Contents lists available at ScienceDirect

Nano Energy

journal homepage: www.elsevier.com/locate/nanoen

Full paper

Nanotransducers on printed circuit boards by rational design of high-density, long, thin and untapered ZnO nanowires

Giuseppe Arrabitoa,1,2, Vito Erricoa,2, Zemin Zhangb, Weihua Hanb, Christian Falconia,⁎

a Department of Electronic Engineering, University of Rome Tor Vergata, Via del Politecnico 1, 00133, Rome, Italyb School of Physical Science and Technology, Lanzhou University, Lanzhou 730000, China

A R T I C L E I N F O

Keywords:PiezotronicsFlexible microheatersNanotransducersZnO nanowiresPrinted circuit boards

A B S T R A C T

Nanotransducers can offer crucial advantages in comparison with conventional sensors and actuators. However,interfacing and packaging nanostructures into complete electronic systems is very complex. Here we describe awet chemical method for cointegrating arrays of ZnO nanowires into systems on printed circuit boards (PCBs).First, we deposit on the PCB a MnOOH layer for reproducibly increasing the nanowires density. Afterwards, wenumerically demonstrate that the ligand ethylenediamine, at the isoelectric point of the ZnO nanowires tips, caneffectively control, at very low concentrations, both zinc speciation and supersaturation in the nutrient solution.Accordingly, we combine potassium chloride and ethylenediamine, produced in situ from a safer precursor(ethanolamine), for concurrently thinning the nanowires (top width around 60 nm) and stabilizing their topfaces. Our synergic approach enables the solution growth of ZnO nanowires which are untapered and have highdensities (> 8/μm2) and record length (> 15 µm) and aspect ratio (> 200) for plastic substrates which may notwithstand high temperatures. These characteristics permit to simultaneously add the top electrode, package thenanowires, improve their mechanical robustness and connect to the electronic interface by a simple flip chipadhesive bonding procedure. As proofs of concept we describe flexible PCB heaters (electrical to thermal energyconversion) with power densities comparable with state of the art flexible heaters and a wearable piezotronicheartbeat detector comprising both ZnO nanowires (mechanical to electric energy conversion) and electronics ona single flexible PCB. Our results open the way to the cointegration of nanotransducers and electronics onconventional PCBs.

1. Introduction

Nanowires (NWs) offer unique opportunities for electronics, in-cluding energy harvesting [1,2], ring oscillators [3], photosensor arrays[4], low threshold lasers [5], mechanical sensors [6], and electronicskins [7]. However, though NWs can be grown in solution on almostany substrate [8,9], the integration of high-quality NWs into completesystems is still elusive. In fact, the co-integration of devices fabricatedwith different technologies is often an intricate challenge. For instance,only recently a sensor array for the in situ sweat analysis could be fullyintegrated on a Printed Circuit Board (PCB) [10], which is the standardsubstrate for assembling electronic components. In order to embedNWs-based devices in PCBs, for easy packaging [2], it would be possibleto grow vertically aligned arrays of NWs on the PCB copper, deposit apolymer to wrap the nanowires (for improving the mechanical ro-bustness and preventing short circuits [2]) and, finally, add the top

electrode and connections. Zinc oxide (ZnO) NWs are likely the easiestquasi-1D nanostructures to be grown in solution and also have re-markable properties, including piezoelectricity, pyroelectricity andsemiconductivity [11]. However, plastic substrates such as PCBs maynot withstand the high temperature annealing of ZnO seed layers. Thisis an important issue because, in absence of a pre-deposited ZnO seedlayer, conventional wet-chemical procedures for growing ZnO NWs onmetals tend to result in low reproducibility [12,13] and in short, sparseand low aspect-ratio NWs. Moreover, though ammonium hydroxide canincrease the length and density of ZnO NWs directly grown on metals,the resulting NWs are tapered [14]. All these characteristics are pro-blematic as, for instance, short and sparse NWs increase the risks ofshort circuits during packaging and tapered ends hinder good electricalcontacts.

To the best of our knowledge, the integration of ZnO NWs by arational approach on complete packaged PCBs with fully functional

https://doi.org/10.1016/j.nanoen.2018.01.029Received 17 November 2017; Received in revised form 14 January 2018; Accepted 15 January 2018

⁎ Corresponding author.

1 Present address: Dipartimento di Fisica e Chimica, Viale delle Scienze, Parco d′Orleans II, 90128, Palermo, Italy.2 G.A. and V.E. contributed equally.

E-mail address: falconi@eln.uniroma2.it (C. Falconi).

Nano Energy 46 (2018) 54–62

2211-2855/ © 2018 Elsevier Ltd. All rights reserved.

T

devices has not yet been reported. Here we rationally design a wet-chemical procedure for reproducibly growing high-density, long, thinand untapered ZnO NWs on PCBs and then package and interface theNWs to conventional electronics. First, we obtain both high reprodu-cibility and high NWs densities by pre-functionalizing the PCB copperwith a manganese oxohydroxide (MnOOH) layer, similar to an ap-proach previously applied to glass substrates [13]. Afterwards, we se-lect a pH value close to the isoelectric point of the ZnO NWs tips, thuspreventing significant electrostatic interactions between the NWs tipsand ions in solution. Subsequently, we adopt MEA (monoethanolamine)in order to in situ produce the more harmful EDA (ethylenediamine)and, thus, thin the NWs and, at the same time, tune the supersaturationdegree. Finally, since chloride ions may not appreciably change thespeciation and solubility of zinc in solution, we demonstrate that tinyamounts of potassium chloride are already sufficient to effectivelystabilize the NWs tips and, thus, synergistically counter-act the taperinginduced by EDA. This strategy minimizes harmful reagents and con-sistently results in high density (> 8 NWs/µm2), long (> 15 µm in a12 h growth) and untapered NWs which have the highest reported as-pect-ratio (> 200) for NWs grown without using temperatures whichare not compatible with conventional flexible substrates and PCBs.These NWs on a PCB are easy to be packaged and co-integrated withcomplete electronic systems. As proofs of concept, we demonstrateflexible PCB nanoheaters and a flexible PCB comprising both electronicsand an ultra-sensitive ZnO NWs piezotronic sensor for heart-beat de-tection. Our results open the way to the co-integration of NWs-baseddevices and electronics on PCBs.

