Vol. 134 (2018) ACTA PHYSICA POLONICA A No. 6

Applications of ZnO Nanorods and Nanowires — A ReviewB.S. Witkowski∗

Institute of Physics, Polish Academy of Sciences, Aleja Lotnikow 32/46, PL-02668 Warsaw, Poland(Received December 11, 2018)

This paper contains a review describing the various potential applications of ZnO nanorods in many differentdevices such as light emitters (light emitting diodes and lasers) and detectors (on single nanorod or nanorods matri-ces), photovoltaic cells (dye-sensitized, perovskite-based, thin film, organic and silicon solar cells), photocatalyticapplications, field emitters, field effect transistors, chemical and biological sensors, and electric nanogenerators.The review describes state of art in this field from more than twenty years perspective, presenting problems andchallenges to be solved.

DOI: 10.12693/APhysPolA.134.1226PACS/topics: ZnO nanorods, ZnO nanowires, photovoltaic cells, solar cells, LED, laser, detector, sensor, photo-catalytic, field emission, field effect transistor, nanogenerator

1. Introduction

Zinc oxide is a semiconductor with a direct band gap of3.37 eV of high transparency and high n-type conductiv-ity [1]. This means that ZnO is suitable for use in a widerange of optical and electronic applications. It is alsonotable that ZnO is piezoelectric and exhibits high me-chanical, thermal, and chemical stability [2]. ZnO is alsobiocompatible, and thus does not pose a threat to peo-ple and animals [3]. The great advantage of ZnO is theease of crystallization of various forms of nanostructuresby many methods (chemical vapor deposition (CVD) [4–9], thermal evaporation [7, 10–14], molecular beam epi-taxy (MBE) [15], pulsed laser deposition [1], hydrother-mal [16], atomic layer deposition (ALD) [17] and othermethods [18]), some of which are extremely cost-effectiveand environmentally friendly [19]. Thus, a wide rangeof possible applications of ZnO layers and nanostruc-tures are under investigation. In this review, we focuson possible applications of ZnO nanorods and nanowires(NRs). This review is organized as follows: light emitters(light-emitting-diodes (LED) and lasers), light detectors,photovoltaic (PV) cells, photocatalytic applications, fieldemission (FE), field effect transistors (FET), chemicaland biological sensors, and power nanogenerators.

2. Light emitters2.1. Light emitting diodes

LED technology was originally developed from high-efficiency electroluminescence (EL) semiconductor de-vices in early 1962 [20]. The high brightness, high ef-ficiency and long durability of LEDs renders them idealfor displays and solid-state lighting and their share of themarket for light emitters is growing constantly.

ZnO NRs, after GaN structures, are the top choicefor new short wavelength LED and laser diode appli-cations not only because of their superior properties,

∗e-mail: bwitkow@ifpan.edu.pl

with their wide band-gap (3.37 eV) and a large exciton-binding energy (60 meV) [21], but also due to their be-ing easy to grow via chemical and physical vapor-phaseapproaches [22, 23]. Independent ZnO-based nanostruc-tures can easily be formed, including NRs. It is knownthat ZnO NRs can be grown following a designed patterninto aligned arrays on almost any substrate in any shape.The energy bandgap can be expanded by substituting asmall amount of Mg or reduced by adding Cd [24, 25].However, p-type ZnOmaterials with sufficiently high con-ductivity and carrier density are still no guarantee of sta-bility and reproducibility, and the ZnO homojunctionsare not yet practically applicable to LEDs.

Heterojunction diodes based on ZnO film/GaN filmstructures have been suggested in order to exploit theadvantages of ZnO, but heterojunction diodes have lowefficiency because of the high energy barrier at thejunction interface. This limitation of current thin-film technology can be surmounted by employing ZnONRs which have superior properties, such as high crys-talline quality, large surface area, and the quantumconfinement effect.

The first attempts to create luminescent diodes usingZnO were heterostructures, in which ZnO NRs arraysand polycrystalline n-ZnO were grown on Mg-doped p-GaN films by the CVD method [26]. Electrolumines-cence emission at 386 nm wavelength was observed un-der forward bias in the heterojunction diodes and UV-violet light emerged from the ZnO NRs. Satisfactoryemission was achieved only after thermal treatment inhydrogen ambient to increase the electron injection ratefrom the n-ZnO films into the ZnO NRs. The high con-centration of electrons supplied from the n-ZnO films ac-tivated the radiative recombination in the ZnO NRs andincreased the light-emitting efficiency. Very similar ap-proaches were also used in the following years, i.e. inwork [27], but in this case, due to impurities and de-fects in the materials, the light emission was poor andin a wide visible range. This kind of structure is stillunder investigation, but with minor changes. For ex-ample, in work [28], the method for growing ZnO NRswas nanoparticle assisted pulsed laser deposition. The

(1226)

http://doi.org/10.12693/APhysPolA.134.1226mailto:bwitkow@ifpan.edu.pl

Applications of ZnO Nanorods and Nanowires — A Review 1227

radiative electron–hole recombination was taken in theZnO NRs due to the small energy barrier for holes.The p-GaN/n-ZnO heterojunction diodes with use ofNRs exhibited improved EL emission in UV and bluecolor compared to those of the film-based GaN/ZnOheterojunction diodes.

A high-brightness LED based on GaN/ZnO NRs het-erojunction was achieved in a similar structure, but thespaces between the NRs were filled with organic material— A11 poly(methyl methacrylate) (PMMA), althoughthis material did not take an active part in the junctionand emission [29]. A blue shift was observed in the ELwith the increase of bias voltage, indicating the modifi-cation of external voltage to the band profile in the de-pletion region. Nevertheless, efficient and stable emissionin the near UV/violet range was obtained and presentedthe first hybrid heterojunction LED device.

ZnO NRs were also used in GaN LEDs but not as anactive part in the p–n junction, but as an emission am-plifier. Work [30] presented the enhanced light output ofGaN-based LEDs with vertically aligned ZnO NRs arraysgrown by metal organic vapor phase epitaxy (MOVPE)directly on a p-GaN layer (next to the electric contact).Compared to conventional GaN LEDs, the light outputof GaN LEDs with ZnO NRs arrays increased up to 50%and 100% at applied currents of 20 and 50 mA, respec-tively. The authors claimed that the electrical field ofGaN LEDs can be changed by growing ZnO nanorods onp-GaN, and that with the reduction in internal polariza-tion, GaN LEDs with ZnO nanorods are very promisingas high-efficiency LEDs. These results indicate that theZnO NR arrays on GaN LEDs result in a decrease in thetotal internal reflection of the generated light in the activeregion of the LED. Also note that the operating voltageof the GaN LEDs with ZnO NRs is higher than that ofthe LEDs without ZnO NRs. This example shows thatwithout changing the LED structure, by growing ZnONRs it is easy to achieve higher-brightness LEDs. Therewere also reports about a dramatic increase in the lightextraction efficiency of GaN-based blue LEDs by ZnONRs arrays grown on a planar indium tin oxide (ITO) —transparent electrode [31]. The light output efficiency ofthese LEDs was increased by about 57% with no increasein forward voltage over the conventional LEDs with pla-nar ITO. The increased light extraction by the ZnO NRarrays is due to the formation of sidewalls and a roughsurface, resulting in a multiple photon scattering at theLED surface.

There were also some reports of application of ZnONRs in different semiconductor heterostructures used forLED construction, i.e. work [32] presents a LED basedon n-ZnO NRs/p-Si heterojunction. ZnO nanorods weregrown on p-type Si substrates employing an easy low-temperature aqueous solution method. Spaces betweenNRs were filled with high-molecular-weight polymers. Atypical EL spectrum of such an n-ZnO/p-Si heterojunc-tion under forward bias was composed of a ultravioletemission centered at 387 nm and a broad green band

at 535 nm, consistent with the photoluminescence (PL)spectrum. Unfortunately, the obtained intensities andspectral ranges were not satisfactory.

A very interesting direction of research is in thearea of hybrid structures, in which ZnO NRs arecombined with the p-type organic material. Thefirst work in this field shows EL from the structureZnO NRs/poly(3,4-ethylene-dioxythiophene)(PEDOT)/poly(styrenesulfonate)PSS [33]. The ZnO NRs weregrown in a low-temperature process from aqueous solu-tion and additionally encapsulated in a thin polystyrenefilm deposited from high-molecular-weight solutions.This configuration is mechanically very robust. The near-vertical orientation allows use of sandwich-like contacts,which can easily be patterned. Light emission was ob-tained in the near-UV region (low intensity of ZnO ex-citonic luminescence) and a stronger emission coveringthe entire visible area up to the near-infrared. Unfortu-nately, the structure showed stability for only an hour.This is a main problem in many applications of organicsemiconductors, which are to be solved by introducingcoating layers.

Better time stability can be achieved by tightly clos-ing the structure, e.g. reference [34] presents an LEDconsisting of SnO2/ZnO/PEDOT structures with an in-ternal polystyrene insulator layer. Moreover, as-grownfilms show a broad electroluminescence band over thevisible spectrum, but annealing at quite low tempera-ture of 300 ◦C increases the emission and strongly raisesthe excitonic part of emission. The complete processingprocedure was compatible with large-area fabrication onflexible substrates.

In 2008, the possibility of creating a LED based onZnO NRs (grown from aqueous solution) in junction withpoly(3-methylthiophene (PMT) was presented [35]. Thetypical EL spectra obtained from this heterostructure de-vice under forward bias show a strong ultraviolet peakand a weak visible emission. The emission of the het-erostructure device is much stronger than that of theZnO-only, however, time stability continues to be a sig-nificant problem.

Another and very interesting approach from the stand-point of physics were investigations related to single NRs.In 2006, a novel technique was presented for reliableelectrical injection into single semiconductor NRs forLEDs [36]. The first demonstration of this technique wasconstructing the first ZnO single-NR LED. The deviceexhibits broad sub-bandgap emission at room tempera-ture. This is a very interesting approach; however, due tothe complexity of this technique and the necessity to usea negative electron-beam resist and direct electron-beampatterning (which is associated with very high costs) thecommercial use of this approach is currently excluded.

