Retained Carrier-Mobility and Enhanced Plasmonic-Photovoltaics of

of pz orbitals in carbon atoms that form the π-cloud filled with electron and vacant π*-cloud. ..... A SiO2 (285 nm)/Si chip with CVD graphene t...
0 downloads 0 Views 5MB Size
Letter pubs.acs.org/NanoLett

Retained Carrier-Mobility and Enhanced Plasmonic-Photovoltaics of Graphene via ring-centered η6 Functionalization and Nanointerfacing Songwei Che,† Kabeer Jasuja,‡ Sanjay K. Behura,† Phong Nguyen,† T. S. Sreeprasad,§ and Vikas Berry*,† †

Department of Chemical Engineering, University of Illinois at Chicago, 810 S. Clinton Street, Chicago, Illinois 60607, United States Department of Chemical Engineering, Indian Institute of Technology, Gandhinagar, Palaj, Gujarat 382355, India § Center for Materials & Sensor Characterization, College of Engineering, and the Polymer Institute, The University of Toledo, Toledo, Ohio 43606, United States ‡

S Supporting Information *

ABSTRACT: Binding graphene with auxiliary nanoparticles for plasmonics, photovoltaics, and/or optoelectronics, while retaining the trigonal-planar bonding of sp2 hybridized carbons to maintain its carrier-mobility, has remained a challenge. The conventional nanoparticle-incorporation route for graphene is to create nucleation/attachment sites via “carbon-centered” covalent functionalization, which changes the local hybridization of carbon atoms from trigonal-planar sp2 to tetrahedral sp3. This disrupts the lattice planarity of graphene, thus dramatically deteriorating its mobility and innate superior properties. Here, we show large-area, vapor-phase, “ring-centered” hexahapto (η6) functionalization of graphene to create nucleation-sites for silver nanoparticles (AgNPs) without disrupting its sp2 character. This is achieved by the grafting of chromium tricarbonyl [Cr(CO)3] with all six carbon atoms (sigma-bonding) in the benzenoid ring on graphene to form an (η6-graphene)Cr(CO)3 complex. This nondestructive functionalization preserves the lattice continuum with a retention in charge carrier mobility (9% increase at 10 K); with AgNPs attached on graphene/n-Si solar cells, we report an ∼11-fold plasmonic-enhancement in the power conversion efficiency (1.24%). KEYWORDS: Graphene, functionalization, nanoparticles, plasmonics, photovoltaics

G

phononic transport26−28 in graphene. The alternate functionalization routes involving noncovalent π−π interactions between aromatic chemical moieties and graphene and van der Waals interactions are weak (low bond energy) and nonrobust to hold nanoparticles on graphene.29 If NPs can be anchored on functionalized graphene without altering the native planarity and superior properties, it will significantly widen the scope of its applications. In this work, we have developed a route for silver NPs (AgNPs) nucleation and growth on large-area hexahapto (η6)-functionalized graphene, which preserves graphene’s extraordinary electronic properties, including conductivity and carrier mobility. The field emission scanning electron microscopy (FESEM) shows that a high density of AgNPs are incorporated on η6functionalized graphene and that NPs enhance the Raman intensity by 2.3 ∼ fold. We also show that the photovoltaic efficiency of graphene/n-silicon (G/n-Si) heterojunction solar cells exhibit an 11-fold plasmonic-enhancement upon employing hexahapto-functionalized graphene integrated with AgNPs. In η6-functionalization chemistry, the coordination bondings of

raphene is an atomic-thick crystal of sp2 hybridized carbon atoms bonded in a trigonal-planar geometry1 with the conduction band touching the valence band (semimetal) at the high-symmetry Dirac points (K and K′) in the Brillouin zone. This results in the evolution of Dirac Fermions,2 which exhibit ultrahigh charge-carrier mobility3 and fractional quantum Hall effect.4 Graphene interfaced with nanoparticles (NPs) can leverage their plasmonics,5 magnetism,6 biocompatibility,7 photoactivity,8 optoelectronics,9 photovoltaics,10 catalysis,11 and/or specific-interactions.12,13 Other applications include enhanced sensitivity in chemical and biomolecule sensing,14,15 NP spacer to reduce aggregation,16 electrocatalysis (or catalysis) (platinum,17 Fe3O4,18 MoS219) and energy storage.20 However, all these interfaces require chemical functionalization (or defects) of graphene to create active sites for NP attachment and results in loss of the superior properties native to graphene. This is because the conventional covalent chemistries (oxidation,21,22 fluorination,23,24 sulphonation,16 and chlorination25) change the local hybridization of carbon atoms from trigonal-planar sp2 to tetrahedral sp3 thus distorting the lattice planarity of graphene and the associated continuum of delocalized π-clouds. This in turn leads to a drastically increased carrier scattering (decreased carrier mobility), reduced carrier density, and diminished electronic/ © 2017 American Chemical Society

Received: April 7, 2017 Revised: June 2, 2017 Published: June 6, 2017 4381

DOI: 10.1021/acs.nanolett.7b01458 Nano Lett. 2017, 17, 4381−4389

Letter

Nano Letters

Figure 1. Vapor-phase hexahapto functionalization of graphene. (A) Experimental design setup for hexahapto functionalization via vapor-phase process in a quartz tube vacuum furnace. (B) Schematic of hexahapto-functionalized graphene [(η6-graphene)Cr(CO)3]. (C) Molecular orbital diagram of (η6-graphene)Cr(CO)3 in a staggered configuration (bold line, electrons donated from graphene valence band to chromium unoccupied orbitals; dash line, electrons are back-donated from chromium to graphene conduction band, which are less than that of back transfer to carbonyls).