2. Experimental

2.1. ZnO NWs synthesis

We chose conventional, commercially available flexible copper-PCBs (Mega Electronics, Cambridge, UK) as low-cost substrates for allthe experiments. These PCBs consist of a 50 µm thick polyester film,single side epoxy-bonded with an electro-deposited high-ductility(EDHD) 35 µm copper film. We cut each PCB into square samples. Weperformed experiments on control samples and on PCB samples pre-treated with potassium permanganate. In the latter case, we placed thesamples in glass beakers filled with potassium permanganate solution(Sigma Aldrich, KMnO4, low in mercury, ACS reagent, ≥ 99.0%), withKMnO4 concentrations of 0.5 mM, 5 mM and 50 mM. The beakers weresealed and placed onto a hot plate at 90 °C for 20 min, then we ex-tensively rinsed the PCB samples in Millipore water and sonicated themfor 1 min in Millipore water using an Elma SIOH Elmasonic bathsystem. We prepared the nutrient solutions for ZnO NWs growth withultra-pure DI water (Barnstead Easypure II filtration system, resistivityat 25 °C> 18.2 MΩ·cm). We performed all the experiments with a nu-trient solution composed by 7.5 mM zinc nitrate hexahydrate (SigmaAldrich, purum p.a., crystallized, ≥ 99.0% (KT)), 3.75 mM hexam-ethylenetetramine (Sigma Aldrich, HMTA, ACS reagent, ≥ 99.0%),0.10 M NH4OH (Fisher ammonium hydroxide 35% v/v in water), 2 mMpolyethylenimine (Sigma Aldrich, PEI, ethylenediamine branched,average Mw ~800, average Mn ~600). In addition to this nutrientsolution, we differentiate the experiments by varying the content ofpotassium chloride (Fluka, KCl, purum p.a.> 99.0%) with 0 – 0.5 – 5 –50 mM and by adding different amine molecules: monoethanolamine(Sigma Aldrich, MEA, reagent Plus ≥ 99.0%), diethanolamine (SigmaAldrich, DEA, reagent Plus ≥ 99%), triethanolamine (Sigma Aldrich,TEA, ≥ 99%) at 7.5 mM concentration. Monoethanolamine was used atdifferent concentrations (from a minimum of 0.75 mM to a maximum of75 mM) in order to investigate its effects on the NWs growth. Wesoaked the substrates in 250 mL nutrient solutions. The reaction oc-curred for 12 h at 85 °C in Memmert oven. Speciation plots to evaluatezinc chemical species present in solution were obtained from HySS,4.0.31, Hyperquad Simulation and Speciation software. The software

permits to retrieve the concentrations of the zinc species present in thezinc nutrient solution as a function of the pH and of the concentrationsof zinc ligands. Calculations are executed at standard conditions (25 °C)and at 85 °C by estimating conditional equilibrium constants.

2.2. Statistical analyses

The ZnO NW densities are measured from SEM pictures using theImageJ software. ZnO NWs are counted and their lateral sizes aremeasured from squared areas of 2 × 2 µm2 selected from comparableregions of top view SEM images. ZnO NWs densities are expressed asaverage counts per µm2 – this value derives from the average of ZnONWs counts from 6 areas each of 2 × 2 µm2 (see SupportingInformation 1). The NWs lateral sizes (measured at the center of the NWand at the top of the NW) were measured as average values from 15NWs selected from a square area of 2 × 2 µm2. The lengths are aver-aged from 15 NWs selected from side view images. From such values,we could calculate aspect ratios. We compared NWs densities, lengthsand diameters by Analysis of Variance (ANOVA), Tukey's post tests.

2.3. ZnO NWs characterisation

We analysed topographical and morphological features of the grownZnO NWs by acquiring scanning electron microscopy (SEM) images ofthe samples with an FE-SEM (LEO SUPRA 1250, Oberkochen,Germany). We performed Energy Dispersive X-ray spectroscopy (EDX)with a Quanta INCA system. In order to prevent sample surface char-ging, we connected a corner of the copper layer of the PCB to thegrounded SEM chamber, without covering the nanowires with a thingold layer. The XPS analysis was carried out by the Kratos AXIS UltraDLD equipped with an Al Kα source (hν = 1.4866 keV). Each specimenwas analysed at an emission angle of 0° as measured from the surfacenormal. The elements presented on the samples were identified fromtheir survey spectra. The survey spectra were acquired at a pass energyof 80 eV and a step size of 1.0 eV. To obtain more detailed information,high-resolution spectra were also recorded from individual peaks at20 eV pass energy with a step size of 0.05 eV. The components for theelement peaks due to the different chemical species have a Gaussian/Lorentzian shape and are quantified using nonlinear least-squares re-gression. The spectra were analysed by the XPS Peak software (4.1version). Photoluminescence spectra were acquired by the ZLX-PL-Iinstrument (excited by HeCd laser at 340 nm). XRD data were acquiredby Philips XRD spectrometer (X′per pro, Cu Kα). During packaging, thetwo PCBs layers were held in compression (applied pressure: 3.9 kPa)with an adhesive interlayer for 48 h at room temperature. The squarewave generated by the bracelet was acquired by the analog input of themultifunction data-acquisition U2331A Modular Multifunction DataAcquisition instrument (Keysight technologies). The ZnO heaters havebeen characterized by a FLIR i7 thermal camera (140 × 140 pixel),with emissivity set to 0.95.

