Spatially Resolved Imaging of Inhomogeneous ... - ACS Publications

Mar 11, 2005 - I. Correlation of Localized Structural, Electronic, and. Chemical Properties ..... signature spectra acquired at three unique morpholog...
2 downloads 0 Views 747KB Size
Download PDF
Langmuir 2005, 21, 3521-3528

3521

Spatially Resolved Imaging of Inhomogeneous Charge Transfer Behavior in Polymorphous Molybdenum Oxide. I. Correlation of Localized Structural, Electronic, and Chemical Properties Using Conductive Probe Atomic Force Microscopy and Raman Microprobe Spectroscopy Todd M. McEvoy† and Keith J. Stevenson* Department of Chemistry and Biochemistry, Center for Nano- and Molecular Science and Technology, Texas Materials Institute, The University of Texas at Austin, Austin, Texas 78712 Received November 5, 2004. In Final Form: February 7, 2005 A detailed study of electrochemically deposited molybdenum oxide thin films has been carried out after they were sintered at 250 °C. Conductive probe atomic force microscopy (CP-AFM), Raman microscopy, and X-ray photoelectron spectroscopy (XPS) techniques were employed to assess the complex structural, electronic, and compositional properties of these films. Spatially resolved Raman microprobe spectroscopy studies reveal that sintered molybdenum oxide is polymorphous and phase segregated with three types of domains observed comprising orthorhombic R-MoO3, monoclinic β-MoO3, and intermixed R-/β-MoO3. CP-AFM studies conducted in concert with Raman microprobe spectroscopy allowed for correlation between specific compositional regions and localized electronic properties. Single point tunneling spectroscopy studies of chemically distinct regions show semiconducting current-voltage (I-V) behavior with the β-MoO3 polymorph exhibiting higher electronic conductivity than intermixed R-/β-MoO3 or microcrystalline R-MoO3 domains. XPS valence level spectra of β-MoO3 films display a small structured band near the Fermi level, indicative of an increased concentration of oxygen vacancies. This accounts for the greatly enhanced electronic conductivity of β-MoO3 as these positively charged cationic defects (anion vacancies) act to trap excess electrons. Connections between structural features, electronic properties, and chemical composition are established and discussed. Importantly, this work highlights the value of using spatially resolved techniques for correlating structural and compositional features with electrochemical behaviors of disordered, mixedphase lithium insertion oxides.

Introduction The introduction of new redox-active transition metal oxides capable of reversible lithium insertion is of continual interest for advanced lithium ion energy storage devices.1-3 These materials have also drawn significant technological attention for use as catalysts,4 electrochromics,5 and chemical sensors.6 Recently, several soft chemistry routes7 have been explored in an effort to improve metal oxide properties, as well as to prepare unique metastable phases with unusual valence states. In particular, chemical and electrochemical deposition,8 sol-gel9 processing, and surfactant templating10 methods have shown to be efficient in preparing new materials with improved structural * Corresponding author. Fax: 512-471-8696. E-mail: [email protected]. † Current address: Air Products and Chemicals, Inc., 7201 Hamilton Blvd., Allentown, PA 18195. (1) Winter, M.; Besenhard, J. O.; Spahr, M. E.; Nova´k, P. Adv. Mater. 1998, 10, 725. (2) Julien, C. M. Mater. Sci. Eng., R 2003, 40, 47. (3) Long, J. W.; Dunn, B.; Rolison, D. R.; White, H. S. Chem. Rev. 2004, 104, 4463. (4) Weckhuysen, B. M.; Wachs, I. E. In Handbook of Surfaces and Interfaces of Materials; Nalwa, H. S., Ed.; Academic Press: New York, 2001; Vol. 1, p 613. (5) Granqvist, C. G. Handbook of Inorganic Electrochromic Materials, 1st ed.; Elsevier: Amsterdam, 1995. (6) Varghese, O. K.; Grimes, C. A. J. Nanosci. Nanotechnol. 2003, 3, 277. (7) Rouxel, J.; Tournoux, M.; Brec, J. Soft Chemistry Routes to New Materials; Trans Tech Publications: Aedermannsdorf, Switzerland, 1993. (8) Manthiram, A.; Kim, J. Chem. Mater. 1998, 10, 2895. (9) Rolison, D. R.; Dunn, B. J. Mater. Chem. 2001, 11, 963. (10) Liu, P., Zhang, J.-G.; Tracy, C. E.; Turner, J. A. Electrochem. Solid State Lett. 2000, 3, 163.