2.2. Lasers

Similarly as in the case of LEDs, single-crystallinesemiconductor NRs have long been considered an excel-lent candidate for achieving laser emission, for example,

1228 B.S. Witkowski

to realize cost effective Fabry–Perot (FP) type lasers, be-cause of the optical feedbacks provided by the naturallyformed flat facets in the ends of NRs.

Already in 2001, laser action in the UV range wasdemonstrated from a matrix of well-oriented ZnO NRsat room temperature [37]. The ZnO NRs were grownon sapphire substrates using a simple vapor transportand condensation process. ZnO NRs form natural lasercavities with diameters varying from 20 to 150 nm andlengths up to 10 µm. Under optical excitation, surface-emitting lasing action was observed at 385 nm, with anemission linewidth less than 0.3 nm. Optical pumpingwith a wavelength of 266 nm was used, but high pow-ers were needed for the lasing action (a threshold of40 kW/cm2).

In 2004, demonstrated were operating random lasersbased on ZnO NR arrays embedded in the ZnO epilay-ers, which consist of a thick MgO buffer layer and a ZnOthin film [38]. The presence of ZnO epilayers increasesthe optical gain so that random lasing can be sustained.Under 355 nm optical excitation at room temperature,sharp lasing peaks at around 390 nm with a linewidthless than 0.4 nm were observed in all directions. Thedependence of the lasing threshold intensity on the ex-citation area was in good agreement with the randomlaser theory [39]. We thus claim that such emission inZnO NRs arrays was verified.

Finally, in 2011 the fabrication of electrically pumpedZnO NR diode lasers using p-type Sb-doped ZnO NRsand n-type ZnO film was reported [40]. FP-type UVlasing was demonstrated at room temperature with goodstability.

Of course, the presented results require further opti-mization work, i.e. the top contact with the p-type NRmight be engineered to offer both good optical trans-parency and low electrical resistivity; this is, however, avery important step to be solved for the use of ZnO NRsin commercial light emission devices.

3. Light detection applicationsZnO NRs are an example of nanostructured materi-

als which have attracted much attention because of theirmassive surface area to volume ratio, their high sensi-tivity to adsorbed molecules on their surfaces and theirgood electric transport features. ZnO NRs have showngreat promise as UV sensors because of their wide bandgap, visible light blindness (in case of high-quality NRs),low-cost and ease of nanostructure fabrication. However,an essential hindrance to ZnO single NR UV sensors isthe very low photo response current due to the small sizeof individual NRs and chemical processes on the surfacewhich in the case of nanorods play a key role.

3.1. Detectors on a single NR

The first reports on the use ZnO NRs as light (UV)detecting elements were published in 2002. The pos-sibility of creating highly sensitive NR switches by us-ing the photoconductive properties of individual semi-conductor NRs [41] were demonstrated. Individual ZnO

NRs were dispersed directly on pre-fabricated gold elec-trodes. The conductivity of the ZnO NRs is extremelysensitive to ultraviolet light exposure. The light-inducedconductivity increase allows us to reversibly switch theNRs between off-state (without UV illumination) andon-state (under UV illumination). The reaction mech-anism was connected with adsorption and desorption ofoxygen from air. Next, in 2005 there was report onthe photoresponsive characteristics of single ZnO NRsgrown by site-selective MBE [42]. These NRs had a rel-atively fast photoresponse and showed electrical trans-port dominated by bulk conduction. In 2007, ZnO singleNR visible-blind UV photodetectors with internal pho-toconductive gain as high as G ≈ 108 were fabricatedand characterized [43]. The photoconduction mechanismin these devices was clarified by means of time-resolvedmeasurements spanning a wide temporal domain, from10−9 to 102 s, revealing the coexistence of fast and slowcomponents of the carrier relaxation dynamics. The ex-tremely high photoconductive gain was attributed to thepresence of oxygen-related hole-trap states at the NR sur-face, which prevents carrier recombination and prolongsthe photocarrier lifetime, as evidenced by the sensitivityof the photocurrent to ambient conditions.

In the following year (2006), the possibility of engi-neering low-power photodetectors by employing individ-ual Cu-doped ZnO NRs was suggested [44]. Since the Cucenters are optically active in ZnO, use of Cu dramati-cally enhances the photosensitivity levels of the NRs toboth UV and visible light as a result of avalanche photo-multiplication. The paper was more a presentation of thepossibility of the use of Cu-doped NRs than a real pre-sentation of a device, nevertheless the results were verypromising.

In 2009, the first single crossed ZnO NR UV photosen-sor was demonstrated [45]. An in situ lift-out techniquewas employed to fabricate this photosensor based on self-assembled crossed NRs. The photoresponse suggestedthat they can be regarded as candidates for optoelec-tronic devices and may be used for UV detectors. Thepoor sensitivity and necessity of use of an focused ionbeam (FIB) causes that this approach is not attractivefor industrial application.

The devices on individual nanostructures generallyrequire sophisticated processing methods, which makestheir commercial applications extremely difficult. In thisrespect, the use of whole matrices of nanostructures, evenwith the loss of sensitivity or electrical parameters, ismuch more promising.

3.2. Detectors on NRs matrices

The first paper describing the behavior of a ZnO NRsmatrix under UV illumination was published in 2004 [46].In summary, the possible scenario of the photocurrentof ZnO NRs was discussed from the point of adsorptionand desorption of the water molecules giving the hydroxylgroup to the oxygen defect sites with the diffusion processthrough the ZnO NRs.

Applications of ZnO Nanorods and Nanowires — A Review 1229

In 2006, using the bottom-up approach a UV photodi-ode, consisting of heterojunction of n-type ZnO NRs andp-type silicon was demonstrated [47]. The performanceof UV photodiodes could be enhanced due to the goodUV responsivity and larger excitation area to volume ra-tio of ZnO NRs. According to experimental results, thecurrent–voltage characteristics of the device showed thetypical rectifying behavior of heterojunctions, and thephotodiode exhibited a response of ≈ 0.07 A/W for UVlight (365 nm) under a 20 V reverse bias. The sensitivityto UV was rather poor and this approach also requireduse of a quite large reverse bias, but it was the first pro-totype of UV photodiode using ZnO NRs.

In 2008, a method was proposed to fabricate ZnObridging NR structure exhibiting high performance in UVdetection [48]. This method uses a single-step CVD pro-cess, in which the electrodes and sensing elements are fab-ricated at the same time. The electrodes are formed bythe thick ZnO layers covering the Au-catalyst-patternedareas on the substrate, and the sensing elements consistof the ultra-long ZnO NRs bridging the electrodes. Thefabricated nanobridge device shows fast response to UVillumination in ambient air and also the highest sensi-tivity among the ZnO NRs array. A similar sensitivityfor the ZnO NRs matrix was reported only once in sub-sequent years [49]. The detector described in this workwas characterized by a slightly lower sensitivity, but asimpler production technology. The basis of the detectorwas a matrix of nanorods produced on quartz substrateby extremely simple and fast growth from water solu-tion, and the resultant nanorods were defect-free, whichwas the key to their high sensitivity. Detection edgewas close to 400 nm wavelength. Importantly, anneal-ing and/or chemical treatment was not necessary beforereuse. Detection was based on monitoring the sample re-sistance changes which was associated with the adsorp-tion and desorption processes of hydroxyl groups on thesurface of the nanorods. The electric response was scal-ing with light intensity. High sensitivity and extremelysimple manufacturing technology make this detector veryattractive for potential applications.

In 2009, a few papers were published on UV detectorsbased on ZnO NRs matrices produced by MOCVD or arcdischarge method [50, 51]. However, the obtained sensi-tivities were not groundbreaking. In the same years, pa-per [52] was published with photoresponse measurementsof ZnO NRs under different relative humidity to studythe surface effects of oxygen and water. The proposedmodel can help to better understand gas desorption andadsorption processes on the surface of metal oxide semi-conductors, which play a central role in photodetectionand gas sensing.

In 2010, a highly interesting work was publishedabout well-oriented ZnO NRs distributed onto the sur-face of ZnO nanofibers. The matrix was producedby the combination of electrospinning and hydrother-mal reaction [53]. Both the current and the UV re-sponsiveness were enhanced significantly after grow-

ing ZnO NRs onto the surface of ZnO nanofiber.After surface functionalization with a layer of N719(di-tetrabutylammonium cis-bis(isothiocyanato)bis(2,2′-bipyridyl-4,4′-dicarboxylato)ruthenium(II)) dye mole-cules, the current of the ZnO NR/ZnO nanofiber devicecould also be regulated by visible light.

In 2011, a bottom-up approach to fabricating macro-scopic integrated ZnO NR UV sensors was demon-strated [54]. Macroscopic multi-NR-based UV sensorswere fabricated by transferring vertically aligned ZnONRs to a parallel-aligned layer on a polydimethylsilox-ane (PDMS) surface, followed by integration with elec-trodes patterned on rigid glass and flexible PET sub-strates. This multi-NR sensor design saved all the advan-tages of single-NR devices, including high on/off currentratio and fast response and recovery times.

In 2012, another approach was reported — high qualityZnO NRs were grown on p-type Si (111) substrates viamicrowave-assisted chemical solution methods [55]. Theunbiased operation mode of this UV detector offered ex-cellent stability over time. The high sensitivity and fasterresponse and recovery times of this ZnO NR array/p-SiUV detector make this device a promising self-poweredUV detector.

A few years later in very similar approach, a photore-sistor based on ZnO NRs and silicon substrate (very low— p-type) was presented [56]. The wavelength detectionrange is between 400 and 1100 nm which reflects lightabsorption by the silicon partner of the junction. Im-portantly, post-illumination annealing or chemical treat-ment is not necessary before reuse of the detector. Theabsorption of light near the junction of two materialscauses generation of electron–hole pairs which were sep-arated by the band bending in the depletion layer. Theband bending caused the flow of electrons to ZnO NRs(slightly n-type) which were not actively involved in thephotocurrent. The holes gathered on the Si side, thus,creating an active conducting channel. High sensitiv-ity and the extremely simple manufacturing technologymade this detector very attractive for potential applica-tions. The poor relationship between the response andthe light intensity enabled usage as an optical switch.