3. Results and discussion

3.1. Temperature, pH, container volume, growth time and definition of thereference nutrient solution

A key obstacle to the rational design of wet-chemistry methods forsynthesizing NWs is the very high number of critical parameters, in-cluding temperature, pH, container volume, growth time, and con-centrations of all the reagents. Therefore, we first identified the para-meters whose optimal values could be, approximately, predicted. Inorder to achieve a good length (e.g. longer than 10 µm, as required forminimizing the risks of short-circuits during packaging [2]), a 12 hgrowth and a 80 mL solution volume can be sufficient. Standard PCBsmay not withstand, especially for prolonged times, high temperatures(e.g. even 95 °C for 12 h can already result in deformations of the plastic

G. Arrabito et al. Nano Energy 46 (2018) 54–62

55

support of the PCBs); therefore, we set the synthesis temperature at85 °C which ensures the hexamethylenetetramine (HMTA) decomposi-tion (which occurs at temperatures higher than 70 °C) and increase theNWs density [15] without damages to PCBs. We considered, in all theexperiments, a 7.5 mM concentration of zinc nitrate hexahydrate and3.75 mM HMTA, 0.10 M ammonium hydroxide (NH4OH) and 2 mMpoly-ethylenimine (PEI); in fact, these amounts of zinc can result in asupersaturation degree which is likely ideal to favor the growth of thinNWs [15], while PEI is able to optimize the zinc release [16]. Anothercrucial parameter is pH. Excessive pH values (> 11) decrease super-saturation and could significantly etch the PCB copper. Additionally,the pH affects the electrical charge densities on both the lateral and thevertical surfaces of the NWs. These electrical charges can attract ionswhich may competitively limit the access of reactive zinc species, thusreducing the growth rate in the orthogonal direction [17]. Therefore, inorder to preserve the maximum degree of freedom in the selection ofadditional reagents, we set the pH around the isoelectric point of thetop faces (tips) of the NWs, thus effectively minimizing the charges atthe NWs tips and preventing electrostatic reductions of the axial growthrate. Since the isoelectric point of the ZnO NWs tips is around 8.8 [18],we kept the pH during NWs growth around 8.9–9.0 by taking advantageof ammonia buffering; this pH value also ensures sufficient super-saturation and is not aggressive for the PCB (Table S1).

3.2. MnOOH-functionalization of the copper surface before the solution-growth of NWs

In order to facilitate packaging, high NWs densities are required.However, unless a seed layer is pre-deposited, the direct wet-chemistrygrowth of ZnO NWs onto metals typically result in low density and poorreproducibility [12,19]. On the other hand, seed layers typically requireannealing at high temperature and, therefore, may not be used onconventional PCBs. As an alternative, at the temperature of the NWssynthesis (85 °C), the absorption of metal ions – like zinc – on manga-nese oxides, when in colloidal form, is much more energetically con-venient than on bare metal surfaces [20]. For this reason, we soaked thePCB in potassium permanganate (KMnO4) solutions (90 °C for 20 min;the temperature, slightly higher than 85 °C, is tolerable because of thevery short duration of the procedure), thus forming a thin manganeseoxyhydroxide (MnOOH) layer [13]. After the MnOOH deposition, wesuccessfully grew ZnO NWs on the PCBs (12 h, 85 °C), whereas in ab-sence of MnOOH functionalization, the NWs growth was not re-producible (Supporting Information 1–2, Fig. S1-S2). Our results, inagreement with recent reports on glass substrates [21], confirm thatlow amounts of KMnO4 (0.5 mM) result in the deposition of thinMnOOH films which facilitate the heterogeneous nucleation of ZnONWs, whereas high KMnO4 concentrations (5–50 mM) result in thickerMnOOH films, higher nucleation in the solution phase, and lower NWsdensity (see Supporting Information 1–2). In all the following

Fig. 1. Modelling of EDA-assisted control on zinc speciation, supersaturation degree and solubility during the wet-chemical synthesis of ZnO NWs. (a), Relative concentrations of zinccomplex species as a function of the EDA concentration (0 – 75 mM) for a nutrient solution containing 7.5 mM Zn(NO3)2·6H2O, 3.75 mM HMTA, 100 mM NH4OH, and 2 mM PEI, with pHand temperature equal to 8.9 and 85 °C, respectively. Inset shows EDA concentrations in the range 0 – 10 mM. (b), Degree of supersaturation as a function of the EDA (0 – 75 mM) and Cl-

(0 – 50 mM) concentrations, with pH and temperature equal to 8.9 and 85 °C, respectively. (c), Solubility plot for the nutrient solution as a function of pH for different EDA concentrations(0 – 75 mM) at 85 °C. Inset shows the same calculations and the degree of supersaturation (Δc) in the pH range 8–10.

G. Arrabito et al. Nano Energy 46 (2018) 54–62

56

experiments, in order to improve reproducibility and to maximize theNWs density (up to 8.3 NWs/µm2), we employed MnOOH-functionali-zation with 0.5 mM KMnO4.

3.3. EDA-based control of zinc speciation and supersaturation in thenutrient solution

The potassium permanganate pre-treatment allows to reproduciblysynthesize high density NWs. However, for fabricating ZnO nanode-vices on PCBs, ideally, the NWs should also be long, thin and untapered.Ethylenediamine (EDA, H2NCH2CH2NH2) may help to grow thin NWsat alkaline pHs [22], but easily forms zinc complexes because of its highformation constants with Zn and, therefore, can change the zinc spe-ciation and reduce the supersaturation degree (i.e. the difference be-tween the zinc experimental concentration, 7.5 mM in our case, and thezinc solubility). In fact, high EDA concentrations can even impede thegrowth of NWs on the substrate. Moreover, EDA can also result in ta-pered NWs [22]. This problem could be solved by the addition ofchloride ions which can stabilize the top axial faces of the NWs [23]without significantly modifying the zinc speciation and the super-saturation degree because of their very low formation constants withzinc (see Supporting Information 3). As a simple approach, adding KClto the solution can introduce chloride ions but, of course, will also in-troduce K+ ions which, at higher pH, e.g. 11, after being attracted bythe negatively charged NWs tips, could limit the access of zinc ions andslow down the axial growth rate [17]; nevertheless, at our pH, aroundthe isoelectric point of the NWs tips, no significant electrostatic at-tractions between ions and the NWs tips are expected. Based on theseconsiderations, in order to obtain quantitative insights for process de-sign, we computed the zinc speciation, the zinc solubility and the su-persaturation degree of possible nutrient solutions (see SupportingInformation 3). For our calculations, we, first, considered the formationconstants for zinc complexes with zinc ligands, at 25 °C and, then, es-timated the corrected constants at 85 °C [24]. As an advantage, sincewe have already set the pH, in contrast with typical speciation plots[17], we can compute zinc speciation as a function of the EDA and KClconcentrations rather than as a function of pH. Fig. 1a shows that in oursystem EDA effectively changes the zinc speciation and, in particular,forms Zn(EDA)22+, which can help to grow thin NWs, but, at highconcentrations, would dramatically reduce the zinc hydroxide ions andthus prevent the NWs growth. Consistently, Fig. 1b shows that EDA canfinely tune both the zinc solubility and the supersaturation degree (Δc).Moreover, Fig. 1b confirms that, as expected, the supersaturation de-gree is almost independent on the KCl concentration. Finally, Fig. 1cshows the zinc solubility and the supersaturation degree as a function ofpH, for different EDA concentrations. As evident, Zn–EDA complexes,which help to grow thin NWs, are formed at these low EDA con-centrations because of the relatively mild pH (at higher pH, e.g. 11,small EDA concentrations would only negligibly affect the speciation ofzinc, Fig. S4c, and would only form tiny amounts of zinc-EDA com-plexes). During our syntheses, after a transient, the pH reaches andmaintains a value close to 8.9 (Supporting Information 4, Fig. S6).