pheric pressure to 20 mTorr in 2 min) at the reaction temperature (130 °C, see Methods section for reaction details). The functionalization reaction can be expressed as

a group VI−B transition metal tricarbonyl [Cr(CO)3] with the benzenoid ring in graphene are involved, such that there is a donation of conjugated π-electrons from the six-membered ring to the unoccupied d-orbitals of Cr. In theory, the grafting of Cr(CO)3 or chromium benzene complexes on graphene surface not only lowers the conduction band of graphene and increase its conductivity investigated via density functional theory (DFT)30 but also leads to much lesser electron−phonon scattering in contrast to covalently functionalized graphene.31,32 This is indeed affirmed by our observation that the charge carrier mobility is sustained and the conductivity is enhanced in graphene post η6-functionalization. Furthermore, the highly stable nature of hexahapto bonds between graphene and Cr(CO)3 (binding energy ranging from 20−23 kcal/mol33 via a DFT simulation) makes it superior to other noncovalent functionalization approaches. While nanoparticle-interfacing on η6-functionalized graphene has not been shown, the haptic functionalization of graphene (and other graphitic materials) has been realized earlier through a liquid phase route.34−36 Such approaches, however, require long reaction time in organic medium, which eventually leads to adsorbed impurities and defects on graphene surface.37,38 In contrast, the vapor-phase functionalization route for graphene demonstrated here allows shorter reaction time and contamination-free surface functionalization, which can be advantageous for large-area integration required for photovoltaic and lithography applications. The carbonyl moieties obtained after the nondestructive η6functionalization of graphene can be used to tune the surface chemistry and allow integration with other materials to form a wide range of composites in which the superior properties of graphene remain unperturbed.39,40 Results and Discussion. Here, vapor-phase hexahapto functionalization was carried out on graphene sheet (produced via chemical vapor deposition (CVD) on copper foil, see details in Supporting Information Section 1) transferred onto a SiO2(285 nm)/Si wafer of 1 × 1 cm2 area. The functionalization was carried out in a quartz tube furnace under vacuum with chromium hexacarbonyl [Cr(CO)6] powder (100 mg) as the precursor (Figure 1A). Because Cr(CO)6 may be lost at a high rate due to its sublimation (important for the reaction) under vacuum, the chamber was gradually evacuated (from atmos-

Graphene([C6]n ) + Cr(CO)6 → (η6 − Graphene)Cr(CO)3 ([C6]n − 1 − C6 − Cr(CO)3 ) + 3CO

The complete process involves sublimation of Cr(CO)6 followed by mass-transfer onto the graphene surface and finally reaction with the benzenoid rings on graphene. The reaction temperature was optimized to be 130 °C for our system (see details in Supporting Information Section 2) by considering the interplay between the rates of sublimation, mass transport, and surface reaction. Because the haptic functionalization is selflimiting and ceases after a monolayer formation of Cr(CO)3 on graphene, the excess solid precursor is expected to sublimate and flow out without reacting with graphene’s surface. Thus, the reaction time was optimized to be approximately 40 min to completely consume the Cr(CO)6 precursor. This was followed by NP attachment as discussed later. To understand the influence of hexahapto functionalization on graphene’s properties and to further leverage this unique chemistry, it is critical to understand the charge transfer mechanism between carbon and the zerovalent transition metals (Cr0) in the resultant chemical structure (Figure 1B). Post hexahapto functionalization of graphene, the Cr(CO)3 moiety places itself on the top of benzenoid ring’s center, resembling an inverted piano stool, as shown in the inset of Figure 1B. Concurrently, the other three carbonyls detach from Cr and are expected to form carbon monoxide (CO) as byproducts.41 The native electronic structure of graphene results from an overlapping of pz orbitals in carbon atoms that form the π-cloud filled with electron and vacant π*-cloud.42 To some extent, the bonding of Cr(CO)3 moieties to graphene sheet is similar to the interaction within (η6-benzene)Cr(CO)3 complex,43 in which σ-bonding and π-back-bonding exit synchronously. The σ-coordination bonding involves a transfer of the π electrons from the graphene valence band to the unoccupied a1 and e orbitals of Cr(CO)3 group.44 As a result, the valence shells (3d, 4s, and 4p orbitals) of the transition metal atom, Cr([Ar]3d54s1), are occupied by six electrons from the bound 4382

DOI: 10.1021/acs.nanolett.7b01458 Nano Lett. 2017, 17, 4381−4389

Letter

Nano Letters

Figure 2. Raman spectroscopic analysis and X-ray photoelectron spectroscopy of pristine graphene and η6-functionalized graphene. (A,D) Raman maps of D/G intensity ratio (ID/IG, ranging from 0 to 0.9) for graphene (A) and η6-functionalized graphene (D). (B,E) G-band position mapping (ranging from 1580 to 1592 cm−1) for graphene (B) and η6-functionalized graphene (E), scale bar is 500 nm. (C) Raman spectra obtained by averaging the circle area of graphene without (blue line) and with hexahapto functionalization (red line). (F) G-band Raman peak-shift before (blue line) and after η6-functionalization (red line), respectively. (G) High resolution of Cr 2p peaks located at 586.56 and 576.86 eV. (H,I) High resolution C 1s peaks with deconvolution by Gaussian function for graphene (H) and η6-functionalized graphene (I), respectively.

distance of 2.95 ± 0.05 Å from graphene’s planar surface.51 Herein, the hexahapto bonding in (η6-graphene)Cr(CO)3 is expected to retain the planarity and thus preserves the superior properties of graphene. To study graphene’s structural planarity and charge carrier mobility after hexahapto functionalization, we have conducted various spectroscopic and low-temperature electrical transport measurements on graphene before and after bonding of Cr(CO)3 moieties. The first insight was obtained by Raman spectroscopy, an inelastic scattering-based vibrational spectroscopic technique, routinely used for probing two-dimensional nanomaterials’ phonon vibration modes,53 which can be influenced by doping levels.54 The Raman spectra of graphene on SiO2/Si chip before and after the vapor-phase reaction with Cr(CO)6 are shown in Figure 2. The same region of graphene was scanned to obtain the spatial mapping of graphene’s Gband peak position (from 1580 to 1592 cm−1) and the intensity ratio between D- and G-band peaks (ID/IG) via confocal scanning Raman spectrometer with 532 nm laser excitation (spot-size ∼721 nm). The Raman spectra in Figure 2C,F were obtained by averaging spectra of the circled area in Figure

graphene’s benzenoid ring and another six electrons from the three carbonyl groups (3CO) attached to Cr, which accords with the 18-electron rule.45 The π-acceptor interaction involves a back-transfer of the electrons from the highest occupied molecular orbitals (HOMO) of Cr(CO)3 to the empty antibonding orbitals of graphene (conduction band).46,47 The aspect in which (η6-benzene)Cr(CO)3 complex and the bonding within (η6-graphene)Cr(CO)3 differs is that there are less electrons back-donated to the graphene than carbonyls due to a high density of electrons in graphene31,48,49 as shown by Sarkar et al., and theorized by Armstrong et al. Therefore, there is a net loss of electrons from graphene’s valence band, leading to hole (p-type) doping in graphene. Further, similar to the structure of (η6-arene)Cr(CO)350−52 and as proved by DFT electronic structure calculations,40 all six carbons of graphene’s benzenoid ring are at a same bonding length (2.25 ± 0.05 Å) from Cr and have the same angle (120 ± 0.01°) between carbon atoms. This suggests that the spatial continuum of sp2 planar lattice of graphene is preserved post η6-functionalization. In this structure, the oxygen atoms in Cr(CO)3 are at a plane parallel to the graphene surface with a 4383