4. ZnO NRs in PV applications

A wide energy band gap in the UV area prevents directabsorption of visible light while retaining high crystallo-graphic quality, which is necessary for effective separa-tion of carriers. Despite this, the advantages of the NRgeometry can improve the efficiency of many types ofphotocells, dye-sensitized solar cells (DSSC), perovskite,organic, or even silicon based.

4.1. Dye-sensitized solar cells

Among various applications, the utilization of ZnOnanostructures as a photo-electrode for DSSCs has re-ceived a considerable attention and interest recently dueto their compatibility with the commonly used TiO2 ma-terials [57]. ZnO shows higher electronic mobility and

1230 B.S. Witkowski

a similar energy level of the conduction band to TiO2which make ZnO suitable for use as a photo-electrodematerial for the fabrication of efficient DSSCs.

The NR dye-sensitized solar cell is an exciting variantof excitonic photovoltaic devices. As an ordered topol-ogy that increases the rate of electron transport, a NRelectrode may provide a means to improve the quantumefficiency of DSSCs. Raising the efficiency of the PV cellto a competitive level depends on achieving higher dyeloadings through an increase in surface area.

The first demonstration of the possibility of using NRsin DSSC instead of metal nanoparticles (NPs) was pub-lished in 2005 [58]. The idea of integrating nanorods withDSSC was under development for many years. The firstwork with the use of ZnO NRs appeared in 2006 [59].ZnO NRs were used as a photo-electrode. Typical so-lar cell photocurrent was 1.3 mA/cm2, photovoltage —0.7 V, fill factor — 0.35–0.40 and overall efficiencieswere 0.3%. Electron transport through the NRs wasnot constrained by the NR dimensions. ZnO NR-basedDSSC were prepared by adsorbing N719 dye onto theNR surface.

In 2008, an article was published with studies of theinterface capacitance of a DSSC made from chemicallyprepared ZnO NRs by time-resolved photocurrent andelectrical impedance measurements [60]. While the elec-trical response was in good agreement with the spacecharge capacitance theory, the variation of the measuredcapacitance with the applied bias was found to deviatesignificantly from the simple electrostatic model. This isin strong contrast to the predictions of the electrostatictheory that is usually applied to model the space chargecapacitance. Even the numerical solution of the full elec-trostatic problem in quasi 3D did not provide a goodapproximation. This work suggested the need for fur-ther investigations to better understand the quasi staticcharge distribution in the ZnO NR assemblies preparedfrom solution.

In the same year, a work appeared with a compar-ison of the performance of DSSC based on ZnO NRsprepared by hydrothermal and vapor-deposition meth-ods [61]. DSSCs based on hydrothermally grown rodsexhibit higher power conversion efficiency, which can beattributed to the higher dye adsorption. The photo-voltaic performance exhibits a complex relationship tothe presence of surface OH groups and defect emissions,and optimal annealing conditions are strongly dependenton the growth method used. Efficiency of 0.22% was ob-tained for both as-grown hydrothermally grown NRs andvapor deposited nanorods annealed in oxygen at 200 ◦C.

Research on conjunction ZnO NRs with this sensitiz-ing dye has been under further investigation. In 2009,a paper was published with results on vertically alignedZnO NR arrays coated with gold nanoparticles used inSchottky barrier PV cells [62]. After incorporation ofN719 sensitizing dye, the open circuit voltage increasedto 0.63 V from 0.5 V being measured for dye-sensitizedsolar cells based on the bare ZnO NRs. The efficiency

of the gold-coated ZnO NR dye-sensitized solar cells wasthus increased from 0.7% to 1.2%. In the same year therewas another paper on well-crystallized hexagonal-shapedZnO NRs used as photo-anode materials, also in conjunc-tion with N719, to fabricate the DSSCs which exhibitedan overall light to conversion efficiency of 1.86% witha very high fill factor of 74.4%, short-circuit current of3.41 mA/cm2 and open-circuit voltage of 0.73 V [63].

In 2010, results were presented on DSSC based on ZnONRs and N719, fabricated and modified through the ad-dition of Au NPs [64]. The high quality as-synthesizedZnO NRs were well-dispersed and led to a high cell ef-ficiency of 5.2%. Cells with Au NPs resulted in lowerefficiency (2.5%). The thick layer of Au NPs aggrega-tion may have led to distortion of the plasmonic effect.Also, the addition of Au NPs effectively decreased thesurface area of ZnO NRs with direct contact to the dyemolecules, resulting in a lower amount of adhered dyemolecules to convert sunlight. The electrons generatedby the photo-absorption through the thickly aggregatedAu NPs layer may have a lower injection rate to ZnONRs as compared to those absorbed by the dye.

The last report on DSSC contains description of veryinteresting structures in the form of a nanoforest of highdensity, long branched treelike multigeneration hierar-chical crystalline ZnO NR photoanodes obtained by aselective hierarchical growth sequence [65]. The short-circuit current density and overall light conversion effi-ciency of the branched ZnO NR DSSCs were almost fivetimes higher than the efficiency of standard vertical ZnONRs and reached 2.63%. The efficiency increase is dueto the increased surface area enabling higher dye loadingand light harvesting, as well as to reduced charge recom-bination through direct conduction along the crystallineZnO nanotree branches.

Despite these interesting results, the use of ZnO NRsin DSSC is currently undergoing a period of stagnation.

4.2. Perovskite-based PV cells

Like in the case of DSSC cells, the advantages of ZnONRs can be used in perovskite solar cell (PSC) technol-ogy, i.e. higher electronic mobility and similar energylevel of the conduction band to TiO2. Already in 2014a work was published on high-efficiency ZnO NR-basedperovskite solar cell with power conversion efficiency(PCE) of over 11% with short-circuit current density JSCof 20.08 mA/cm2, open-circuit voltage (VOC) of 991 mVand fill factor of 0.56 [66]. The ZnO NR-based PSCshowed almost ideal external quantum efficiency (EQE)spectral shape with excellent matching of the calculatedJSC with the measured JSC . Compared to a limited de-tection in charge collection due to the slow charge collec-tion rate and low absorbed photon-to-current conversa-tion efficiency for the TiO2 NR-based PSC, the ZnO NRsystem showed fast saturation in charge collection, whichindicates that ZnO NRs are an effective charge collectionsystem in CH3NH3PbI3 based perovskite solar cells.

Applications of ZnO Nanorods and Nanowires — A Review 1231

Two years later, a work was presented on ZnO NRsperovskite-based solar cells on rigid and on flexible sub-strates [67]. Optimizing the ZnO NRs growth con-ditions enabled achievement of a PCE of 9.06% witha current density of 21.5 mA/cm2. Moreover, flex-ible ZnO NRs PSCs were fabricated, demonstratingexcellent stability over 75 bending cycles with PCEof 6.4%. This investigation shows that ZnO NRs canbe used as an efficient alternative to low temperaturephotoanode.

Despite the considerably higher efficiency of inorganic–organic metal hybrid PSCs, electron transport was stilla challenging issue. In 2017, a report was publishedon the use of ZnO NRs prepared by hydrothermal self-assembly as the electron transport layer in PSCs [68].The efficiency was significantly enhanced by passivat-ing the interfacial defects through atomic layer deposi-tion (ALD) of Al2O3 monolayers on the ZnO NRs. Byemploying the Al2O3, the average power conversion ef-ficiency of methylammonium lead iodide PSCs was in-creased from 10.33% to 15.06%, and the highest effi-ciency obtained was 16.08%. The authors suggested thatthe passivation of defects using the ALD of monolayersmight provide a new pathway for the improvement of alltypes of PSCs. The results received are very promis-ing and achieving higher efficiencies in such cells is onlymatter of time.

4.3. Thin film PV cells (based on Cd and Cu)

Very interesting results were also obtained with theapplication of ZnO NRs in Cd-based photovoltaic cells.In 2005, a work was published on an ETA (extra thinabsorber) solar cell fabricated from an electron-acceptinglayer of free standing ZnO NRs, a light-absorbing layer ofCdSe, and a hole-accepting layer of copper thiocyanatewhich fills vacuities between the ZnO/CdSe NRs [69].The ZnO distracts photons within the light-absorbinglayer, increasing the chance for light absorption. Mean-while, the low thickness of the CdSe layer facilities trans-fer of photogenerated electrons to ZnO. This nanostruc-ture creates an output current that is up to 50 timesgreater than in a solar cell made of flat, dense substratematerial. The total efficiency reached 2%.

In 2007, a quantum-dot sensitized NR solar cell basedon photosensitization of ZnO NRs with CdSe quantumdots was demonstrated [70]. Photocurrent was generatedfrom visible light by the excitation of electron hole pairsin the CdSe quantum dots. The electrons were injectedacross the quantum dot-NR interface into the ZnO. Fur-ther improvement on the efficiency and stability of thesedevices would require development of an optimal holetransport medium for this architecture. The received ef-ficiency below 1% is much lower than theoretically pre-dicted efficiencies.

In 2009, a work was presented on ZnO/CdSe core-shellNR array films used as the photo anodes in solar cells [71].A power conversion efficiency of 0.34%, a photocurrentdensity of 3.10 mA/cm2 and an open circuit potential of

0.43 V were achieved under optimum parameters. Al-though the power-conversion efficiency is not sufficientlyhigh in this preliminary work, using CdSe nanocrystalsas sensitizers for the ZnO solar cells is quite encouragingby reducing the thickness of the CdSe shell and tuningthe size of the CdSe nanocrystals in order to improvethe light absorption properties of the electrodes. Despitethe efficiency issue, this study gives some insights intothe fundamental mechanisms. Unfortunately, this didnot translate into further improvement of results in thisconfiguration.

4.4. Organic based PV cells

Another direction for the development of photovoltaiccells is a conjunction of ZnO NRs and organic semicon-ductors. The first paper in this field was published in2006 and described a photovoltaic device based on NRstructure with poly(3-hexylthiophene-2,5-diyl) (P3HT)polymer and showed a power conversion efficiency overfour times greater than that for a similar device basedon NPs [72]. The best ZnO NR:P3HT device usingthe molecular interface layer yields a short circuit cur-rent density of 2 mA/cm2 under AM 1.5 illumination(100 mW/cm2) resulting in a power conversion efficiencyof 0.2%. The low VOC can be attributed partly to shuntlosses due to imperfections in the dense ZnO layer, andthe low JSC to low charge separation efficiency due to thelarge film pore sizes (≈ 100 nm) relative to the P3HT ex-citon diffusion length.