3.4. In-situ formation of EDA from MEA by zinc ions catalysis

Though EDA can effectively control the zinc solubility and the su-persaturation degree, EDA also constitutes a serious hazard upon ex-posure to ambient conditions. As a safer precursor, monoethanolamine(MEA, H2NCH2CH2OH) can produce EDA in presence of zinc nutrientsolutions by reacting with ammonia, according to the possible reactionin aqueous solution:

H2NCH2CH2OH + NH4OH ⇆ H2NCH2CH2NH2 + 2H2O

A possible mechanism of formation of ethylenediamine (EDA) frommonoethanolamine (MEA) is as follows. Since the alcohol hydroxyl

group of MEA is a bad leaving group, this reaction would normally beunlikely, but in presence of Lewis acids such as zinc [25], the alcoholhydroxyl group becomes a good leaving group. In fact, in the nutrientgrowth solution, zinc ions form complex species with ligands (i.e. ZnL,in which L is a generic zinc ligand molecule) which, in turn, can form acomplex with the alcohol through association with one of the two un-shared pairs of electrons of the oxygen atom (1), thus increasing thepossibility that a good nucleophile, like ammonia, displaces the hy-droxyl leaving group and forms EDA (2 and 3). Finally, the hydro-xylated form of zinc complex gives back the initial ZnL specie (4).

The in-situ formation of EDA from MEA has been speculativelyconsidered to explain the formation of ZnO NWs from nanoparticles tonano- and micro-rods in ultrasonic baths at low temperature [26]. Inour nutrient solution, ammonia is in excess (100 mM) with respect toMEA (0.75–75 mM), so that the reaction is favored. By taking into ac-count previous experimental evidences on reductive amination reac-tions in aqueous media [27], we expect a good yield for the conversionof MEA into EDA in aqueous solutions containing zinc ions. As negativecontrols to monoethanolamine (MEA), we performed experiments withthe addition to the nutrient solution of the corresponding secondaryand tertiary amines, i.e. diethanolamine (DEA) and triethanolamine(TEA) instead of MEA (see Supporting Information 3). These com-pounds are not expected to produce, in our reaction conditions, thecorresponding linear amines Bis(2-aminoethyl)amine (DIEN) and Tris(2-minoethyl)amine (TREN) which have affinity constants with zincions higher than EDA (see Supporting Information 3). Instead, mixturesof amines bearing hydroxyl groups and cyclic amines are expected to beformed [28] – which exert little effect on the zinc speciation and so-lubility (see below and Fig. S8, Supporting Information, for experi-mental verification).

3.5. MEA-KCl synergic control of the ZnO NWs synthesis

We experimentally verified the occurrence of the above describedoptimal synthesis conditions. Consistently, in comparison with thecontrol case (Fig. 2a), the synthesis of ZnO NWs in presence of 7.5 mMMEA and without KCl (Fig. 2b) resulted in short (8.0± 0.3 µm), lowaspect ratio (56±9) NWs with clear sharpened “needle-like” mor-phology similar to reports in literature in which EDA is added to thenutrient solution [29]. Therefore, we investigate the ability of chlorideions to stabilize the high surface energy (002) top face [23] andcounteract the EDA–induced NWs tapering. Within the concentrationrange (0.5–50 mM), KCl significantly increased both the NWs lengthand lateral size (Fig. 2c,e). Low concentrations of KCl (0.5–5 mM) sta-bilized the (002) face (untapered NWs), whereas high KCl concentra-tions (50 mM), besides reducing the NWs density, determined needle-shape NWs (Supporting Information 5, Fig. S7) due to the gradual de-creasing surface area of each hexagonal crystallographic atomic layerdirecting towards the c-axis, a mechanism described as crevice andhillock formation in ZnO crystal growth [30]. We therefore chose

G. Arrabito et al. Nano Energy 46 (2018) 54–62

57

0.5 mM KCl as an optimal concentration. Finally, we combined KCl(0.5 mM) and MEA (0.75–75 mM) in the nutrient solution for simulta-neously producing EDA in situ and stabilizing the NWs tips (Fig. 2d). Asschematically shown in Fig. 2f, in the best case (1:1 M ratio Zn: MEA,i.e. 7.5 mM MEA concentration), we obtained 15.3± 0.3 µm length andaspect ratios as high as 208± 48 (with a nanowire density of 8.6± 0.4

NWs/µm2, which is mainly determined by the KMnO4 pre-treatment).Most remarkably, these results were highly reproducible; for instance,in two replicated experiments we found NWs length, aspect ratio, anddensity equal to 14.3±0.2 µm, 180±78, 8.2± 0.4 NWs/µm2 (firstexperiment) and 14.4± 0.2 µm, 220±54, 8.1± 0.4 NWs/µm2

(second experiment). The addition of similar amines (DEA and TEA)

Fig. 2. Synergic MEA-KCl control of the ZnO NWs morphology on MnOOH-functionalized Cu PCB. (a-d), ZnO NWs grown on flexible PCB after MnOOH-Cu functionalization by using thereference nutrient solution (a) or by adding 7.5 mM MEA (b), 0.5 mM KCl (c) or the combination of 7.5 mM MEA and 0.5 mM KCl (d). Scale bars of top view images and insets: 300 nmand 80 nm. Scale bars of 90° tilted images: 2 µm. e–f, Systematic investigation of the NWs aspect ratios (striped light blue bars) and lengths (orange bars) as a function of the KCl (e) andMEA (solution containing 0.5 mM KCl) concentration (f). NWs lengths are compared by ANOVA, Tukey's post test (**: p<0.01).