DOI: 10.1021/acs.nanolett.7b01458 Nano Lett. 2017, 17, 4381−4389

Letter

Nano Letters

Figure 3. Anchoring of AgNPs on the surface of (η6-graphene)Cr(CO)3. (A) Schematic depicting the anchoring of silver nanoparticles on η6functionalized graphene via chemical reduction. (B) Raman enhancement of G band by 230% for AgNPs@η6-G (red curve) in comparison with pristine graphene (black curve). The laser power for Raman spectra was 10 mW. The integration time was 0.2 s for one accumulation (with the same objective 100×) and the data was recorded after 10 accumulations. The Raman data was normalized by the Raman band of silicon wafer at 520 cm−1. (C,D) FESEM images of pristine graphene (C) and η6-functionalized graphene (D) with AgNPs, respectively. The size of AgNPs are ranging from 50 to 100 nm.

2p3/2, respectively.36 For pristine graphene (Figure 2H), the observed C 1s peak can be deconvoluted into three peaks (using Gaussian fit), centered at EB of 284.31, 286.22, and 288.57 eV assigned to sp2 CC bond, CO bond, and carbonyl bond, respectively.58 The presence of oxy-functionalized carbon can be attributed to either poly(methyl methacrylate) (PMMA) residue or adventitious carbon.59,60 An increase in the ratio of carbonyl to sp2 CC peak from 0.08 (pristine graphene) to 0.14 was observed. Moreover, in the case of (η6-graphene)Cr(CO)3, the main peak of C 1s shifted to a higher EB by 0.20 eV (Figure 2I) indicating the incorporation of electron withdrawing functional group (CrCO3) because the extra Coulombic interaction between the X-ray excited electrons and carbon ion core consumes more energy.61 For comparison, the XPS spectrum for (η6graphene)Cr(CO)3 obtained by a solution phase is presented in the Supporting Information Figure S3. Next, we leveraged the carbonyl functional groups (−CO) in (η6-graphene)Cr(CO)3 as anchoring sites62 to attach metal NPs on graphene. A schematic of the mechanism underlying the in situ formation of AgNPs on (η6-graphene)Cr(CO)3 is shown in Figure 3A, where Ag+ is reduced to Ag0 by hydroxylamine (3AgNO3 + NH2OH → 3Ag + 3HNO3 + NO).63 These Ag0 radicals can nucleate, combine with other radicals and form nanocrystalline particles. The accessibility to unshared electron pairs of the oxygen atoms on the hexahapto-ligand can aid in the nucleation of Ag0 or attachment of the Ag-nuclei on graphene surface.62 These graphene-anchored nuclei continue to grow into larger particles by allowing incorporation of more Ag0 radicals onto the growing nanocrystal lattice and/or by allowing aggregation of nuclei from the solution.64 Because the NP coverage at low density is not complete, this implies that

2A,B,D,E. The D-band is known to originate from the breathing mode of sp2 carbon atoms. From Figure 2A,C,D, the ID/IG ratio (ranging from 0 to 0.9) changes insignificantly (increased by only 1.1% ± 0.6%) after hexahapto functionalization, which implies that η6-functionalization is nondestructive to the continuous sp2 carbon lattice of graphene. Furthermore, doping would result in the stiffening of in-plane vibrations and sharpening of the G-band.55 Figure 2B,E,F shows a blue shift in G-band position after functionalization (∼3 cm−1 upshift). Moreover, the full width at half-maximum (fwhm) of G peak (Lorentzian fitting) decreased after functionalization with an accompanying decrease in the ratio between intensity of 2D and G peaks (I2D/IG) by 35% ± 2%, which is consistent with the Raman spectra obtained in electrochemically gated graphene.54 On the basis of the above results, we have estimated a doping level of ∼3 × 1012 cm−2 as per the theory developed by Das et al.54,56 These observations combined with electrical properties (shown later) suggest that upon hexahapto functionalization with electron-withdrawing groups Cr(CO)3, graphene starts exhibiting hole (p) doping.57 The functionalization of graphene with metal carbonyls is expected to shift the charge neutrality point (CNP) toward positive biases, as shown through ab initio investigations by Plachinda et al., who have theorized that metal carbonyls act as weak p-dopants.31 For comparison, the Raman spectrum of η6-functionalized graphene via solution-based process is shown in the Supporting Information Figure S1. To understand the binding attributes of Cr(CO)3 moieties upon hexahapto functionalization, X-ray photoelectron spectroscopy (XPS, Kratos AXIS-165) was performed (Figure 2G− I). The high-resolution spectra for Cr core level peaks, as represented in Figure 2G, indicate peaks at binding energies (EB) of ∼586.56 and ∼576.86 eV, assigned to Cr 2p1/2 and Cr 4384

DOI: 10.1021/acs.nanolett.7b01458 Nano Lett. 2017, 17, 4381−4389

Letter

Nano Letters

Figure 4. Graphene/n-Si heterojunction photovoltaic cells. (A) Schematic depicting graphene/n-Si solar cell structure with Ti/Au as top contacts and gallium−indium eutectic as back electrode. (B) Optical image of the solar cell device with a 0.04 cm2 active area. (C) Energy band diagram of AgNPs@η6-G/n-Si Schottky junction. ΦG and ΦSi, work function of η6-funtionalized graphene and n-Si, respectively; ΦSBH, barrier height; χSi, electron affinity of n-silicon. (D) J versus V characteristics curves of solar cells with AM 1.5G illumination; the inset is an enlarged view of G/Si and η6-G/Si. (E) Series resistance (Rs) values extracted from dV/d ln I versus J curves under dark condition.