In 2007, a paper appeared on the effect of polymerprocessing on P3HT/ZnO NR photovoltaic device per-formance [73]. The best device was characterized by aPCE of 0.28%, a VOC of 440 mV, a JSC of 1.33 mA/cm2,and a fill factor of 48%. These data clearly show that theoptimization of solvent selection and polymer processingcan significantly improve the performance of P3HT/ZnONR composite devices.

In the same year, the performance of the hybrid pho-tovoltaic devices was improved as the average length ofthe ZnO NR was increased [74]. From the dependence ofthe photovoltaic performance on the ZnO NR length andthe organic layer thickness, it was clearly demonstratedthat the introduction of the direct transport channel tothe electrode could enhance the performance of photo-voltaic devices with thick active layers. Such thick activelayers could be advantageous especially when polymerswith small extinction coefficients and broad absorptionspectra are used for the absorption layers. The fill fac-tor of the devices increased from 38% to 50% when anarray of the ZnO NRs was introduced, which directlycontributed to the improvement of the power conversionefficiencies up to 2.7%.

In the next year, another version of hybrid flexibleorganic PV cells was demonstrated. ZnO NRs weregrown on transparent and conductive single-walled car-bon nanotube (CNT) thin films [75]. The structurewas created by directly growing ZnO NRs from solutionon the single-walled CNT electrodes and spin coating

1232 B.S. Witkowski

the P3HT polymer. A maximum PCE of ≈ 0.6% wasachieved. There was observed enhancement of the re-sponse in the blue–near UV region through absorptionby the ZnO NRs.

Very interesting results were published on theuse of ZnO NRs grown on reduced graphene ox-ide (rGO) [76]. This structure was used to fab-ricate inorganic–organic hybrid solar cells witha layered structure of quartz/rGO/ZnO NR/P3HT/poly(3,4-ethylenedioxythiophene) (PEDTOT):poly(styrenesulfonate) (PSS)/Au. The observed PCE of0.31% was higher than that reported in previous solarcells by using graphene films as electrodes.

In 2010, a structure was presented using P3HT:phenyl-C61-butyric acid methyl ester (PCBM) as an active poly-mer blend, which was found to be compatible with a pre-pared ITO substrate base [77]. The structure ITO/ZnOfilm/ZnO NR/P3HT: PCBM/Ag showed a relativelyhigh efficiency of 2.44%.

Also in 2010 a work was published presentinghybrid bulk heterojunction PV cells based onZnO nanorods blended with P3HT [78]. Tetra(4-carboxyphenyl)porphyrin (TCPP) molecules weregrafted onto the surface of ZnO NRs to enlarge theabsorption spectrum of the blend. However, the overallphotocurrent of the device decreases gradually withTCPP concentration at the ZnO NR surface. TCPPgrafting is beneficial for additional photocurrent gen-eration in the P3HT:ZnO blend but introduces strongmodification of the blend morphology and charge carrierdynamics at the P3HT/ZnO interface, which finallyreduces the overall photocurrent generation.

In the next year a method was shown for the enhance-ment in photovoltaic performance of hybrid solar cellsconstructed from ZnO NRs and P3HT by Li incorpora-tion to ZnO crystal [79]. The oxygen-enrichment of theZnO surface caused by Li doping, observed by red PLemission, yields the effective charge transfer at the inter-face. This leads to the increase of JSC . The obtainedimprovement of VOC was obtained. However, furtherLiZn leads to increased reverse current densities of mi-nority carriers decreasing VOC after the maximum valueat 5 at.% incorporation. The maximum power conversionefficiency of 0.37% was obtained at 5 at.% doping.

In recent years, work in these areas has been continued,but no breakthroughs have occurred.

4.5. Silicon based PV cellsOne of the most promising applications of ZnO NRs

is their inclusion in the architecture of silicon-basedPV cells. In recent years, there has been many workson this subject, and the results are extremely interest-ing. The first advantages were connected with anti-reflectance properties of an NRs array. In 2012 a pa-per was published on the growth of ZnO NRs on aZnO/textured c-Si photovoltaic device through the hy-drothermal method [80]. The reflectance spectra andI–V characteristics indicated that the ZnO/textured c-Si photovoltaic device with ZnO NRs had the lowest

reflectance among the tested devices, especially in therange of UV and green light (350 nm to 590 nm). Com-pared to SiNx/textured c-Si and ZnO/textured c-Si pho-tovoltaic devices, the proposed device exhibited PCE im-provements of around 7% and 6.3%, respectively.

In the next year, a novel approach was introduced ofusing transparent conducting oxide (TCO) NR arraysas a replacement for conventional textured TCO lay-ers in the configuration of silicon p–i–n solar cells [81].Enhanced PCE and improved I(V )-characteristics wereshown from 5.9% to 6.7%.

In the same year, the use of ZnO NRs was demon-strated as an anti-reflecting (AR) coating on planar andtextured, industrial scale Si solar cells [82]. For planarand textured single-crystalline Si solar cells, photovoltaicefficiencies of 7.5% and 11% have been obtained withoutARs, while these values were improved to 10% and 12.7%through the growth of ZnO NRs, respectively. An almostone order of magnitude improvement in the minority car-rier lifetime indicated the passivation effect of ZnO NRs.

A similar solution was under further investigation.In 2016, a multi-crystalline solar cell with two func-tional layers — ZnO nanostructures and an Al2O3 spacerlayer was demonstrated [83]. Results show that theZnO nanostructures synthesized from the seed layer withoptimized thickness exhibit excellent photon harvestingability in the visible light range, leading to an increasein short circuit current of the PV cells. Eventually, amaximized conversion efficiency of 15.8% was achievedby the multi-crystalline solar cell with the optimizedAl2O3/ZnO nanostructures stack, which exhibit com-parable performance to the previously reported single-crystalline Si cell based on ZnO nanostructures with ahighest PCE of 16.0%.

In 2014, a first paper was presented about ZnO NRsused not only as an AR layer, but also as a junctionpartner with Si in PV cell [84]. The structure consistsof p-type Si substrate, ZnO NR as n-type partner andAZO (Al doped ZnO) film (grown at a low depositiontemperature in the ALD process). A wide spectral rangeof the absorption (from 350 nm to 1200 nm) was ob-served. The wide working spectrum and the good junc-tion quality led to 3.6% quantum efficiency. The inves-tigated PV structures are very cheap and easily con-structed. Of course, NRs 3D morphology also leads toincreased light trapping. In this work, the authors useda new method for the growth of ZnO NRs, which pro-duces very high quality NRs despite the very fast (sev-eral minutes) and low-temperature growth process [19] —see Figs. 1 and 2.

The same group published another article in the fol-lowing year — by minor changes in the technology theywere able to obtain almost 11% of efficiency [85]. Theonly change was filling the spaces between NRs with ZnOdeposited by the ALD method (this method was usedin the previous approach to grow an AZO layer on theZnO nanorods). Thanks to this the junction area wasmore continuous and an improvement in efficiency was

Applications of ZnO Nanorods and Nanowires — A Review 1233

Fig. 1. Scanning electron microscopy images of ZnONRs on silicon substrate.

Fig. 2. Room-temperature cathodoluminescence spec-tra of ZnO NRs on silicon substrate.

obtained. Two years later, the same group published an-other important improvement — the ZnO layer, whichpreviously has been grown on NRs (between NRs andAZO layer) was changed to a ZnMgO layer (also grownby the ALD method) [86]. Due to this simple change,the efficiency was increased to 14%, and this result wasobtained with point contact and without optimization

of the thickness of silicon substrate. The results pre-sented there indicate that the ZnO/Si heterojunction so-lar cell can potentially be a less expensive high-efficiencystructure that can be fabricated by an environmentallyfriendly method. What is important, all technologiesused in this approach are very inexpensive and easy scal-able to industrial size.

There were further works on a similar structures (p-type Si and n-type ZnO-NRs), that focused on the ap-plication of metallic NPs to use plasmonic effects to sup-port the process of light absorption and separation ofcarriers [87–89]. However, the final results did not showhigher efficiencies than previously reported for the beststructures of this type.

5. Photocatalytic applications

Photocatalysis is a light-induced catalytic processwhereby photogenerated electron–hole pairs in a semi-conductor undergo redox reactions with molecules ad-sorbed onto the surface, thereby breaking them intosmaller fragments. Photocatalysis with metal–oxide–semiconductor nanostructures has been an area of in-tense research over recent years with TiO2 [90]. Opticalabsorption in the deep ultraviolet region and low photo-efficiency [91] are factors that hold the wide scale use ofTiO2 for photocatalytic activities under sunlight. ZnO,with a high surface reactivity owing to a large numberof native defect sites coming from oxygen nonstoichiom-etry, has found to be an efficient photocatalyst material.ZnO exhibits comparatively higher reaction and miner-alization rates [92]. Surface area and surface defects playan important role in the photocatalytic activity of metal–oxide nanostructures, as the contaminant molecules needto be adsorbed onto the photocatalytic surface for theredox reactions to occur. The higher the effective surfacearea, the higher the adsorption of target molecules willbe, leading to better photocatalytic activity. One dimen-sional nanostructures, such as NRs, offer a higher surfaceto volume ratio compared to nanoparticulate coatings ona flat plate. In 2014, important work was published onthe effect of aspect ratio and surface defects on the pho-tocatalytic activity of ZnO NRs [93]. The studies showedthat as-synthesized ZnO NRs have good photocatalyticperformance (degradation of methylene blue (MB)), andthe photocatalytic activity increased with increasing as-pect ratio. The high aspect ratio of ZnO NRs exhibit ahigh level of surface defects, which is the reason for thehigh photocatalytic performance. Photocatalytic activ-ity comparable to that of doped ZnO was also observedwith engineered defects in ZnO crystals achieved by fastcrystallization during synthesis of the NPs [94]. Compar-ative results of photocatalytic degradation studies on MBwith visible light irradiation demonstrated that ZnO NRsare 12–24% more active than nanostructured films. Anenhancement of 8% in the photocatalytic activity of ZnONRs was achieved through creation of oxygen deficientstructures using a fast crystallization process achievedby microwave assisted hydrolysis.