G. Arrabito et al. Nano Energy 46 (2018) 54–62

58

which cannot produce EDA did not significantly modify the NWsmorphology, thus further confirming the in-situ production of EDA (seeSupporting Information 5, Fig. S3–S5, S8). Low MEA concentrations(0.75 mM and 3.75 mM) improved both the length and aspect ratio,whereas higher concentrations of MEA (15 mM and 75 mM) led to ta-pered NWs with reduced length and aspect ratio (Fig. 2f). All our resultsconsistently confirm the MEA-induced in situ formation of Zn(EDA)22+

and Zn(EDA)(OH)+. In fact, first, MEA, if not converted in EDA, wouldnot appreciably modify the zinc speciation given the very low forma-tion constants with zinc – see Supporting Fig. S4b; second, the reduceddiameter of the NWs unambiguously proves the formation of zinc-EDAcomplexes; third, higher (i.e. ≥ 75 mM) MEA concentrations (i.e higherEDA concentrations), besides reducing both the zinc hydroxide con-centration and the supersaturation degree, form high concentrations ofZn-(EDA)22+ complexes and therefore make the KCl-induced top-facestabilization less effective. In striking contrast with MEA alone (thin buttapered NWs, Fig. 2b), the proper combination of MEA and KCl re-producibly gives almost uniform prism-like NWs with sub-100 nmwidth (Fig. 2d) and aspect ratios as high as 200. We investigated thephase and elemental composition of the ZnO NWs by XPS, XRD andPhotoluminescence (PL) (see Supporting Information 6–9, Fig. S9–11).XRD showed a significantly higher (002) diffraction peak intensity incomparison to (100) and (101), indicating that the general preferentialorientation of the ZnO NWs is along the direction perpendicular to the(001) crystallographic face (c-axis direction). We experimentally ver-ified by XPS that the ZnO NWs do not show significant nitrogen orchloride contamination (see Supporting Information 9), also con-sistently with the negligible electrostatic interactions, at our pH

(around the isoelectric point of the NWs tips) between ions and the NWstips. We detected minute amounts of Mn (6%) deriving from theMnOOH modified PCB copper layer. Remarkably, the synergy of KCland MEA is beneficial for the ZnO NWs morphology as demonstrated bythe significantly higher (002) orientation, the higher crystalline com-ponent (O1) in the O1s XPS peak and the higher near-band-edge (NBE)emission and deep-level emissions (DLE) in comparison to the controlcase (see Supporting Information 6–9, Fig. S9–11).

3.6. ZnO piezotronic heartbeat detector and electronics on a flexible PCB

After growing on PCBs high-density, long, thin, untapered ZnONWs, we simultaneously add the top electrode, package the NWs, im-prove their mechanical robustness, and connect to the electronic in-terface by flip-chip adhesive bonding. As schematically illustrated inFig. 3a, after depositing an adhesive polymer, a second PCB is flippedover, aligned and placed on the first PCB under constant pressure. Boththe PCBs must be of the same type because, as typical with flip-chipbonding, mismatches between the temperature coefficients of expan-sion of the two substrates could degrade reliability. The record length(for substrates which may not withstand high temperatures, such asPCBs, see Table S3 in Supporting Information) and untapered nature ofthe synthesized NWs allow reliable packaging and high-quality elec-trical contacts between the tips of the ZnO NWs and the copper of thePCB top electrode. After packaging, circuits can be soldered as usual. Asa proof of concept, Fig. 3b-c show a bracelet comprising ZnO NWs asultra-sensitive piezotronic pressure transducers, electronics (conven-tional surface-mount devices, SMD) and a cotton wristband for wearing

Fig. 3. ZnO nanodevice on PCB for heartbeat detec-tion. (a), Schematic illustration of the flip-chip ad-hesive bonding. (b), Photograph of the device wornaround a wrist. (c), Photograph of the PCB device,showing the cotton stripe, the sensing part (dark)and the electronic circuits. (d), Frequency of the os-cillator output voltage (vOUT) during measurementstaken on a 35-years old man in good health and atrest (raw data, blue; 9-samples moving average, red).(e), Schematic illustration of the electronic interface(ZnO NWs are represented by the rectangular com-ponent).

G. Arrabito et al. Nano Energy 46 (2018) 54–62

59

the bracelet. Fig. 3d shows the frequency of the oscillator output vol-tage (vOUT) during measurements taken on a 35-years old man in goodhealth and at rest (raw data and 9-samples moving average for low-passfiltering high-frequency disturbances). As shown in Fig. 3e, the elec-tronic interface is an oscillator circuit which transduces the currentthrough the ZnO NWs (represented by the rectangular box), biased at aconstant voltage VREF, into an output frequency (see SupportingInformation 10), so that the heartbeat pulses can modulate the oscil-lator frequency (Fig. 3d). Table S2 reports the values of the componentsused in the electronic interface.