nondestructive nucleation of AgNPs on η6-functionalized graphene/n-Si solar cells enables the realization of a high enhancement of PCE by slightly increasing the carrier mobility (shown later). To fabricate the G/n-Si Schottky junction solar devices (Figure 4A) with a top-grid structure,73 the monolayer graphene synthesized by CVD process was transferred onto nSi surfaces followed by electrode fabrication via standard photolithography (and metallization process), hexahapto functionalization, and AgNPs anchoring. The light scattering image of the fabricated device with an active area of 0.04 cm2 is presented in Figure 4B along with the corresponding electronic band structure diagram (Figure 4C). In this device, upon illumination (AM 1.5G), the photogenerated excitons (holes and electrons) are separated by a built-in electric field and holes transferred toward graphene and electrons toward n-Si.69,73,74 Upon efficient collection of light-induced charge carriers at the respective electrodes, the device exhibits an open-circuit voltage (VOC) and short-circuit current density (JSC). The results for current density (J) versus voltage (V) characteristics for the solar devices in the presence of illumination (AM 1.5G) are presented in Figure 4D. The solar cell employing η6-functionalized graphene interfaced with AgNPs (AgNPs@η6-G/n-Si) exhibits an open-circuit voltage (Voc) of 0.28 V, a short-circuit current density (Jsc) of 21 mA/ cm2, and a fill factor of 0.22 with a PCE of 1.24%, which is 11fold larger than G/n-Si solar cells (PCE = 0.11%) and 5-fold enhanced compared with the efficiency of solar cells that employ η6 functionalized graphene (PCE = 0.26%) (η6-G/n-Si solar cell). This enhancement can be attributed to three mutually inclusive phenomena. First, the attachment of AgNPs is expected to enhance the conductivity of both graphene and η6 functionalized graphene. In Figure 4E, the series resistances (Rs) of AgNPs@η6-G/n-Si and η6-G/n-Si devices were decreased to 6.16 and 8.13 Ω, respectively, from 10.64 Ω for G/n-Si device, as determined from the slope of dV/d(ln J) versus J curve of the dark J−V characteristics.75 Second, there is a possible increase in graphene’s work function (ΦG) postfunctionalization and AgNPs attachment as suggested by

the radicals can move along the surface to incorporate into the low-potential lattice of growing particles Ag m + (η6 − Graphene)Cr(CO)3 Ag n → (η6 − Graphene)Cr(CO)3 Ag n + m

The size and density of AgNPs are dependent on the reaction temperature and duration.65 Here, the spherical AgNPs were produced at temperature of 13 °C (see details in Methods section) with a size ranging from 50 to 100 nm. The FESEM image in Figure 3D indicate a large number of AgNPs anchored on hexahapto-functionalized graphene (AgNPs@η6-G). In contrast, the unfunctionalized graphene (Figure 3C) shows significantly less NPs attachment (presumably on defect-sites). Further, the attachment of NPs on (η6-graphene)Cr(CO)3 results in an enhancement of Raman scattering signal, which can be attributed to excitation of localized surface plasmon resonance (with high enhancement factor of 103) or chemical enhancement from charge-transfer complexes (with low enhancement factor of 2−100).14,65,66 The Raman intensity of the G-band exhibits ∼2.3-fold increase at the regions of AgNPs anchored on η6-functionalized graphene as shown in Figure 3B, suggesting surface chemical enhancement or surface-enhanced Raman scattering. As mentioned before, the anchoring of AgNPs on η6functionalized graphene with retained charge carrier mobility and sp2 lattice planarity will considerably broaden the scope of graphene’s functionality. We show that this system (AgNPs) can be interfaced with n-silicon (n-Si) to create graphene/n-Si (G/n-Si) 2D/3D heterojunction solar cells67 (Figure 4A) with remarkably enhanced power conversion efficiency (PCE). Currently, chemical doping of graphene by boron68 and volatile oxidants including HNO3,69 H2O2, and HCl70 is used to enhance the PCE (by 2−3 fold) via tuning the work function of graphene and reducing its sheet resistance. However, the resultant defects on graphene destroy its sp2 character and reduce charge carrier mobility, which limits the gain in PCE of these graphene-based solar cells.71,72 In contrast, here the 4385

DOI: 10.1021/acs.nanolett.7b01458 Nano Lett. 2017, 17, 4381−4389

Letter

Nano Letters

Figure 5. Electrical transport measurement in graphene and (η6-graphene)Cr(CO)3 FET devices. (A) Schematic of graphene FET device. VDS is source-drain voltage, VDS = 1 mV. VBG is back gating voltage. (B) Sheet conductivity versus VBG at 10 K of graphene before (black curve) and after (red curve) hexahapto functionalization. The inset is an optical image of graphene FET device with source-drain contacts (Cr/Au). Scale bar is 20 μm. (C) Resistivity versus VBG − VD at 10 and 300 K for η6-functionalized graphene.

a higher Voc obtained for AgNPs@η6-G/n-Si and η6-G/n-Si solar cells.76 The p-doping effect due to hexahapto functionalization results in a downshift of Fermi level (EF) by 172 meV (see Supporting Information Section 9 for details), which further increases ΦG and thus the Schottky barrier height (as ΦSBH = ΦG − χSi, where χSi is the electron affinity of n-Si) and built-in electric field (Vbi, linearly proportional to Voc). Third, the plasmonic effects associated with AgNPs in AgNPs@η6-G/ n-Si solar cells are expected to substantially enhance the photocurrent response in solar cells.77 In the case of plasmonics-based enhancement, light is preferentially scattered by localized surface plasmon resonance of AgNPs. It leads to the trapping, coupling, and absorption of light in the AgNPs@ η6-G/n-Si solar cells.78,79 This is clearly observed from the extraordinary improvement of Jsc (4-fold) for AgNPs@η6-G/nSi in contrast to η6-G/n-Si solar devices. The electrical transport measurements were also performed for graphene before and after hexahapto functionalization at different temperatures. Figure 5A depicts the schematic of a back-gated graphene field effect transistor (FET) device. The inset of Figure 5B shows an optical image of a typical device (10 μm L × 5 μm W) with Cr/Au (15/95 nm) as the source and drain contacts. The η6-G device exhibits p-doping characteristics (Dirac voltage, VD = +43 V); at the same time the bare graphene device is almost ambipolar (VD ∼ 0 V) as shown in Figure 5B. This is consistent with the enhanced πback-donation to the carbonyl ligands as explained earlier, as well as the charge transfer mechanism suggested by the Raman spectral analysis presented in the preceding section. We calculated the field effect mobility μFE, which is derived from ⎛ 1 ⎞ dσ the Drude formula,80 μFE = ⎜ C ⎟ dVsh , where σsh is the ⎝ SiO2 ⎠ BG