1234 B.S. Witkowski

Due to a wide band gap of 3.37 eV, poor pho-ton absorption in visible spectra region of ZnO lim-its its application as an in visible light photocatalyst.In order to shift the optical absorption of ZnO intothe visible region, one possible approach is to dopetransition metal ions into ZnO photocatalyst. Thisidea was realized by different dopants, i.e. Co dop-ing [95], La [96], Ni [97], Cr [98], Cu [99], Hg [100]or Ce [101]. In all these examples, doped NRs ex-hibit higher photocatalytic activity than similar NRswithout dopants.

Another interesting approach was based on the combi-nation of ZnO NR and graphene (G) sheets. In 2013, apaper was published on the sandwich-like ZnO/G/ZnOheterostructures with much higher photocatalytic activ-ity under UV light irradiation than that of comparableZnO NRs, with a photocatalytic degradation rate of MBthree times faster than that of the ZnO NRs [102]. Still,the low efficiency of photocatalytic activity for the visibleregion was not solved. One year later, a similar approachwas presented with sensitization of ZnO NRs on rGOwith CdS NPs on ZnO NRs [103]. This approach helpedto achieve an increase in the photocatalytic performanceof the synthesized composites toward degradation of MBin visible light. The idea to sensitize NRs to visible lightby adding NPs was presented. For example, in 2011 apaper was published on the use of CuO NPs to increasephotocatalytic performance [104] and in 2014 the use ofNb2O5 NPs led to excellent photocatalytic performanceon the degradation of phenol [105]. Photocatalytic mech-anism investigations demonstrated that the degradationof phenol under natural sunlight was mainly via OH rad-ical oxidation mechanism. The improvement in photo-catalytic activity was attributed to enhancement of thesunlight absorbing capability stemming from a good in-terface connection between two semiconductors with ap-propriate band structures so that charge transfer couldproceed efficiently.

An alternative approach presented in the literature isbased on coating the NRs with metallic NPs. The firstpaper on this solution appeared in 2006. Photodegra-dation of methyl orange (MO) under 365 nm irradiationwas performed to investigate the photocatalytic activ-ities of the Au NPs/ZnO NRs [106]. Enhancement ofthe photocatalytic activity for degradation of MO wasachieved by loading Au nanoparticles with sizes smallerthan 15 nm. The effect is more distinct as the size ofthe Au NPs is reduced to 5 nm. The charge separationability of the Au NPs may strongly depend on their sizes,resulting in various photocatalytic activities heterostruc-tures. This may be related to the scattering of the inci-dent irradiation by the Au NPs and the formation of theSchottky barriers at the junction of the ZnO NRs andthe Au NPs. In the following years, similar approacheswere presented in the cases of Pt NPS [107] and AgNPs [108]. In both cases, NPs size and density must bealigned, otherwise the decrease of photocatalytic activitymay be observed.

6. Field emission

Thin films of nanostructures with high aspect ratio,such as nanotubes and NRs, are considered to be idealFE electron sources that can emit electrons at low elec-tric field. Among these materials, CNT thin films havereceived much attention. However, relatively few stud-ies have been carried out on the FE from nanostructuredthin films made of other materials such as ZnO NRs.With structural properties similar to CNT, such as ahigh aspect ratio and stability in vacuum environment,it is possible that ZnO NRs could have excellent electronemission properties.

In 2003, the possibility to enhance the FE from a tung-sten tip with ZnO NRs was demonstrated. ZnO NRs weregrown on a sharp tungsten tip as a FE electron sourceby a simple and effective vapor-transport method [109].The results were similar as obtained for CNT.

The first systematic research on ZnO NRs as indepen-dent emitters began in 2002 [110]. In this work verti-cally well-aligned ZnO NRs were grown by the vapor de-position method at a low temperature of 550 ◦C, usingCo as catalyst. The turn-on voltage for the ZnO NRswas found to be about 6 V/µm at current density of0.1 µA/cm2. The emission current density from the ZnONRs reached 1 mA/cm2 at a bias field of 11 V/µm, whichcould provide sufficient brightness as a field emitter in aflat panel display. Therefore, the well-aligned ZnO be-came a research subject concerning the application of aglass-sealed flat panel display in the near future (eventhough the threshold voltage for FE is still higher thanthat of CNTs).

In 2003, it was shown that the FE from the randomlyoriented ZnO NRs can be significantly improved by re-ducing the NRs areal density [111]. It was shown thatthe same screening effect observed on CNT field emit-ters also affects the FE from ZnO NRs. Thin films withthe lowest areal density of ZnO NRs showed much bet-ter FE characteristics, comparable to those of CNT. Theelectric field required to obtain the FE current densityof 1 mA/cm2 from a ZnO NR thin film with very lowareal density was 6.46 V/µm, which is comparable to theresult from CNT field emitters.

Similar results were obtained by different groups [112].The turn-on field of the ZnO NRs was about 6.2 V/µmat a current density of 0.1 µA/cm2, and the emissioncurrent density reached 1 mA/cm2 at an applied field ofabout 15 V/µm.

It is well known that FE mainly depends on tip mor-phology and the density of NRs. Such good FE prop-erties resulted from the small geometry of ZnO NRsand shape engineering. In 2003, well-aligned arrays ofZnO nanoneedles were fabricated using a simple vaporphase growth [113]. The diameters of the nanonee-dle tips are as small as several nanometers, which ishighly in favor of the FE. FE measurements using thenanoneedle arrays as a cathode showed emission cur-rent density as high as 2.4 mA/cm2 under the field of7 V/µm, and a very low turn-on field of 2.4 V/µm.

Applications of ZnO Nanorods and Nanowires — A Review 1235

Such a high emission current density was connected withthe high aspect ratio of the nanoneedles.

The importance of the shape was also confirmed bylater studies [114]. Standard ZnO NRs with a diameterof 150 nm showing poor FE performance were modified(by reducing the tip radius of NRs to 5 nm). The secondmodification was decorating the NRs with small metalNPs. Each of these two approaches produced excellentFE performance, typically turn-on field (the electric fieldat which the current density reaches 10 µA/cm2) as smallas 2 V/µm and current density as large as 1 mA/cm2.For the validity of the concept of decoration with metalparticles, the metal should be stable against oxidationand have a work function smaller than that of the emit-ting material. The larger the work function of the emit-ting materials, the more effective the electron subsidy. Itwas shown that there is a size range within which metalNPs decoration works in enhancing the FE performanceof ZnO nanomaterials.

The influence of metallic NPs has been observed be-fore. In work [115], ZnO NRs synthesized by the vapor-liquid-solid growth mechanism with Cu and Au as thecatalyst were investigated. The Cu-catalyzed ZnO NRswith a high-quality wurtzite structure were grown verti-cally on p-type Si (100) substrate. These ZnO NRs showexcellent FE properties with turn-on field of 0.83 V/mmand current density of 25 µA/cm2. The emitted cur-rent density of the ZnO NRs is 1.52 mA/cm2 at a biasfield of 8.5 V/µm. Cu-catalyzed ZnO NRs exhibited ahigher FE area factor of about 7.18 × 103 and a largerfield adjustment factor of 1.21× 10−5 than the values ofAu-catalyzed NRs, which is due to the vertical directiongrowth of Cu-catalyzed NRs.

The role of metallic structures remained a subject ofresearch. The work [116] shows investigations on FE fromZnO NRs grown on polystyrene and polyethylene foils. Itwas shown that a few-nanometer thick Au film on top ofthe ZnO NR lowers the electric field for electron emissionsignificantly while gold films thicker than 25 nm havelittle effect on the FE properties. This is explained byadditional field enhancement due to Au islands that areformed on top of the ZnO NRs.

In 2006, there were very interesting investigationsabout FE behavior of the ZnO NRs after thermal anneal-ing in a different environments [117]. The Raman and PLresults indicate that the crystal structure was improvedby annealing in oxygen due to oxygen vacancy concen-tration reduction. Surface states correlative with oxygenvacancy in ZnO NRs can play a key role in determiningthe FE performance. The sample annealed in oxygen hadbetter FE properties due to its better crystal structure,lower work function, and increased conductivity.

In 2004, two papers (by the same authors) appearedon the subject of FE from ZnO NRs grown on carboncloth [118, 119]. Extremely good results were obtained:an emission current density of 1 mA/cm2 at an operat-ing electric field of 0.7 V/µm. A comparative study ofFE from NRs of various morphologies and site densities

led to the conclusion that length, site density, and sub-strate geometry are vital factors in determining the FEproperties.

In 2011, there were reports on studies of ZnO NRsgrown on G as field emitters. This is extremely attrac-tive for flexible applications, such as flexible displays andsolar cells. In work [120] a hybrid material was demon-strated consisting of vertically aligned ZnO NRs grownhydrothermally on reduced G/PDMS substrates. TheFE of this hybrid material showed low turn-on field valuesof 2.0 V/µm, 2.4 V/µm, and 2.8 V/µm (for convex, flat,and concave deformations, respectively). Such variationsof FE properties were attributed to the modification ofZnO NRs emitter density upon mechanical deformation.The convex bending exhibited one of the lowest valuesof 2.0 V/µm, due to the reduced screening effect fromneighboring field emitters.

Similar results were obtained in 2013. In work [102],ZnO NR arrays were grown on a single side of flexi-ble reduced graphene sheets (rGs) forming two-layeredheterostructures of ZnO/G. Due to the outstanding me-chanical and electrical properties of the rGs, two-layeredZnO/G heterostructures were demonstrated to have ex-cellent FE properties (turn-on field as low as 2.1 V/µm,the emitting current 470 µA/cm2 at 3 V/µm).

Although commercial field emitters are not built on aZnO basis, it is believed this is a matter of time as ZnONRs are one of the most promising nanostructures andsoon we will see their practical applications extent to FEapplications as well.

7. Field effect transistor

ZnO NRs, because of their unique electrical proper-ties, are under investigation as an attractive candidatefor a broad range of applications in new types of trans-parent electronics that operate with low power require-ments, i.e. in FET.

The first true FET structure based on single ZnO NRsas a channel was fabricated using NRs grown by siteselective MBE [121]. In the dark, this depletion-modetransistor exhibited good saturation behavior, a thresh-old voltage of −3 V, and a maximum transconductanceof order 0.3 mS/mm.