3.7. ZnO nanoheaters on flexible PCBs

We also used the same PCB ZnO nanodevices, with active areasequal to 3 × 3 mm2, as flexible heaters. In contrast with conventionalserpentine-shaped resistive heaters, the current flow is vertical to thesubstrate (rather than parallel), thus resulting in a much larger cross-section, and the voltage across the heater includes the voltage dropsacross two metal-ZnO junctions (rather than being a purely ohmicvoltage). Fig. 4a shows a PCB heater and its temperature distributionwhen the heating power is around 0.25 W (heating power per unit areaaround 2.8 W/cm2); in these conditions the thermal resistance betweenthe heater and the environment, RTH, may be estimated around 320 K/W. In order to clearly show the temperature differences within theheater, the color scale bars are saturated (i.e. no color changes above anupper threshold temperature or below a lower threshold temperature);the maximum temperature, shown at the top left corner, corresponds tothe white cross mark at the center of the heater. As shown in Fig. 4b, wealso mounted the PCB heater on a conventional heat sink by usingthermal paste, thus reducing RTH to about 70 K/W; in these conditions,Fig. 4c shows the values of both the maximum temperature and thepower per unit area as a function of the input voltage applied to theheater. Finally, as shown in Fig. 4d, we immersed the heater attached tothe heat sink in water, thus further reducing the thermal resistance andallowing to reach power per unit area up to about 20 W/cm2, which iscomparable or better than state of the art flexible heaters which,however, require special processing and may not be integrated onstandard PCBs but must be fabricated on different substrates which areseparately connected to the electronics. As with conventional flexibleheaters, the typical variability of the current transport through theheater (e.g. spread, bending) may be counteracted by including a tem-perature sensor and a proper feedback control loop. The high values of

the currents (e.g. up to 0.4 A/cm2) through the NWs confirm that therational control in ZnO NWs synthesis results in excellent electricalcontacts between the PCB copper layer and the top faces of the un-tapered NWs. Similar NW-based heaters could be useful for in-vestigating biological and chemical processes at micro- and nano- scale,e.g. protein synthesis [31] or cell lysis [32]. We have also packagedseveral control devices comprising non-optimal NWs. These experi-ments confirmed that our optimized, long and thin NWs are better forboth packaging (length is crucial to minimize the risk of short circuits)and sensitivity to pressure (consistent with piezoelectric size effects inzinc oxide and better crystallinity, see Supporting Information 6–9, Fig.S9–11). Moreover, we found it extremely difficult to create high-qualityand reliable electrical contacts when using samples with tapered NWs,likely because of the reduced contact area (i.e. high parasitic seriesresistance) between the sharp NWs tips and the top electrode; instriking contrast, the excellent electric contact between the top elec-trode and the untapered tips of our optimized NWs easily allowed toobtain very large current densities with reasonably small applied vol-tages (Fig. 4c).

4. Conclusions

Here we have developed a rational approach to control the ZnONWs synthesis, thus enabling the low temperature (< 90 °C) solutiongrowth on PCBs of NWs with high densities (> 8/µm2) and recordlength (> 15 µm) and aspect-ratio (> 200). In fact, though even longerZnO NWs have been previously grown in solution (see Table S3 inSupporting Information), all these reports involved the deposition andsubsequent high temperature annealing (i.e. 350–400 °C) of ZnO seed-layers, which is only possible on rigid substrates which can safelywithstand high temperatures (e.g. ITO coated glass or silicon). Inpractice, we preliminarily fixed the synthesis parameters whose optimalvalues could be determined, such as growth time, solution volume,temperature and concentrations of HMTA, PEI, ammonium hydroxideand zinc nitrate hexahydrate. Afterwards, we set the pH, after the initialtransient, at the isoelectric point of the ZnO NWs tips (8.9), in order topreserve the integrity of the copper surfaces and to prevent significantinteractions between the NWs tips and ions which could otherwisecompetitively limit the access of reactive zinc species. We numericallycomputed zinc speciation in solutions containing amine-ligands, inparticular the bidentate ligand ethylenediamine (EDA), for under-standing the effects induced on the formation of complexes with zinc

Fig. 4. Flexible ZnO nanoheaters on PCB. a–b,Optical (top) and thermal (bottom) image of a flex-ible PCB heater (a) and of the same device attachedto a conventional heat-sink by thermal paste (b). c,Temperature and power per unit area as a function ofthe input voltage for the flexible PCB heater attachedto the heat sink (b). d, Optical image of a flexiblePCB heater mounted on a heat sink and immersed inwater for further reducing the heater-to-environmentthermal resistance.

G. Arrabito et al. Nano Energy 46 (2018) 54–62

60

and the effect on supersaturation at the pH 8.9 and at 85 °C, in order tofavor the growth of ZnO NWs on copper and to reduce homogeneousgrowth. These calculations show that at our pH, 8.9, and with 7.5 mMzinc ions concentration, EDA is surprisingly able – already at very lowconcentrations (0.75–7.5 mM) – to trigger a significant reduction in theconcentration of the zinc hydroxide species needed for growing NWs onthe substrate because of the formation of Zn(EDA2)2+ species, whereasat higher pH values, its effect becomes less relevant – i.e. higher per-centage of zinc hydroxide species are formed which would in additioneasily give rise to homogeneous growth. Importantly, in this range ofEDA concentrations (0.75–7.5 mM), zinc supersaturation is not af-fected, thus permitting to exert a sufficient driving force for the ZnONWs formation on the PCB whilst reducing homogeneous growththanks to the formation of Zn(EDA2)2+. However, since EDA is highlytoxic and dangerous, we exploited zinc ions as Lewis acid catalysts tofavor amination of MEA (a safer precursor) by ammonia in order to in-situ produce EDA. This choice is a convenient alternative to conven-tional ligand-based control on materials growth, permitting an easierand much safer integration of the ZnO NWs onto PCBs, following agreen chemistry approach. Then, we combined EDA, produced in situfrom monoethanolamine (MEA), with potassium chloride in order tostabilize the top faces of the NWs, thus achieving the reproduciblegrowth of high-density, long, thin and untapered NWs which arecharacterized by the highest reported density, length and aspect ratiofor syntheses at temperatures compatible with flexible substrates(Supporting Information 11, Table S3). These characteristics make theresulting NWs easy to be packaged and connected to electronics by flip-chip adhesive bonding, thus enabling the co-integration of ZnO nano-devices and electronics on PCBs. As a first demonstration, we haveshown a flexible PCB comprising both a ZnO piezotronic heartbeatdetector and its electronic interface. Moreover, we have reported flex-ible ZnO nanoheaters on flexible PCBs with extremely high powerdensities, which are comparable with state of the art flexible heaters. Infact, the untapered ends of the NWs allow to obtain high-quality elec-trical contacts with the top PCB added during packaging, so that veryhigh current densities (e.g. up to 0.4 A/cm2) can flow through the NWs.The wearable bracelet for heart-beat detection and the flexible heatersdemonstrated here may represent just two examples among manypossible ZnO nanodevices (e.g. piezoelectric/pyroelectric energy har-vesting, pressure/strain sensing, heaters, temperature sensors, recti-fiers,…) that can in principle be co-integrated with electronics on PCBs.Our synergic method for the wet-chemical synthesis of ZnO NWs can bea milestone towards the widespread inclusion of ZnO nanodevices incomplete electronic systems.