functionalization, which is also observed in other two devices, can be explained by two dominant phenomena: (1) as described above, because the hexahapto functionalization does not alter the continuum of graphene’s sp2 structure, the shortrange scattering (from point defects) is unchanged; and (2) uniform attachments of Cr(CO)3 functional groups provide dipole screening, which neutralizes the charge impurity potential63,83−85 from the underlying SiO2 substrate. This screening also preserves the minimum conductivity, and the voltage peak width in our devices. Consequently, the η6functionalization can either maintain or reduce the Columbic scattering in CVD graphene. While preserving the carrier mobility of graphene devices, the hexahapto functionalization also maintains other device performance metrics. The temperature dependence of the (η6-graphene)Cr(CO)3 device’s resistivity on applied back-gate voltage (VBG − VD) at low temperature (T = 10 K, black curve), and at room temperature (T = 300 K, red curve) are shown in Figure 5C. As temperature increases, the (η6-graphene)Crdρ (CO)3 device exhibits a nonmetallic behavior (dT < 0) in the low density regime (n < |n*|, near Dirac point), and a dρ pronouncedly metallic behavior (dT > 0) in the high density 86,87 regime (n > |n*|). In addition, the concentration of the holes doped because of the hexahapto functionalization was estimated to be 3.1 × 1012 cm−2 (see Supporting Information Section 11 for detailed derivations). This hole doping contributes to an increase in the sheet conductivity from 0.26 mS (graphene) to 0.55 mS (η6-functionalized graphene). These observations support that vapor-phase hexahapto functionalization of large area graphene not only preserves its high charge carrier mobility with a slight increase but also enhances its electrical conductivity. Conclusions. In summary, we have demonstrated a scalable, vapor-phase process to functionalize the benzenoid rings of graphene (ring-centered functionalization) to create sites to incorporate AgNPs (50 to 100 nm) while retaining sp2 planar lattice, which results in sustained charge carrier mobility. The heterojunction photovoltaic device with AgNPs anchored graphene interfaced with n-Si exhibited an 11-fold enhancement of power conversion efficiency. Further, the AgNPs attachment resulted in a 230% enhancement in Raman G-band intensity indicating a clear signature of surface-enhanced Raman spectroscopy. The hexahapto functionalization is

( )

sheet conductance and calculated as σsh =

1 L 81 , R tot − R con W

Rtot is

the total resistance, and Rcon is the contact resistance. After subtracting the contact resistance obtained via four-point probe measurements (10800 Ω μm, see details in Supporting Information Section 10), μFE was estimated to be 3398 and 3111 cm2 V−1 s−1 at n = 2 × 1012 hole/cm2 for (η6graphene)Cr(CO)3 and graphene devices, respectively. Mobility measurements on AgNPs laden graphene is known to show an unaccountable capacitive gating from AgNPs82 (nongrounded) and was therefore not conducted. The slight enhancement of carrier mobility in graphene post hexahapto4386

DOI: 10.1021/acs.nanolett.7b01458 Nano Lett. 2017, 17, 4381−4389

Letter

Nano Letters

1165 remover and IPA. The samples were annealed under vacuum at 200 °C to remove residuals. The back-gating measurement in Figure 5 was carried out with a source meter (Keithley 2612). All the electrical transport data in the main text were obtained under vacuum (∼10−3 Pa) in a variable temperature probe station (ARS cryostat). Fabrication of Solar Cell Devices and Photovoltaic Measurement. Top electrodes (Ti/Au, 10 nm/50 nm) were deposited via electron beam evaporation on n-Si substrates and subsequently, electrode patterns were fabricated by photolithography and etching processes. Gallium−indium eutectic was screen-printed on rear side of silicon as back contact electrodes. Current density versus voltage (J−V) measurements of solar cells were performed under 100 mW/cm2 illuminations (AM1.5G filter) and presented in Figure 4D.

found to result in hole (p-type) doping of graphene as affirmed by the blue shift of Raman G-band (3 cm−1), the decrease of I2D/IG (35% ± 2%), and sharpening of G-band (22% ± 5%). Furthermore, the ID/IG changes negligibly (1.1% ± 0.6%) and the charge carrier mobility was enhanced by 9%, which further reflects the unique feature of η6-functionalization to preserve graphene’s planar structure and its high charge carrier mobility. This scalable, nondestructive graphene functionalization and NP incorporation chemistry developed here can enable significant expansion of its scope of graphene interfaces and applications. Materials and Methods. Synthesis of (η6-Graphene)Cr(CO) 3. The vapor-phase hexahapto functionalization of graphene was carried out in a quartz tube furnace connected to a mechanical pump. A SiO2 (285 nm)/Si chip with CVD graphene transferred onto it (see details of graphene growth and transfer in the Supporting Information Section 1) was placed inside the quartz tube near the center. Chromium hexacarbonyl (Cr(CO)6) powder, weighing 100 mg, was placed in a quartz boat at the center of the heating zone at a distance of 3 cm apart from the graphene sample. Initially, the furnace temperature was increased to 130 °C at the rate of 3 °C/min maintaining 1 atm pressure. Then the chamber was slowly evacuated by a gradual opening of the vacuum valve (from atmospheric pressure to 20 mTorr in 2 min). In the next step, the vapor phase reaction with Cr(CO)6 was carried out for 40 min at 20 mTorr. After the reaction, the furnace was allowed to cool down at a rate of 3 °C/min with the chamber at a pressure of 4 mTorr. To compare with reaction in solution, we also carried out the process in a Parr high pressure chamber (see details in the Supporting Information Section 3). Silver Nanoparticles Attachment. The η6-functionalized graphene and pristine graphene chips were immersed in 50 mL of silver nitrate solution (0.2 mM). Then 1.3 μL of hydroxylamine (NH2OH, 50 wt %) was immediately added to reduce the metal ions in the solution for 12 h at a low temperature of 13 °C. The samples were washed thoroughly by deionized water and isopropyl alcohol (IPA) to remove the unbound nanoparticles. The surface morphology of samples was further analyzed by FESEM (JSM-6320F). Raman Spectroscopy and Mapping. Raman spectroscopy and spatial Raman mapping were performed using confocal Raman microscope (Raman-AFM, WITec alpha 300 RA, laser wavelength of 532 nm). The laser spot size was 721 nm with a 100× objective lens (numerical aperture = 0.90). All the Raman maps had a pixel size of 0.08 μm for both x- and y-directions. The same 3 × 3 μm2 region of graphene on SiO2/Si was scanned to obtain the spatial mapping of G peak position (from 1580 to 1592 cm−1) and the intensity ratio between D-band and G-band (ID/IG). Device Fabrication and Electronic Transport Measurements. The electrical contacts (Cr/Au, 15 nm/95 nm) were deposited on transferred graphene chips via electron beam evaporation. After the positive photoresist (AZ 1818) was spincoated onto the samples at 4000 rpm for 40 s. The samples were then baked on the hot plate at 110 °C for 1 min. The contact pattern was introduced onto the samples using photolithography (Karl Suss MA6 mask aligner, 900 W UV power). The graphene channel with length of 10 μm and width of 5 μm (as shown in the manuscript Figure 5B) was defined via oxygen reactive ion etching (RIE, Oxford) to remove the nonprotected graphene. Subsequently, the covered photoresists were removed by submerging the samples into the baths of