In the same year, a work was published on bottom-upintegration of ZnO NRs to obtain a vertical surround-gate FET [122]. The vertical device architecture has thepotential for use in ultra-high-density nanoscale memoryand logic devices. More significantly, scaling of the verti-cal channel length was lithographically independent anddecoupled from the device packing density. A bottomelectrical contact to the NR was uniquely provided by aheavily doped underlying lattice-match substrate.

Also in 2004, a very interesting result was presented— by simple polymer passivation of n-channel ZnONR-based FET significantly improved its characteristics,and increased the current on/off ratio from 7 to 104–105 and also increased transconductance from 140 nSto 1.8 µS [123]. The electron mobility estimated from

1236 B.S. Witkowski

transconductance exhibited a maximum value of 1000–1200 cm2/(V s), which confirms the practicability ofsemiconductor oxide-based NRs for electronic nanodeviceapplications.

After 2 years, in 2016, a much better FET based onZnO NR was presented [124]. The applied SiO2/Si3N4passivation processes proved to significantly enhance de-vice performance in the subthreshold swing, on/off ra-tio, and mobility. On average, the field effect mobilityof ZnO NR FETs is dramatically increased by two or-ders of magnitudes, with the maximum measured mobil-ity exceeding 4000 cm2/(V s). The origin of the supe-rior FET performance was attributed to passivation ofsurface defect states, which act as scattering and trap-ping centers, and reduction of surface adsorption pro-cesses at oxygen vacancy sites. Since mobility determinesthe carrier velocity and thus the switching speed, theseNR FETs are of great potential for future applicationsin high-frequency integrated electronics, such as memoryelements and logic gates.

Also in 2016, there was a very interesting work on theuse of a ZnO NR-based transistor as a force detector inthe nanoscale [125]. A new type of device, a piezoelec-tric FET composed of a ZnO NR bridging two ohmiccontacts, was demonstrated, in which the source to draincurrent is controlled by the bending of the NR. A possibleexplanation of this phenomenon is suggested by associ-ation with the unique semiconducting and piezoelectriccoupling effect of ZnO. This device was demonstratedas a force sensor for measuring forces in the nanonew-ton range and even smaller with the use of smaller NRs.An almost linear relationship between the bending cur-vature and the conductance was found at small bend-ing regions, demonstrating the principle of NR-basednanoforce sensors.

In 2008, a work was published on the high-current driv-ing multiple ZnO NRs FETs [126]. The devices were sim-ply formed by assembling ZnO NRs on a Si substrate us-ing an optimized AC dielectrophoresis technique. Thesemultiple ZnO NRs FETs exhibited very good electricalcharacteristics with a transconductance of 311.5 µS at adrain voltage of 15 V. This approach could find uses ina number of applications in the field of high-frequencyintegrated electronics.

ZnO is also very known for its sensor properties, i.e.the great influence of the chemical state of the sur-face (adsorption and desorption of multiple molecules)on electrical properties, which is particularly visible inthe case of nanostructures, where the depleted layer cancover the entire volume of structures. Therefore, it is wellsuited for creating sensors based on a transistor with anopen gate (based on the ZnO matrix) subjected to chem-ical coefficients.

The first demonstration of this idea was presentedin 2004 [127]. n-type single crystal ZnO NRs were syn-thesized by a vapor trapping CVD and fabricated intoNR FETs. O2 adsorption on the NR surface was shownto have a considerable effect on the measured conduc-

tance, and an O2 sensing study showed that the sensi-tivity depends on NR diameter as well as applied gatevoltage. In the following years, FET-based sensors werepresented for different chemical compounds; e.g. in 2009it was shown that the functionalized ZnO NRs on the sur-face of a silver wire connected as an extended gate area ofan ordinary MOSFET can be used for the electrochem-ical detection of Ca2+ ions [128]. In 2013, the solution-gated FET-based sensors for the detection of cholesterolwere demonstrated [129]. The FET sensors showed agood sensing performance in terms of wide-linear rangeand high sensitivity with a detection limit of ≈ 0.05 µM.In 2014, a FET hydrazine sensor was fabricated based onsolution grown vertically-aligned ZnO NRs. In the caseof ZnO NRs, however, the FET structure is not requiredfor high-sensitive sensors, which allows sensor structureto be greatly simplified. These applications are presentedin the next section.

8. Chemical and biological sensors

Oxide semiconductor gas sensors exhibit a conductanceafter adsorption of oxidizing gas or the electrochemicalinteraction between the reducing gas and adsorbed oxy-gen with a negative charge. The main strengths of oxidesemiconductor gas sensors are high sensitivity, cost effec-tiveness, and reliability, all of which originate from theirsimple chemisorption sensing mechanism. However, theoxidative or reductive interactions of gases with the ox-ide semiconductor surface are often insufficient to recog-nize and quantify a specific gas. Thus, over the previousdecades, many efforts have been directed at achieving gasselectivity as the addition of precious metals, oxide cat-alysts or control of sensor temperature or the use of aselective filtering layer. This chapter contains review ofdifferent approaches and sensor architectures.

8.1. Detection of reducing gases

ZnO is a typical n-type semiconductor and its con-ductivity is influenced by the surface depletion layer,especially for nanostructures such as NRs. Therefore,the mechanism of a ZnO NR gas sensor is based on thechange in resistance. The target gas molecules adsorbonto or desorb from the working material surface, caus-ing resistivity changes, that is the sensor response andrecovery process. When the sensor is placed in air, oxy-gen molecules will be adsorbed on the surface of ZnOand capture electrons from the surface layer and changeinto oxygen species (O−2 , O

− and O2−) creating deple-tion layer on the surface. The concentration of elec-trons, as the majority carrier, is very low and the re-sistance is relatively large. When some reducing com-pound, e.g. ethanol is injected into surface of NRs,oxygen species (O−2 , O

− and O2−) will react with thereducing gas [130]:

C2H5OH+ 6O− → 2CO2 + 3H2O+ 6e−.

Oxygen species react with ethanol and release electronsback to the conduction band of ZnO. The width of the

Applications of ZnO Nanorods and Nanowires — A Review 1237

electron depletion layer decreases, leading to decrease inthe sensor resistance. This is the main sensing mecha-nism of ZnO-based sensors.

The first demonstration of ZnO NRs as a highly sensi-tive gas sensor was published in 2004 [131]. The sensorswere fabricated with the microelectromechanical system(MEMS) technology. High sensitivity and fast responsetime were found at a working temperature of 300 ◦C. Thenext paper on this subject was published in 2005 [132].Presented in 2006 was a sensitivity of 18.9 to 200 ppm ofethanol, response time and recovery times were 10 and30 s respectively [133]. In the same year there was pre-sented sensor based on ZnO NRs working at 350 ◦C thatcan detect 1 ppm of ethanol [134]. As the above ex-ample shows, the detection mechanism is mainly depen-dent on surface conditions that change with temperature.This way, the sensitivity is related with growth technol-ogy. Therefore, the selectivity of this type of sensors (ofcourse, apart from the examples of specific functionaliza-tion) can be obtained by selecting the operating temper-ature. Also in 2006 a paper was presented demonstrat-ing nanorods, which exhibit good sensing characteristicsfor 1000 ppm of ethanol at or below 150 ◦C [135]. Fur-thermore, the sensitivity remained essentially the sameup to 1000 cycles. This application was under intenseinvestigation over the next year. In 2007, sensors werepresented based on the ZnO NRs with a sensing responseto 1 ppm ethanol of 3.2 [136]. The response times andthe recovery times for ethanol gas were no more than 10 swhen operating at 450 ◦C. In the same year, a very in-teresting study of the size dependence of gas sensitivityof ZnO NRs was presented [137]. While the high surfaceadsorption of oxygen is the key to obtaining excellent gassensing performance by the sensor, it was found that thethinner ZnO NR sensors have significantly better sens-ing performance due to the larger quantity of electrondonors related with oxygen vacancies (VO) and larger ef-fective surface areas, resulting in a larger quantity of ad-sorbed oxygen than the thicker NRs sensors. Therefore,the thin well-aligned ZnO NRs would be of great poten-tial for further practical applications in gas sensors. Onlyone year later, another paper described well aligned ZnONRs with 20 nm diameter at the edge of terrace stepson the sapphire substrate [138]. This sensor exhibitedthe high surface-to-volume ratio of 58.6 µm−1 and thesignificantly high sensitivity of about 10 even for the lowethanol concentration of 0.2 ppm in room temperature.In the following years, several papers were presented onthe use of pure ZnO NRs as an ethanol sensor [139–142].

In addition to modifying the size of the NRs, amplifi-cation of the sensor response can be obtained by addingmetal NPs to the surface of the NRs. Metal NPs onNRs could increase the adsorption of oxygen moleculesdue to its spillover effect. This chemical mechanism ofAu NPs will cause more electrons to be trapped and theelectron depletion layer will be thicker. It will generatemore oxygen species and more reactive points comparedwith the pure ZnO. The first practical use of this solu-

tion was published in 2007 [143]. The sensors fabricatedfrom In-doped ZnO NRs exhibited high sensitivity andvery fast response and recovery times (≈ 2 s). In thesame year, a paper was presented on high-density singlecrystalline ZnO NRs with Pd NPs adsorbed on NR sur-faces [144]. Compared with pure ZnO NRs, it was foundthat the sensor response of the ZnO NRs with Pd ad-sorption was much larger. Similar results were presentedon Ag NPs embedded-ZnO NRs [145]. Gas-sensing prop-erty measurements had shown that, compared with thepure ZnO, the synthesized Ag nanoparticle embedded-ZnO NRs had an enhanced ethanol-sensing property. Fi-nally, papers were published describing sensors based onAu-functionalized ZnO NRs [130, 146]. The Au/ZnO sen-sor manifests high response, fast response–recovery timeand good repeatability, resulting from the combinationof 1D nanostructure with unique properties of Au NPs.The superior sensing features indicate that the presentAu/ZnO NRs are promising as gas sensors.

