complementary XRD and FTIRcharacterization of the ZnO samples after the exposure to H2

to find any correlation of the 483 cm−1 band mode with theformation of surface hydroxyl groups on the ZnO nanopowdersurfaces.

A similar procedure for H2 exposure was applied to ZnOnanopowder sample B. No change of peak was observed in thisexperiment. It seems that the high crystallinity and grain sizeof nanopowder sample B is an important factor to inhibit theH2 gas interaction with ZnO.

A series of XRD measurements were carried out toidentify the presence of surface hydroxyl groups in ZnOsamples. The XRD patterns of sample A, before andafter exposure to H2, are shown in figures 5(a) and (b).Figure 6 shows the XRD patterns of samples A and Bannealed at 300 ◦C. Similarly, the XRD patterns of sampleB, before and after exposure to H2, can be found in thesupplementary information (available at stacks.iop.org/Nano/23/015705/mmedia). From the XRD data (figure 5(a) andsupplementary information available at stacks.iop.org/Nano/23/015705/mmedia), a decrease in most of the peaks can beseen when samples A and B were exposed to H2 as compared tothe annealed samples. For both samples the strongest reflectionwas attributed to a (1011) peak relative to (1010), whichsuggests the dominance of hexagonal nanostructures [18]. Ourdiffraction intensity ratio (I(0002)/I(1010)) of the (0002) polarplane to the (1010) plane was calculated to be ∼0.735 for bothsamples A and B. Diffraction intensity ratio (I(2110)/I(1010)) isthe ratio that describes the portion of ZnO(1010):ZnO(2110).The low ratio indicates the presence of a larger portion ofnon-polar planes as opposed to polar planes, i.e. a greaterproportion of (1010) and (2110) terminated ZnO surface facetsexist in our nanopowder samples. The existence of (1011) alsocorresponds to the presence of semi-polar planes. Furthermore,sample B is shown to be more crystalline as indicated by thegreater intensity of the peaks in the XRD pattern (figure 6).The same discussion that was presented for the XRD patternsin figures 5 is also similarly applicable for the figure 6 XRDpatterns. Basically the crystal phases and the peak intensity

5

Nanotechnology 23 (2012) 015705 K W J Wong et al

Figure 7. Infrared spectra of sample A recorded immediately after(a) annealing with zero air at 300 ◦C for 30 min and (b) annealingwith 3% H2 in N2 balance at 300 ◦C for 30 min.

ratios are quite alike. However, sample B does not present anypeaks that indicate the formation of hydroxyl groups.

Both ZnO samples were exposed to H2 gas in N2 balancefor 30 min, with the XRD plots being taken immediately after.While no extra peak was identified for the larger grain sizesample (sample B), an additional peak at 44◦ was found forsample A (figure 5(a)). This peak can also be seen in thenon-annealed ZnO nanopowder (hydrated) sample that wasdried before being exposed to H2. As also suggested byMuruganandham and Wu [24], the peak at 44◦ indicates thepresence of surface hydroxyl in ZnO [24]. Additionally inagreement with our Raman spectroscopy results, the XRDresults show that the adsorption stability of H2 on ZnO surfacesis directly related to the particle size and crystallinity of theZnO material.

An XRD of sample A, taken after several days ofstorage under atmospheric conditions, showed no peak at 44◦(figure 5(b)). Our XRD plot showed that the critical loss ofa hydroxyl group from the surface of the ZnO samples thatwere stored in a room temperature environment is possiblewithin days. To support our result, Lavrov et al [33] have alsoobserved that hydrogen leaves bulk ZnO after a period of amonth or more [7].

The IR absorption spectrum of sample A, which wasexposed to H2 gas, is shown in figure 7. Again, the coarsepowder (sample B) did not show any changes after H2 exposurecompared to the finer grained sample A, which demonstrateda change in the spectrum. Figure 7 also shows the effect ofintroducing H2 gas to the ZnO sample at 300 ◦C, where theincrease in the absorbance at 1618 cm−1 is associated with theemergence of surface bound hydroxyl groups, as has also beenpreviously reported [8].

In this work, we demonstrated that H2 gas can efficientlyinteract with ZnO nanopowders with small grain dimensionsand low crystallinity. While previously the change inthe Raman spectra of such an interaction has been mostlyassociated with ZnO interstitials and oxygen vacancies, wenow strongly believe that surface bound hydroxyl groups are

formed during the interaction and that the change in the Ramanpeak at 483 cm−1 can be strongly correlated with this change.