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.7b01458. Graphene growth by CVD and graphene transfer process, vapor-phase η6-functionalization of graphene temperature testing, synthesis method of η6-functionalized graphene obtained via an in-solution process, Raman characteristic of η6-functionalized graphene obtained via an in-solution process, methods to scan same area of graphene in Raman measurements, XPS characteristics of η6-functionalized graphene obtained via an in-solution process, UV−vis spectroscopy of silver nanoparticles, the attachment of silver nanoparticles on unfunctionalized graphene, estimation of Fermi energy changed due to η6-functionalization, estimation of contact resistance via four-point probe measurements and doping concentration (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Kabeer Jasuja: 0000-0002-0766-2192 Vikas Berry: 0000-0002-1102-1996 Author Contributions

V.B. conceived the project. V.B., S.C., and K.J. led the experiment design. S.C. performed the vapor-phase reaction, Raman spectroscopy, XPS, AgNPs synthesis, FESEM, and analyzed the experimental data. S.B and S.C. performed the photovoltaic experiments and analysis. P.N. and S.C. conducted the electrical transport measurements and analysis. All authors contributed to writing the manuscript. K.J. and T.S.S. worked on the hexahapto reaction in solution phase. All of the authors contributed to the scientific discussion. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work made use of instruments in the Electron Microscopy Service (Research Resources Center, UIC) and Nanotechnology Core Facility of University of Illinois at Chicago. S.C. acknowledges Mr. Shikai Deng for his help in FESEM. V.B. acknowledges financial support from National Science 4387

DOI: 10.1021/acs.nanolett.7b01458 Nano Lett. 2017, 17, 4381−4389

Letter

Nano Letters

(32) Plachinda, P.; Evans, D.; Solanki, R.; Plachinda, P.; Evans, D. R. J. Chem. Phys. 2011, 135, 044103. (33) Liu, S.; Lei, Y.; Li, Y.; Zhang, T.; Wang, H.; Lan, Y. RSC Adv. 2014, 4, 28640. (34) Tian, X.; Sarkar, S.; Moser, M. L.; Wang, F.; Pekker, A.; Bekyarova, E.; Itkis, M. E.; Haddon, R. C. Mater. Lett. 2012, 80, 171− 174. (35) Sarkar, S.; Zhang, H.; Huang, J.-W.; Wang, F.; Bekyarova, E.; Lau, C. N.; Haddon, R. C. Adv. Mater. 2013, 25, 1131−1136. (36) Kalinina, I.; Bekyarova, E.; Sarkar, S.; Wang, F.; Itkis, M. E.; Tian, X.; Niyogi, S.; Jha, N.; Haddon, R. C. Macromol. Chem. Phys. 2012, 213, 1001−1019. (37) Wang, H.; Lu, S.; Zhang, Y.; Lan, F.; Lu, X.; Xiang, Y. J. Mater. Chem. A 2015, 3, 6282−6285. (38) Xu, L.; McGraw, J. W.; Gao, F.; Grundy, M.; Ye, Z.; Gu, Z.; Shepherd, J. L. J. Phys. Chem. C 2013, 117, 10730−10742. (39) Avdoshenko, S. M.; Ioffe, I. N.; Cuniberti, G.; Dunsch, L.; Popov, A. a. ACS Nano 2011, 5, 9939−9949. (40) Dai, J.; Zhao, Y.; Wu, X.; Zeng, X. C.; Yang, J. J. Phys. Chem. C 2013, 117, 22156−22161. (41) Atkins, P. W. Shriver Atkin’s Inorg. Chem. 2010, 851. (42) Liu, H.; Liu, Y.; Zhu, D. J. Mater. Chem. 2011, 21, 3335. (43) Rossi, A. R.; Sanda, P. N.; Silverman, B. D.; Ho, P. S. Organometallics 1987, 6, 580−585. (44) Jiménez-Halla, J. O. C.; Robles, J.; Solà, M. J. Phys. Chem. A 2008, 112, 1202−1213. (45) Spessard, G. O.; Miessler, G. L. Organometallic chemistry; Oxford University Press, 2010; pp 53−56. (46) Nunzi, F.; Mercuri, F.; De Angelis, F.; Sgamellotti, A.; Re, N.; Giannozzi, P. J. Phys. Chem. B 2004, 108, 5243−5249. (47) Nunzi, F.; Mercuri, F.; Sgamellotti, A.; Re, N. J. Phys. Chem. B 2002, 106, 10622−10633. (48) Armstrong, R. S.; Aroney, M. J.; Barnes, C. M.; Nugent, K. W. J. Mol. Struct. 1994, 323, 15−28. (49) Sarkar, S.; Niyogi, S.; Bekyarova, E.; Haddon, R. C. Chem. Sci. 2011, 2, 1326. (50) Rees, B.; Coppens, P. Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem. 1973, 29, 2516−2528. (51) Ercoli, R.; Calderazzo, F. Communications 1959, 287, 2271− 2272. (52) Patra, M.; Merz, K.; Metzler-Nolte, N. Dalton Trans. 2012, 41, 112−117. (53) Ferrari, A. C.; Basko, D. M. Nat. Nanotechnol. 2013, 8, 235−246. (54) Das, A.; Pisana, S.; Chakraborty, B.; Piscanec, S.; Saha, S. K.; Waghmare, U. V.; Novoselov, K. S.; Krishnamurthy, H. R.; Geim, a K.; Ferrari, a C.; Sood, a K. Nat. Nanotechnol. 2008, 3, 210−215. (55) Basko, D.; Piscanec, S.; Ferrari, A. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 80, 165413. (56) Pisana, S.; Lazzeri, M.; Casiraghi, C.; Novoselov, K. S.; Geim, A. K.; Ferrari, A. C.; Mauri, F. Nat. Mater. 2007, 6, 198−201. (57) Ziegler, T.; Heber, U. Beilstein J. Org. Chem. 2012, 8, 1059− 1070. (58) Yang, S.-Y.; Chang, K.-H.; Tien, H.-W.; Lee, Y.-F.; Li, S.-M.; Wang, Y.-S.; Wang, J.-Y.; Ma, C.-C. M.; Hu, C.-C. J. Mater. Chem. 2011, 21, 2374−2380. (59) Lin, Y. C.; Lin, C. Y.; Chiu, P. W. Appl. Phys. Lett. 2010, 96, 133110. (60) Pirkle, A.; Chan, J.; Venugopal, A.; Hinojos, D.; Magnuson, C. W.; McDonnell, S.; Colombo, L.; Vogel, E. M.; Ruoff, R. S.; Wallace, R. M. Appl. Phys. Lett. 2011, 99, 122108. (61) Zhang, S.; Li, L.; Kumar, A. Materials characterization techniques; CRC Press, 2008; pp 42−43. (62) Tolstov, A. L.; Lebedev, E. V. Theor. Exp. Chem. 2012, 48, 213. (63) Sreeprasad, T. S.; Nguyen, P.; Kim, N.; Berry, V. Nano Lett. 2013, 13, 4434−4441. (64) Polte, J.; Ahner, T. T.; Delissen, F.; Sokolov, S.; Emmerling, F.; Thünemann, A. F.; Kraehnert, R. J. Am. Chem. Soc. 2010, 132, 1296− 1301. (65) Jasuja, K.; Berry, V. ACS Nano 2009, 3, 2358−2366.