Another approach is related to the change of ZnO bandstructure through doping. In 2008 there were presentedsensors based on the ZnO and Au-doped ZnO NRs [147].It was found that the sensitivity of the sensor based onAu-doped ZnO NRs exhibits higher values than that ofthe sensor based on undoped ZnO. Other authors usedCo [148] or Nd ions as a dopant [149]. In the lattercase, the gas-sensing properties of ZnO NRs were im-proved. The gas response of pure and 2 at.% Nd-dopedZnO NRs to 100 ppm ethanol vapor were 20.3 and 37.8,respectively.

Very interesting results were obtained for ZnO NRsin conjunction with graphene (G). In 2011, a flexiblegas sensor was presented using hybrid vertically grownZnO NRs and free-standing G sheets [150]. This gas sen-sor enabled ppm level detection for ethanol vapor. Theresult suggests that the combination of 1D nanocrys-tals and two-dimensional G improves the performanceof the sensors and imparts additional mechanical or op-tical functions to the devices. Studies on this subjectwere continued in the subsequent years, and in 2013 apaper was published with an analysis of single and dou-ble side ZnO/G structure [102]. When the sensors areexposed to reducing gases, they can catalytically reactwith the adsorbed oxygen, freeing electrons to return tothe ZnO surface and increasing the sensors’ conductivity,thus, gases are detected. Gas molecules adsorbing on theactive surface benefit from the ZnO/G array structures,which will simplify molecular absorption, gas diffusion,and mass transport. Direct growth of the NRs onto therGs substrates, highly aligned ZnO nanorods offer quan-tities of active centers and efficient electron pathways foramperometric sensors. Electrons flow from the rGs tothe ZnO NRs, which causes the surface of the ZnO NRsto become more n-type than at the original state. TherGs substrates create a contact at the interface with ZnOwhich reduces the energy barrier between the gas and therGs and make the electron transfer occur more easily. Inaddition, due to the high charge carrier mobility of rGs,

1238 B.S. Witkowski

it provides direct conduction paths for carriers to trans-port from the junction to the external electrode, and thusthe electrical signals link closely and propagate rapidly.

Almost all the above examples presented a simpleethanol sensor, based on resistance measurement. Other,more sophisticated configurations are also possible. Forexample, in 2011 an ethanol sensor was presented using ap-type Co3O4-decorated n-type ZnO NR network sensorvia the formation of p–n junctions [151]. The less ag-glomerated and porous network structures facilitate notonly the rapid and effective diffusion of analyzed gasesto the entire nanostructured surface but also the uni-form loading of a sensitizer. Therefore, the coating ofa discrete configuration of p-type semiconductors ontoan n-type semiconductor NRs provides a simple, com-prehensive, and powerful tool to control both the gasselectivity and the gas sensitivity. Two years later, awork was published describing synthesis of highly densep-ZnO NPs/n-ZnO NRs on a glass substrate [152]. Thereaction produced more electrons at a higher tempera-ture. These electrons increase the conductivity of n-typeNR and reduce that of p-type NPs. The gas responses ofp-ZnO NPs/n-ZnO NRs are dominated by p-type sensingat 25 ◦C and n-type sensing above 200 ◦C.

The commercially suitable approach was presentedin 2010. It was an NRs based detector grown on siliconon insulator (SOI) complementary metal oxide semicon-ductor (CMOS) membranes [153]. The basic gas sensordevice is a micro-hotplate structure containing a siliconmicro-heater embedded in a thin dielectric membrane,which allows us to achieve high temperatures with lowpower consumption. The sensors were exposed to dif-ferent concentrations of ethanol vapor in air (relativehumidity 3000 ppm) and the sensitivities of 0.2%/ppmand 0.1%/ppm at 809 ppm and 4563 ppm, respectively,were observed. The measurements were repeated atfour different humidities and it was found that humid-ity has a significant effect on ethanol detection overthe tested range.

All the above detectors were presented as ethanol sen-sors, but of course a similar mechanism works with otherreducing gases, e.g. H2 gas. In 2005, a ZnO NRs-baseddetector sensitive to hydrogen at temperatures higherthan 100 ◦C was presented [154]. In recent years, therehave been few papers on the detection of H2 by pure ZnONRs [155–157].

Interesting results on H2 sensing were obtained by us-ing single NR detectors. In 2008, paper was publishedon fabrication of ZnO NRs using nanoscale spacer lithog-raphy gas-sensing properties to H2 in the concentrationrange of 500–5000 ppm [158], but the technologies usedrather preclude commercial application. Presented twoyears later was a single ZnO NRs as a sensing materialin nanosensors with a response to hydrogen gas at roomtemperature [159]. The same year, a study was presentedon the influence on response and selectivity of the sur-face defect concentration and the diameter of ZnO NR insingle NR-detector at room temperature [160].

Another method to obtain selectivity to H2 at roomtemperature is coating Pt or Pd NPs on the surface ofNRs to enhance the sensing mechanism. These NPsappear to be effective in catalytic dissociation of theH2 to atomic hydrogen. In the presence of hydrogen,chemisorbed oxygen ions are reduced to release electronsto the conduction band of the ZnO surface resulting in anincrease in conductivity. In 2005, two papers were pub-lished by the same authors regarding Pt and Pd coatedZnO NRs detectors [161, 162]. These detectors were well-suited to detection of ppm concentrations of hydrogenat room temperature. The recovery characteristics werefast upon removal of hydrogen from the ambient. Butthe slow response of the sensors at room temperaturewas a major issue to be solved. In 2006, another pa-per was published on the hydrogen sensing characteris-tics of different types of ZnO nanostructures, with andwithout impregnation of Pt [135]. The sensitivity to H2sensors was upgraded in comparison to previous works.In 2010, a work was published on growth of ZnO NRson fabric and their use as a hydrogen sensor [163]. Theresponse and selectivity were enhanced and tuned withadditives such as Pt-catalyst. The morphology and elec-trical characteristics of the NRs-on-fabric device werefound to be robust against mechanical handling. Thelow cost and scalable fabrication process, room temper-ature sensing capabilities are very promising for futureapplications.

Similar mechanisms and solutions can be applied incase of detection of liquefied petroleum gas (LPG). Thefirst information about the use of ZnO NRs to detectLPG was published in 2005 [164], where in order to en-hance the sensitivity and selectivity to LPG, NRs werecoated with PdO. The same solution was used in furtherstudies [133]. In subsequent years also presented was theuse of Pd coating to enhance the reaction to LPG [165].The Pd-sensitized vertically aligned ZnO NRs were moreselective to LPG in the presence of CO2 than verticallyaligned ZnO NRs. In 2014, a paper was published pre-senting studies of the effects of aspect ratio of ZnO NRson the LPG sensing properties [166]. The LPG responseincreases along with the increasing aspect ratio of ZnONRs. The high surface adsorption of oxygen is the im-portant factor in the gas sensing phenomenon. It wasfound that sensors based on ZnO NRs with high as-pect ratio provide high effective surface area and a largernumber of electron donors related to VO, resulting in alarger number of oxygen adsorbed than ZnO NRs withlow aspect ratios.

Recently, it was shown that ZnO NRs can also be ap-plied to detect coal mine methane (CMM) [167]. It wasshown that under UV irradiation, the operating temper-ature decreased from 225 ◦C to 100 ◦C for ZnO NRs. Thesensor at room temperature (25 ◦C) under UV excitationshowed an acceptable response (≈ 23%) which makes itsuse in coal mines possible.

An example of another chemical compound which canbe detected by ZnO NRs is H2S. There are several

Applications of ZnO Nanorods and Nanowires — A Review 1239

reports in the literature about detection of this com-pound [134, 137, 164], also with use of Pd nanoparti-cles onto the surfaces of ZnO NRs [168], which leads tofabrication of H2S sensors with a highly enhanced re-sponse and selectivity. In 2012, it was shown that se-lectivity to H2S can be achieved by selecting the detec-tor’s operating temperature [169]. At 50 ◦C, the responseto H2S was much higher than to LPG, CO, CO2, O2,ethanol, and H2.

Similar results were shown for carbon monoxide. Thereare many papers on the use of ZnO NRs to detectCO [155, 164, 170]. In paper [141] an example waspresented how to increase selectivity by selecting oper-ation temperature. Other methods reported for increas-ing sensitivity and selectivity to CO were Co-doping ofNRs [148] or coating NPs with gold, which act as a cat-alyst in chemical sensitization and improve the sensingcharacteristics [171, 172].

Another example of a reducing chemical com-pound, which ZnO NRs can be applied to detectis ammonia. There are many papers on simple-architecture structures which present a response toNH3 [155, 157, 160, 164, 173, 174], but particularly inter-esting are studies using ZnO NRs-coated quartz crystalmicrobalance [175, 176]. The as-developed sensor showeda good response to NH3 gas at room temperature and itcould be used to detect low level concentration of NH3gas. Additionally, the sensor has a good selectivity toNH3 gas over various gases such as LPG, N2O, NO2,CO, and CO2.

ZnO NRs are also sensitive to hydrazine. In 2008, thefirst paper was published on an electrochemical hydrazinesensor with a response time less than 5 s and a detec-tion limit of 0.2 µM [177]. One year later, a paper waspresented about an ultra-sensitive hydrazine sensor fab-ricated by modifying a gold electrode with a high-aspect-ratio ZnO NRs grown by a simple thermal evaporationprocess at low temperature [178]. Response time was lessthan 5 s and a detection limit of 84.7 nM was achieved.In 2014, a work on a FET hydrazine sensor based onvertically-aligned ZnO NRs was published [179]. TheFET sensor exhibited good sensitivity with a low limitof detection (≈ 3.86 nM), and a linear dynamic range(1 nM–60 µM). These results, with their low-cost andease of fabrication, show the great potential of ZnO NRsbased detectors.

ZnO NRs can also serve as humidity sensor. Itis known that water vapor will be adsorbed on thesurface of ZnO NRs and reacts reversibly with ZnOas follows [180]:

H2O+ 2Zn⇔ 2(OH–Zn) + 2VO + 2e−,

where VO is the vacancy created at the oxygen site. Witha strong local charge density and a strong electrostaticfield, the ionic Zn2 will induce chemisorption followedby physisorption of water molecules. At high relativehumidity, the naturally n-type ZnO NRs should becomemore conductive due to the increased number of free elec-

trons. Due to high defect concentration, the first lit-erature reports on humidity measurement by ZnO NRswere achieved by doped NRs, i.e. by Cd [181] or Ptdoping [182]. Presented a few years later was a simplehumidity sensor on pure ZnO NRs [183, 184] based onresistivity monitoring. It was also shown that detectionof humidity based on quartz crystal microbalance coatedwith ZnO NRs is possible [185].