Acknowledgments

G. Arrabito and V. Errico contributed equally to this work. Thisresearch was funded by IIT (Project Seed – API NANE) and MIUR (FIRB– Futuro in Ricerca 2010 Project RBFR10VB42) and also partiallysupported by NSFC (Grant No. 61404066). The authors acknowledgeM. Palmacci for help in fabrication and characterization, G. Carrisi, G.Mastrosimone, S. Gulino, A. Pettignano and A. Orsini for useful dis-cussions. The authors also acknowledge B. Pignataro and M. Scopellitifor their precious help and collaboration.

Appendix A. Supporting information

Supplementary data associated with this article can be found in theonline version at http://dx.doi.org/10.1016/j.nanoen.2018.01.029.

References

[1] X. Wang, J. Song, J. Liu, Z.L. Wang, Direct-current nanogenerator driven by ul-trasonic waves, Science 316 (2007) 102–105.

[2] S. Xu, Y. Qin, C. Xu, Y. Wei, R. Yang, Z.L. Wang, Self-powered nanowire devices,

Nat. Nanotechnol. 5 (2010) 366–373.[3] R.S. Friedman, M.C. McAlpine, D.S. Ricketts, D. Ham, C.M. Lieber, Nanotechnology

high-speed integrated nanowire circuits, Nature 434 (2005) 1085.[4] Z. Fan, J.C. Ho, Z.A. Jacobson, H. Razavi, A. Javey, Large-scale, heterogeneous

integration of nanowire arrays for image sensor circuitry, Proc. Natl. Acad. Sci. USA105 (2008) 11066–11070.

[5] H. Zhu, Y. Fu, F. Meng, X. Wu, Z. Gong, Q. Ding, M.V. Gustafsson, M.T. Trinh, S. Jin,X.-Y. Zhu, Lead halide perovskite nanowire lasers with low lasing thresholds andhigh quality factors, Nat. Mater. 14 (2015) 636–642.

[6] W. Wu, X. Wen, Z.L. Wang, Taxel-addressable matrix of vertical-nanowire piezo-tronic transistors for active and adaptive tactile imaging, Science 340 (2013)952–957.

[7] K. Takei, T. Takahashi, J.C. Ho, H. Ko, A.G. Gillies, P.W. Leu, R.S. Fearing, A. Javey,Nanowire active-matrix circuitry for low-voltage macroscale artificial skin, Nat.Mater. 9 (2010) 821–826.

[8] Y. Xia, P. Yang, Y. Sun, Y. Wu, B. Mayers, B. Gates, Y. Yin, F. Kim, H. Yan, One-dimensional nanostructures: synthesis, characterization, and applications, Adv.Mater. 15 (2003) 353–389.

[9] S. Xu, Z.L. Wang, One-dimensional ZnO nanostructures: solution growth andfunctional properties, Nano Res. 4 (2011) 1013–1098.

[10] W. Gao, S. Emaminejad, H.Y.Y. Nyein, S. Challa, K. Chen, A. Peck, H.M. Fahad,H. Ota, H. Shiraki, D. Kiriya, D.-H. Lien, G.A. Brooks, R.W. Davis, A. Javey, Fullyintegrated wearable sensor arrays for multiplexed in situ perspiration analysis,Nature 529 (2016) 509–514.

[11] S. Xu, Z. Wang, One-dimensional ZnO nanostructures: solution growth and func-tional properties, Nano Res. 4 (2011) 1013–1098.

[12] S. Xu, N. Adiga, S. Ba, T. Dasgupta, C.F.J. Wu, Z.L. Wang, Optimizing and improvingthe growth quality of ZnO nanowire arrays guided by statistical design of experi-ments, ACS Nano 3 (2009) 1803–1812.

[13] M. Kokotov, G. Hodes, Reliable chemical bath deposition of ZnO films with con-trollable morphology from ethanolamine-based solutions using KMnO4 substrateactivation, J. Mater. Chem. 19 (2009) 3847–3854.

[14] X. Wen, W. Wu, Y. Ding, Z.L. Wang, Seedless synthesis of patterned ZnO nanowirearrays on metal thin films (Au, Ag, Cu, Sn) and their application for flexible elec-tromechanical sensing, J. Mater. Chem. 22 (2012) 9469–9476.

[15] C. Xu, D. Gao, Two-stage hydrothermal growth of long ZnO nanowires for efficientTiO2 nanotube-based dye-sensitized solar cells, J. Phys. Chem. C 116 (2012)7236–7241.

[16] K. Liu, W. Wu, B. Chen, X. Chen, N. Zhang, Continuous growth and improved PLproperty of ZnO nanoarrays with assistance of polyethylenimine, Nanoscale 5(2013) 5986–5993.

[17] J. Joo, B.Y. Chow, M. Prakash, E.S. Boyden, J.M. Jacobson, Face-selective elec-trostatic control of hydrothermal zinc oxide nanowire synthesis, Nat. Mater. 10(2011) 1–6.

[18] B.Y. Chow, J. Joo, M. Prakash, Methods and Apparatus for Control of HydrothermalNanowire Synthesis, US8367435 B2, 2011.

[19] B.H. Kim, J.W. Kwon, Metal catalyst for low-temperature growth of controlled zincoxide nanowires on arbitrary substrates, Sci. Rep. 4 (2014) 4379, http://dx.doi.org/10.1038/srep04379.

[20] R.R. Gadde, H.A. Laitinen, Heavy metal adsorption by hydrous iron and manganeseoxides, Anal. Chem. 46 (1974) 2022–2026.