Foundation (Grants CMMI-1030963 and CMMI-1503681) and University of Illinois at Chicago.



REFERENCES

(1) Allen, M. J.; Tung, V. C.; Kaner, R. B. Chem. Rev. 2010, 110, 132−145. (2) Castro Neto, A. H.; Guinea, F.; Peres, N. M. R.; Novoselov, K. S.; Geim, A. K. Rev. Mod. Phys. 2009, 81, 109−162. (3) Mayorov, A. S.; Gorbachev, R. V.; Morozov, S. V.; Britnell, L.; Jalil, R.; Ponomarenko, L. A.; Blake, P.; Novoselov, K. S.; Watanabe, K.; Taniguchi, T.; Geim, A. K. Nano Lett. 2011, 11, 2396−2399. (4) Zhang, Y.; Tan, Y.-W.; Stormer, H. L.; Kim, P. Nature 2005, 438, 201−204. (5) Lim, D.-K.; Barhoumi, A.; Wylie, R. G.; Reznor, G.; Langer, R. S.; Kohane, D. S. Nano Lett. 2013, 13, 4075−4079. (6) Su, J.; Cao, M.; Ren, L.; Hu, C. J. Phys. Chem. C 2011, 115, 14469−14477. (7) Crisan, L.; Crisan, B.; Soritau, O.; Baciut, M.; Biris, A. R.; Baciut, G.; Lucaciu, O. J. Appl. Toxicol. 2015, 35, 1200−1210. (8) Wu, T.; Liu, S.; Luo, Y.; Lu, W.; Wang, L.; Sun, X. Nanoscale 2011, 3, 2142. (9) Bonaccorso, F.; Sun, Z.; Hasan, T.; Ferrari, A. C. Nat. Photonics 2010, 4, 611−622. (10) Tjoa, V.; Chua, J.; Pramana, S. S.; Wei, J.; Mhaisalkar, S. G.; Mathews, N. ACS Appl. Mater. Interfaces 2012, 4, 3447−3452. (11) Scheuermann, G. M.; Rumi, L.; Steurer, P.; Bannwarth, W.; Mülhaupt, R. J. Am. Chem. Soc. 2009, 131, 8262−8270. (12) Hu, C.; Liu, Y.; Qin, J.; Nie, G.; Lei, B.; Xiao, Y.; Zheng, M.; Rong, J. ACS Appl. Mater. Interfaces 2013, 5, 4760−4768. (13) Dutta, S.; Ray, C.; Sarkar, S.; Pradhan, M.; Negishi, Y.; Pal, T. ACS Appl. Mater. Interfaces 2013, 5, 8724−8732. (14) Lu, G.; Li, H.; Liusman, C.; Yin, Z.; Wu, S.; Zhang, H. Chem. Sci. 2011, 2, 1817. (15) Zhang, Y.; Liu, S.; Wang, L.; Qin, X.; Tian, J.; Lu, W.; Chang, G.; Sun, X. RSC Adv. 2012, 2, 538−545. (16) Si, Y.; Samulski, E. T. Chem. Mater. 2008, 20, 6792−6797. (17) Qiu, J.-D.; Wang, G.-C.; Liang, R.-P.; Xia, X.-H.; Yu, H.-W. J. Phys. Chem. C 2011, 115, 15639−15645. (18) Wu, Z.-S.; Yang, S.; Sun, Y.; Parvez, K.; Feng, X.; Müllen, K. J. Am. Chem. Soc. 2012, 134, 9082−9085. (19) Li, Y.; Wang, H.; Xie, L.; Liang, Y.; Hong, G.; Dai, H. J. Am. Chem. Soc. 2011, 133, 7296−7299. (20) Pumera, M. Energy Environ. Sci. 2011, 4, 668. (21) Gómez-Navarro, C.; Weitz, R. T.; Bittner, A. M.; Scolari, M.; Mews, A.; Burghard, M.; Kern, K. Nano Lett. 2007, 7, 3499−3503. (22) Mkhoyan, K. A.; Contryman, A. W.; Silcox, J.; Stewart, D. A.; Eda, G.; Mattevi, C.; Miller, S.; Chhowalla, M. Nano Lett. 2009, 9, 1058−1063. (23) Feng, W.; Long, P.; Feng, Y. Y.; Li, Y. Adv. Sci. 2016, 3, 1500413. (24) Leenaerts, O.; Peelaers, H.; Hernández-Nieves, A. D.; Partoens, B.; Peeters, F. M. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 82, 195436. (25) Li, B.; Zhou, L.; Wu, D.; Peng, H.; Yan, K.; Zhou, Y.; Liu, Z. ACS Nano 2011, 5, 5957−5961. (26) Niyogi, S.; Bekyarova, E.; Itkis, M. E.; Zhang, H.; Shepperd, K.; Hicks, J.; Sprinkle, M.; Berger, C.; Lau, C. N.; DeHeer, W. A.; Conrad, E. H.; Haddon, R. C. Nano Lett. 2010, 10, 4061−4066. (27) Sreeprasad, T. S.; Berry, V. Small 2013, 9, 341−350. (28) Georgakilas, V.; Otyepka, M.; Bourlinos, A. B.; Chandra, V.; Kim, N.; Kemp, K. C.; Hobza, P.; Zboril, R.; Kim, K. S. Chem. Rev. 2012, 112, 6156−6214. (29) Georgakilas, V.; Tiwari, J. N.; Kemp, K. C.; Perman, J. A.; Bourlinos, A. B.; Kim, K. S.; Zboril, R. Chem. Rev. 2016, 116, 5464− 5519. (30) Gloriozov, I. P.; Marchal, R.; Saillard, J. Y.; Oprunenko, Y. F. Eur. J. Inorg. Chem. 2015, 2015, 250−257. (31) Plachinda, P. Diss. Theses 2012. 4388