Our results are also in agreement with some of thedensity functional theory (DFT) calculations that show theadsorption of atomic H onto ZnO [38, 40] forms two possiblearrangements: one where the H atoms are adsorbed to surfaceZn and O atoms and the other (corresponding to highertemperatures) where the H atoms adsorb to only surface Oatoms. For both structures, a hydroxyl species is formed.Hydrogen prefers to adsorb to the O site on the ZnO(1010) and(2110) surfaces [34, 38, 40]. The interaction of H2 with theZnO(2110) surface has also been investigated using DFT, andatomic hydrogen was found to chemisorb to the surface [38].The H atoms were found to adsorb preferably to surfaceO atoms, again indicating that the formation of a surfacehydroxyl group is favourable. The effect of surface oxygenconcentration on the ZnO(2110) surface was also studied,where adsorption of hydrogen on the O-rich surface was foundto cause the surface to become metallic, whereas adsorptiononto the stoichiometric surface kept the semiconducting natureof the material. In this study, two types of energetically stableadsorptions were found. The first, where H atoms adsorbed onboth Zn and O sites (ZnO–2H), and the second where H atomsadsorbed on the O atom of the ZnO(2110) surface (ZnO–H),where the ZnO–H type adsorption has been identified as partialhydrogen adsorption.

Specifically, water molecules were found to continuouslydissociate and associate on the ZnO(1010) surface employingab initio molecular dynamics (MD) simulations, therefore,strongly suggesting an environment where water moleculesand surface hydroxyls can coexist [3]. The small activationbarrier (0.05 eV), which is required to trigger the dissociationof water, also suggests that the association of surface hydroxylsto form water molecules on the ZnO(1010) surface ispossible [3].

5. Conclusions

This study has demonstrated, via three different characteriza-tion approaches, the presence of surface hydroxyl groups insmall sized ZnO grain nanopowder samples at elevated tem-peratures. Raman spectroscopy, XRD and FTIR experimentsall strongly suggest that the interaction of H2 with the ZnOnanopowders leads to the formation of hydroxyl groups onlyin small sized powders. Importantly, our experiments indicatethe importance of the size and crystallinity of the ZnO grainsfor their interaction with hydrogen, where the presence of hy-drogen was not observed for highly crystalline ZnO nanopow-ders with larger grain sizes, while for the smaller grain sizedpowders, clear peak shifts confirm the presence of hydrogenand formation of hydroxyl groups. The presence of surface hy-droxyl groups on ZnO surfaces was found to be more prevalenton the non-polar surfaces, as supported by changes in the XRDpatterns and Raman spectra, as well as the FTIR signal.

The experiments suggest that ZnO nanopowder can beused to detect the presence of hydrogen at low concentrations,if the crystallinity and the grain size of the ZnO sampleare reduced. The relatively low intensity of the E2 (high)

6

Nanotechnology 23 (2012) 015705 K W J Wong et al

Raman peak can possibly be used as a good indicator of thematerial’s suitability. Conclusively, the existence of surfacehydroxyls reported in this work and the effect of ZnO grainsize established a benchmark for the identification of surfacebound hydrogen in ZnO nanopowders.

Acknowledgments

The authors acknowledge the facilities, and the scientificand technical assistance of the Australian Microscopy &Microanalysis Research Facility at the RMIT Microscopy andMicroanalysis Facility of RMIT University. The authorsalso gratefully acknowledge S Ippolito and F Antolasic forproviding assistance in the FTIR experiment.

References

[1] Cusco R, Alarcon-Llado E, Ibanez J, Artus L, Jimenez J,Wang B G and Callahan M J 2007 Phys. Rev. B 75 165202

[2] Du M H and Biswas K 2011 Phys. Rev. Lett. 106 115502[3] Dulub O, Meyer B and Diebold U 2005 Phys. Rev. Lett.

95 136101[4] Eriksson J, Khranovskyy V, Soderlind F, Kall P O,

Yakimova R and Spetz A L 2009 Sensors Actuators B137 94

[5] Ip K, Overberg M E, Heo Y W, Norton D P, Pearton S J,Stutz C E, Luo B, Ren F, Look D C and Zavada J M 2003Appl. Phys. Lett. 82 385

[6] Janotti A and Van de Walle C G 2007 Phys. Rev. B 76 165202[7] Jokela S J and McCluskey M D 2005 Phys. Rev. B 72 113201[8] Kleinwechter H, Janzen C, Knipping J, Wiggers H and

Roth P 2002 J. Mater. Sci. 37 4349[9] Klingshirn C 2007 ChemPhysChem 8 782

[10] Klingshirn C, Fallert J, Zhou H, Sartor J, Thiele C,Maier-Flaig F, Schneider D and Kalt H 2010 Phys. StatusSolidi b 247 1424

[11] Knapp M, Crihan D, Seitsonen A P, Lundgren E, Resta A,Andersen J N and Over H 2007 J. Phys. Chem. C 111 5363

[12] Krishnan D and Pradeep T 2009 J. Cryst. Growth 311 3889[13] Lavrov E V, Borrnert F and Weber J 2005 Phys. Rev. B

72 085212[14] Lavrov E V, Herklotz F and Weber J 2009 Phys. Rev. Lett.