8.2. Detection of oxidizing compoundsThe mechanism of the electrical response to oxidizing

gases like O2 or NO2 is very simple. These gases adsorbdirectly on the surface of ZnO NRs. Therefore, the con-centration of electrons on the surface of ZnO nanorodsdecreases and, correspondingly, the resistance of the ZnOlayer increases. The process of the reaction can be de-scribed in the following example:

NO2(gas)+ e− → NO2(ads)−.

The first publication presenting an NO2 detector basedon ZnO NRs was published in 2006 [186]. At the op-erating temperatures of 300 and 350 ◦C, the ZnO NRsshowed high NO2 sensitivity (1.8 for 1 ppm of NO2)with negligible CO sensitivity (1.00–1.02), showing po-tential for use to detect NO2 in the presence of CO. Insubsequent years, a few papers have appeared on theuse of ZnO NRs as an NO2 sensor based on resistivitychanges [187–189], but a very interesting solution waspublished in 2009 [190] — ZnO-NRs gas sensors werepresented, in which NR bridges were self-assembled be-tween two electrodes by a selective growth of ZnO NRson patterned Au catalyst layers. The sensors demon-strated high gas response to NO2 down to the sub-ppmlevel and fast response/recovery behaviors. The gas re-sponse showed a maximum value at 225 ◦C and linearlyincreased with the concentration of NO2 in the range of0.5–3 ppm. The enhanced response of that sensor couldbe related with the formation of a potential barrier atthe NR/NR junctions in addition to the effect of the in-crease in the surface depletion region of each NR and bythe fact that ZnO NRs are floated above the substrateand thus the adsorbed molecules can be easily desorbedfrom the surface of ZnO.

An interesting approach was presented in 2007 [191],in which, next to the electrical response even at low op-erating temperatures, with detection limits lower than200 ppm, optical properties were concerned. ZnO NRsPL spectra were reversibly quenched by the introductionof a few ppm of NO2 at RT. NO2 adsorbed over the sur-face introduces non-radiative recombination paths thatmodify the emission properties of the NRs. These resultswere very promising for gas sensing applications even atroom temperature.

In 2012, a study was published on the role of gold cat-alyst on the sensing behavior of ZnO NRs for NO2 [192].The presence of Au NPs on the surface of ZnO NRs doesnot enhance the response in case of NO2 gas. It wasfound that precious metals are not suitable for the oxi-dizing gases.

1240 B.S. Witkowski

An interesting approach was presented in 2011 [151]— selective detection of NO2 was achieved using a p-type Co3O4-decorated n-type ZnO NRs network sensor(this paper was also presented in part regarding ethanolsensors). The gas selectivity was explained by the cat-alytic effect of nanocrystalline Co3O4 and the extensionof the electron depletion layer via the formation of p–njunctions and the variation of sensor temperature. Thecoating of a p-type semiconductors onto n-type semicon-ductor NRs provides a simple tool to enhance selectivityand the gas sensitivity. One year later, a similar solutionwas published based on ZnO NRs modified with SnO2NPs [193]. A ZnO/SnO2 composite sensor exhibits ahigher response to NO2 gas at room temperature un-der UV light irradiation. The ZnO/SnO2 heterojunctionmodel and the increased photo-generated electrons arebeneficial for testing NO2 gas.

Of course, ZnO NRs can also be applied to detect oxy-gen, the adsorption of which on NRs surface is the firststep to detecting reducing gases. The first report onO2 detection was achieved by FET with NRs as a chan-nel [127]. O2 adsorption on the NR surface was shownto have a considerable effect on the measured conduc-tance, and an O2 sensing study showed that the sensi-tivity depends on NR diameter as well as applied gatevoltage. The following year, a report was presented onthe use of individual ZnO NR transistors fabricated andcharacterized in a vacuum chamber at different oxygenpressures [194]. The transistors exhibited high sensitiv-ity to oxygen, which led to a change in the source draincurrent and a shift of the threshold voltage. It was shownthat the sensing properties of the transistors are relatedto the trapping and releasing of the carriers in the NR.

ZnO NRs appear well suited to detection of O3. It wasshown [154] that they can be sensitive at temperaturesas low as 25 ◦C for percent levels of O3 in O2. The recov-ery characteristics are quite slow at room temperature,indicating that the NRs can be used only for initial detec-tion of ozone. In 2010, a paper was presented on studiesof various ZnO nanomaterials as sensitive to ozone, alsolow-defect ZnO single-crystal NRs [195]. At room tem-perature, responses reach about 1000 with the shortestresponse T = 60 s and recovery times T = 5 s. The fastresponse and recovery time were attributed to the lowdefect density and high electron mobility of the single-crystal nanorods. The sensor response was measured inthe range from 1 to 2.5 ppm ozone. The sensing mecha-nism of the NO2 gas is like ozone gas, but ozone possessesmore oxidative ability than NO2, so the reaction in ozoneis faster than NO2. Thus, the sensor response to ozonewas higher.

8.3. Biological sensors

ZnO NRs have great potential not only in sensingsimple gases, but also for more sophisticated detectors.In 2009, it was shown that the functionalized ZnO NRson the surface of a silver wire connected as an extendedgate area of an ordinary metal-oxide semiconductor FET

(MOSFET) can be used for the electrochemical detec-tion of Ca2+ ions [128]. That same year, a paper waspresented on formaldehyde detection at room tempera-ture [196]. The results revealed that the response in-creases by more than two orders of magnitude to 110 ppmformaldehyde with UV light irradiation. Such a highresponse was supposed to be caused by the photocat-alytic reaction occurring between formaldehyde and ad-sorbed oxygen molecules. In 2010, the possibility waspresented of detecting hydrogen peroxide by modifiedelectrode with a composite film of ZnO NRs and AgNPs [197]. It was shown that the electro-reduction ofH2O2 is a diffusion-controlled process catalyzed by AgNPs. A linear relationship could be obtained betweenthe current density and the concentration of H2O2 up to1 mM, suggesting successful construction of a sensor forthe detection of H2O2 for this concentration range. Inturn, in 2012 hydroquinone detection by Ce-doped ZnONRs was presented [198], and one year later solution-gated FET based sensors for the detection of cholesterolwere demonstrated. The FET sensors showed a goodsensing performance in terms of a wide-linear range andhigh sensitivity with a detection limit of ≈ 0.05 µM. Inaddition, the sensor shows low interference from elec-troactive species and was found to be dependable forcholesterol detection in human serum and whole bloodsamples [129].

The beginning of ZnO NRs in biological sensors startedin 2006, when a paper on immobilization of uricaseon ZnO NRs for a reagentless uric acid biosensor waspublished [199]. It was demonstrated that a three-dimensional assembly with ZnO NRs achieved the di-rect electron transfer of uricase and showed excellentthermal stability. Two years later, detection of DNAsequences using nanoscale ZnO NRs arrays was pre-sented [200]. The authors used covalent and non-covalentlinking schemes in order to attach probe DNA strands tovarious ZnO nanoplatforms. After carrying out a duplexformation reaction with target DNA strands, fluorescenceintensity was measured and compared. The presence ofthe underlying ZnO nanomaterials was critical in achiev-ing increased fluorescence detection of hybridized DNA.This collective approach allowed detection of the targetspecies B. anthracis at sample concentrations as low asa few femtomolar level and, therefore, permitted highlysensitive identification of the biothreat agent.

Also presented in 2006 was an enzymatic glucosebiosensor based on ZnO NRs array functionalized byglucose oxidase (GOx) with a low limit of detection of0.01 mM, a response time of less than 5 s, a linear rangefrom 0.01 to 3.45 mM [201]. These parameters demon-strate that ZnO NRs can provide a favorable microen-vironment for GOx immobilization and stabilize its bio-logical activity. This idea for a glucose sensor was un-der further development in subsequent years [202] andin 2010 a GOx functionalized sensor was presented witha fast response of less than 1 s and a quite wide linearrange from 0.5 to 1000 µM [203]. This sensor was charac-

Applications of ZnO Nanorods and Nanowires — A Review 1241

terized by stability, selectivity, and reproducibility. Thesame functionalization was used in the ZnO NR-gated re-gion of an AlGaN/GaN high-electron-mobility transistorstructure with a very low limit of detection for glucose of0.5 nM [204]. This electronic detection of biomoleculesis a significant step toward a compact sensor chip, whichcan be integrated with a commercially-available handheldwireless transmitter to realize a portable, fast response,and high sensitivity glucose detector.

A paper was presented in 2009 on a trimethylamine(TEA) gas sensor based on ZnO NRs [205]. Well-crystallized ZnO NRs were employed by a facile solutionroute with benzene sulfonic acid sodium salt (DBS) as amodifying agent. The results demonstrated highly sensi-tive and selective to low-concentration (0.001–1000 ppm)TEA among the gases of ethanol, benzene, toluene,and acetone at the operating temperature of 150 ◦C.The response of the ZnO sensor to 0.001 ppm TEAwas as high as 6.

A very interesting approach was presentedin 2015 [206]. Demonstrated was a novel electrochemi-luminescence (ECL) immunosensor for sensitive andselective detection of carbohydrate antigen15-3(CA15-3)by using polyamidoamine (PAMAM)-functionalized ZnONRs as carriers. The ZnO NRs did not only catalyzeH2O2 to produce various reactive oxygen species withan enhanced ECL signal but also had rich binding sitesfor loading large amounts of luminol. This immobilizedluminol was further introduced onto the electrode surfacethrough a sandwich-type immunoreaction resulting inhigher luminol ECL efficiencies. Finally, Au NPs elec-trodeposited on the electrode not only provided a largesurface area for the immobilization of abundant primaryanti-CA15-3, but also facilitated electron transfer withincreased ECL intensity. This procedure may provide apromising way for the simple and sensitive detection ofother proteins as well.

8.4. pH detectorsIn 2005, it was shown that ZnO NRs can also be used

for pH measurements [207]. ZnO NRs show dramaticchanges in conductance upon exposure t