DOI: 10.1021/acs.nanolett.7b01458 Nano Lett. 2017, 17, 4381−4389

Letter

Nano Letters (66) Cheng, C.-E.; Lin, C.-Y.; Chang, H.-Y.; Huang, C.-H.; Lin, H.Y.; Chen, C.-H.; Hsu, C.-C.; Chang, C.-S.; Sen Chien, F. S. Opt. Express 2013, 21, 6547−6554. (67) Behura, S. K.; Nayak, S.; Mukhopadhyay, I.; Jani, O. Carbon 2014, 67, 766−774. (68) Li, X.; Fan, L.; Li, Z.; Wang, K.; Zhong, M.; Wei, J.; Wu, D.; Zhu, H. Adv. Energy Mater. 2012, 2, 425−429. (69) Liu, X.; Zhang, X. W.; Meng, J. H.; Yin, Z. G.; Zhang, L. Q.; Wang, H. L.; Wu, J. L. Appl. Phys. Lett. 2015, 106, 233901. (70) Cui, T.; Lv, R.; Huang, Z.-H.; Chen, S.; Zhang, Z.; Gan, X.; Jia, Y.; Li, X.; Wang, K.; Wu, D.; Kang, F. J. Mater. Chem. A 2013, 1, 5736. (71) Bi, H.; Huang, F.; Liang, J.; Xie, X.; Jiang, M. Adv. Mater. 2011, 23, 3202−3206. (72) Yang, L.; Yu, X.; Xu, M.; Chen, H.; Yang, D. J. Mater. Chem. A 2014, 2, 16877−16883. (73) Wang, Y.; Chen, C.; Fang, X.; Li, Z.; Qiao, H.; Sun, B.; Bao, Q. J. Solid State Chem. 2015, 224, 102−106. (74) Miao, X.; Tongay, S.; Petterson, M. K.; Berke, K.; Rinzler, A. G.; Appleton, B. R.; Hebard, A. F. Nano Lett. 2012, 12, 2745. (75) Cheung, S. K.; Cheung, N. W. Appl. Phys. Lett. 1986, 49, 85−87. (76) Liu, X.; Zhang, X. W.; Yin, Z. G.; Meng, J. H.; Gao, H. L.; Zhang, L. Q.; Zhao, Y. J.; Wang, H. L. Appl. Phys. Lett. 2014, 105, 183901. (77) Nakayama, K.; Tanabe, K.; Atwater, H. A. Appl. Phys. Lett. 2008, 93, 121904. (78) Atwater, H. A.; Polman, A. Nat. Mater. 2010, 9, 205−213. (79) Pillai, S.; Catchpole, K. R.; Trupke, T.; Green, M. A. J. Appl. Phys. 2007, 101, 093105. (80) Dean, C. R.; Young, A. F.; Meric, I.; Lee, C.; Wang, L.; Sorgenfrei, S.; Watanabe, K.; Taniguchi, T.; Kim, P.; Shepard, K. L.; Hone, J. Nat. Nanotechnol. 2010, 5, 722−726. (81) Di Bartolomeo, A.; Giubileo, F.; Romeo, F.; Sabatino, P.; Carapella, G.; Iemmo, L.; Schroeder, T.; Lupina, G. Nanotechnology 2015, 26, 475202. (82) Sreeprasad, T. S.; Nguyen, P.; Kim, N.; Berry, V. Nano Lett. 2013, 13, 4434−4441. (83) Shishir, R. S.; Ferry, D. K. J. Phys.: Condens. Matter 2009, 21, 232204−4. (84) Liang, S.-Z.; Chen, G.; Harutyunyan, A. R.; Sofo, J. O. Phys. Rev. B: Condens. Matter Mater. Phys. 2014, 90, 115410. (85) Liao, L.; Duan, X. Mater. Sci. Eng., R 2010, 70, 354−370. (86) Heo, J.; Chung, H. J.; Lee, S.-H.; Yang, H.; Seo, D. H.; Shin, J. K.; Chung, U.-I.; Seo, S.; Hwang, E. H.; Sarma, S. D. Arxiv Prepr. arXiv 2010, 35421, 14. (87) Bolotin, K. I.; Sikes, K. J.; Hone, J.; Stormer, H. L.; Kim, P. Phys. Rev. Lett. 2008, 101, 96802.

4389

DOI: 10.1021/acs.nanolett.7b01458 Nano Lett. 2017, 17, 4381−4389