102 185502[15] Lavrov E V, Herklotz F and Weber J 2009 Phys. Rev. B

79 165210[16] Law M, Greene L E, Johnson J C, Saykally R and Yang P D

2005 Nature Mater. 4 455[17] Li F, Wang L, Dai J, Pu Y, Fang W and Jiang F 2007

J. Lumin. 124 162[18] Li G R, Hu T, Pan G L, Yan T Y, Gao X P and Zhu H Y 2008

J. Phys. Chem. C 112 11859

[19] Lin K F, Cheng H M, Hsu H C and Hsieh W F 2006 Appl.Phys. Lett. 88 263117

[20] Lindsay R, Muryn C A, Michelangeli E and Thornton G 2004Surf. Sci. 565 L283

[21] Lo S S, Huang D, Tu C H and Jan D J 2009 J. RamanSpectrosc. 40 1694

[22] Losurdo M and Giangregorio M M 2005 Appl. Phys. Lett.86 091901

[23] McCluskey M D, Jokela S J, Zhuravlev K K, Simpson P J andLynn K G 2002 Appl. Phys. Lett. 81 3807

[24] Muruganandham M and Wu J J 2008 Appl. Catal. B 80 32[25] Paraguay D F, Miki-Yoshida M, Morales J, Solis J and

Estrada L W 2000 Thin Solid Films 373 137[26] Powell D A, Kalantar-zadeh K and Wlodarski W 2004 Sensors

Actuators A 115 456[27] Rout C S, Raju A R, Govindaraj A and Rao C N R 2006 Solid

State Commun. 138 136[28] Sadek A Z, Choopun S, Wlodarski W, Ippolito S J and

Kalantar-zadeh K 2007 IEEE Sensors J. 7 919[29] Scepanovic M, Grujic-Brojcin M, Dohcevic-Mitrovic Z,

Vojisavljevic K, Sreckovic T and Popovic Z V 2007 ActaPhys. Pol. A 112 1013

[30] Scepanovic M, Grujic-Brojcin M, Vojisavljevic K,Bernik S and Sreckovic T 2009 J. Raman Spectrosc. 41 914

[31] Schmidt-Mende L and MacManus-Driscoll J L 2007 Mater.Today 10 40

[32] Shi G A, Saboktakin M, Stavola M and Pearton S J 2004 Appl.Phys. Lett. 85 5601

[33] Shi G A, Stavola M, Pearton S J, Thieme M, Lavrov E V andWeber J 2005 Phys. Rev. B 72 195211

[34] Spencer M J S, Yarovsky I, Wlodarski W andKalantar-zadeh K 2011 Nanotechnology 22 135704

[35] Stuart B 2000 Kirk-Othmer Encyclopedia of ChemicalTechnology (New York: Wiley)

[36] Tamaekong N, Liewhiran C, Wisitsoraat A andPhanichphant S 2009 Sensors 9 6652

[37] Torbrugge S, Ostendorf F and Reichling M 2009 J. Phys.Chem. C 113 4909

[38] Wang C C, Zhou G, Li J, Yan B H and Duan W H 2008 Phys.Rev. B 77 245303

[39] Wang X, Li Q Q, Liu Z B, Zhang J, Liu Z F and Wang R M2004 Appl. Phys. Lett. 84 4941

[40] Wang Y, Meyer B, Yin X, Kunat M, Langenberg D, Traeger F,Birkner A and Woll C 2005 Phys. Rev. Lett. 95 266104

[41] Wang Z L 2009 Chin. Sci. Bull. 54 4021[42] Windisch C F, Exarhos G J, Yao C H and Wang L Q 2007

J. Appl. Phys. 101 123711[43] Xu X L, Lau S P, Chen J S, Chen G Y and Tay B K 2001

J. Cryst. Growth 223 201[44] Yuri M S, Howard L M, David C L, Donald C R, Cole W L,

Nelson Y G, Nancy C G, Larry E H, Shigeru N andLeonard J B 2004 Appl. Phys. Lett. 84 2545

[45] Zhang S B, Wei S H and Zunger A 2001 Phys. Rev. B63 075205

7