[21] T. Dixit, A. Bilgaiyan, I.A. Palani, D. Nakamura, T. Okada, V. Singh, Influence ofpotassium permanganate on the anisotropic growth and enhanced UV emission ofZnO nanostructures using hydrothermal process for optoelectronic applications, J.Sol.-Gel Sci. Technol. 75 (2015) 693–702.

[22] B. Liu, H.C. Zeng, Hydrothermal synthesis of ZnO nanorods in the diameter regimeof 50 nm, J. Am. Chem. Soc. 125 (2003) 4430–4431.

[23] J.M. Downing, M.P. Ryan, M.A. Mclachlan, Hydrothermal growth of ZnO nanorods:the role of KCl in controlling rod morphology, Thin Solid Films 539 (2013) 18–22.

[24] M. Kersten, D.A. Kulik, Modeling of elemental species, Handb. Elem. Speciat. II –Species Environ. Food, Med. Occup. Heal. John Wiley & Sons, Ltd, 2005, pp.651–689.

[25] P.T. Buonora, K.G. Rosauer, L. Dai, Control of the aqueous Aldol addition underClaisen-Schmidt conditions, Tetrahedron Lett. 36 (1995) 4009–4012.

[26] P. Rai, M. Song, J. Kim, Y. Kim, H. Song, Y. Yu, Effect of ammonia water on themorphology of monoethanolamine-assisted sonochemicaly synthesized ZnO na-nostructures, Nanosci. Nanotechnol. 12 (2012) 1380–1385.

[27] G.B. Giovenzana, D. Imperio, A. Penoni, G. Palmisano, Reductive amination withzinc powder in aqueous media, Beilstein J. Org. Chem. 7 (2011) 1095–1099.

[28] T. Schaub, B. Buschhaus, M.K. Brinks, M. Schelwies, R. Paciello, J.P. Melder, M.Merger, Process for the preparation of primary amines by homogeneously catalyzedalcohol amination, US20120232309A1, 2012.

[29] B. Liu, H.C. Zeng, Room temperature solution synthesis of monodispersed single-crystalline ZnO nanorods and derived hierarchical nanostructures, Langmuir 20(2004) 4196–4204.

[30] R.A. Laudise, E.D. Kolb, A.J. Caporaso, Hydrothermal growth of large sound crystalsof zinc oxide, J. Am. Ceram. Soc. 47 (1964) 9–12.

[31] P. Ginet, K. Montagne, S. Akiyama, A. Rajabpour, A. Taniguchi, T. Fujii, Y. Sakai,B. Kim, D. Fourmy, S. Volz, Towards single cell heat shock response by accuratecontrol on thermal confinement with an on-chip microwire electrode, Lab Chip 11(2011) 1513–1520.

[32] N. Privorotskaya, Y.-S. Liu, J. Lee, H. Zeng, J.A. Carlisle, A. Radadia, L. Millet,R. Bashir, W.P. King, Rapid thermal lysis of cells using silicon-diamond micro-cantilever heaters, Lab Chip 10 (2010) 1135–1141.

G. Arrabito et al. Nano Energy 46 (2018) 54–62

61

Giuseppe Arrabito received his B.Sc. degree (2006) inChemistry and M.Sc. degree (2008) in BiomolecularChemistry from University of Catania, Italy. He received hisPh.D. (2012) in Nanoscience from Scuola Superiore diCatania. He spent Post-Doctoral fellowships in the group inProf. C.M. Niemeyer in TU-Dortmund (2013) and in thegroup of Dr. C. Falconi in University of Rome Tor Vergata(2014). He is currently a Post-Doctoral Scientist in thegroup of prof. B. Pignataro, University of Palermo. His re-search interests are in the field of nano-biological arraysfabrication, Dip Pen Nanolithography, DNA nano-technology, ZnO nanostructures for biosensing.

Vito Errico received his M.Sc. (2005) in micro-electronicfrom Polytechnic of Bari, Bari, Italy, and his Ph.D. degree(2008) in materials and innovative technologies from CNR/INFM ISUFI - Università degli Studi del Salento, Lecce, Italy.In 2008 he was appointed to post-doctoral researcher po-sition at the University of Birmingham, United Kingdom.From 2013, he holds post-doctoral research grant positionat University of Rome Tor-Vergata, Italy and his researchinterests involve several different areas among all biology,nano-structures, sensors and medical devices.

Zemin Zhang received his B.S. degree in Microelectronicsfrom School of Physical Science and Technology at LanzhouUniversity in 2014. Currently, he is a is a joint-trainingPh.D. student of Lanzhou University and Joint Center forArtificial Photosynthesis (JCAP) in Lawrence BerkeleyNational Laboratory (LBNL). His current major isCondensed Matter Physics and his research interests focuson energy conversion and storage devices, including dye-sensitized solar cells, self-powered photodetectors, andphotoelectrochemical water splitting and CO2 reduction.

Weihua Han is an Associate Professor of School of PhysicalScience & Technology at Lanzhou University. He receivedhis Ph.D. in Condensed Matter Physics from LanzhouUniversity in 2008. Then he joined the faculty of School ofPhysical Science & Technology at Lanzhou University. Hehas visited Professor Zhonglin Wang's group at GeorgiaTech in 2011–2012, Professor Christian Falconi's group atUniversity of Rome “Tor Vergata” in 2013 and Oak RidgeNational Lab (CNMS) in 2014. His research mainly focuseson Nanomaterials for Energy Harvesting and Storing. Healso has an interest in Force Sensors based on Piezoresistiveeffect.

Christian Falconi (CF) received his M.Sc. and Ph.D. de-grees in Electronic Engineering from the University ofRome Tor Vergata in, respectively, 1998 and 2001. Since2002 CF is Assistant Professor at the University of Rome TorVergata. Since 2013 CF is Adjunct Professor at theSungkyunkwan University (SKKU). Since 2017 CF isAdjunct Professor at the Beijing Institute of Nanoenergyand Nanosystems – Chinese Academy of Sciences. His re-search interests include electronics, sensors, nanoscience,nanogenerators, microsystems and nanosystems.

G. Arrabito et al. Nano Energy 46 (2018) 54–62

62