properties (e.g., increased surface area and porosity) and electrochemical characteristics (e.g., increased ionic transport and higher rate capabilities). For example, Smyrl and co-workers11,12 have demonstrated that sol-gel derived V2O5 aerogels exhibit significantly higher lithium insertion capacities (>4 Li+ per V2O5) compared to coarsegrained crystalline analogues (∼1 Li+ per V2O5). Even though a great number of experimental studies have been devoted to the structural and physical characterization of these materials, the mechanism by which lithium is inserted is not clear. Understanding of electron and lithium transport within the oxide lattice is crucial for optimizing lithium capacities. Additionally, knowledge of the controlling mechanisms associated with electrochemically induced phase transformations that occur during lithium insertion/extraction is vital for improving cycling lifetimes and for alleviating capacity fading problems. Furthermore, this information is essential for establishing material design parameters in the optimization of future lithium ion cell architectures. Fundamental understanding of lithium insertion/ extraction processes has been significantly hindered by the lack of suitable surface-analytical techniques for resolving structure/composition/reactivity relationships. This is chiefly due to the fact that most effective lithium insertion oxides are disordered and do not possess longrange order, contain morphological inhomogeneities (e.g., networks, tunnels, and dislocations), and have unique (11) Le, D. B.; Passerini, S.; Tipton, A. L.; Owens, B. B.; Smyrl, W. H. J. Electrochem. Soc. 1995, 142, L102. (12) Le, D. B.; Passerini, S.; Guo, J.; Ressler, J.; Owens, B. B.; Smyrl, W. H. J. Electrochem. Soc. 1996, 143, 2099.

10.1021/la047276v CCC: $30.25 © 2005 American Chemical Society Published on Web 03/11/2005

3522

Langmuir, Vol. 21, No. 8, 2005

compositions and defect chemistry (e.g., mixed valency, oxygen deficiencies, and cation vacancies). Additionally, very few experimental techniques have been developed to follow lithium insertion reactivity in situ, with even fewer methods suitable for kinetic studies. With this in mind, we have been developing new spectroelectrochemical methodologies13-15 to better correlate how structural and compositional features influence electrochemical charge storage and coloration processes in disordered, heterogeneous metal oxides. In a preliminary investigation,14 we described the physical and electrochemical characterization of MoO3 thin films prepared by electrodeposition. These films constitute a suitable model system because the electrodeposition route allows for systematic regulation of structural and compositional properties. The electrochemical responses of these films toward lithium insertion/extraction and electrocoloration were observed to be strongly dependent upon thermally induced changes in micro-/nanocrystallinity. In particular, MoO3 films sintered at 250 °C were found to consist of a poorly defined mixed phase material that exhibited a more stable and reversible lithium insertion response over amorphous and more crystalline phases. Previously, we were unable to fully understand the influence of the structural and compositional characteristics of this mixed-phase material because the standard characterization techniques employed only reported on the ensemble averaged properties. Specifically, we were unable to ascertain which phases and/or structural features are accessible for electrochemically driven lithium insertion and which phases and/or domains suffer irreversible structural changes during specific insertion/ extraction conditions. We feel better comprehension of controlling factors can be significantly aided by spatially resolved measurements to reduce ambiguity and allow for direct correlation of structure-property relationships. Here we report, in a two-part study, the development of state-of-the-art imaging strategies for the spatially resolved interrogation of lithium ion transport behavior at polymorphous MoO3 thin films. In part I, we detail the correlation of localized structural, electronic, and compositional properties with electrochemical lithium insertion behavior using a combination of conductive probe atomic force microscopy (CP-AFM), Raman microprobe spectroscopy, and spectroelectrochemical microscopy techniques. This approach enables us to directly probe physical and chemical features and to acquire knowledge about relative distributions and arrangements of associated materials properties. In part II of this study, we report on a newly established spectroelectrochemical imaging methodology that relies on simultaneously controlling the applied electrochemical potential while measuring the current and spatially monitoring the intrinsic electrocoloration response of polymorphous MoO3 thin films during lithium insertion. The electrocoloration response serves as a localized reporter on the insertion behavior because these are coupled processes. As we demonstrate, this enables us to directly assess lithium ion insertion energetics and kinetics and to monitor morphological changes occurring at uniquely identified phases. Importantly, for the first time, we are able to simultaneously quantify lithium diffusion coefficients, ionic conductivities, and lithium insertion capacities at energetically and compo(13) Stevenson, K. J.; Hupp, J. T. Electrochem. Solid State Lett. 1999, 2, 497. (14) McEvoy, T. M.; Stevenson, K. J.; Hupp, J. T.; Dang, X. Langmuir 2003, 19, 4316. (15) McEvoy, T. M.; Stevenson, K. J. J. Am. Chem. Soc. 2003, 125, 8438.

McEvoy and Stevenson

sitionally distinct lithium insertion sites contained within mixed-phase, polymorphous MoO3. Experimental Section Electrodeposition of Molybdenum Oxide. Molybdenum oxide thin films were electrochemically deposited onto pre-cleaned transparent conductive indium tin oxide (ITO) coated glass substrates (Delta Technologies, Ltd., 15 Ω/0) from acidic peroxopolymolybdate solutions as previously described.16 As detailed in a previous report,14 post-deposition sintering allows for tuning of the desired film properties (e.g., crystallinity, film thickness, water content, and valency). In the current study, we employed three different post-deposition heat treatments to modify the as-deposited MoO3 films. Method 1 was used to prepare polymorphous films comprising mixtures of R- and β-MoO3 with the β-MoO3 component present at 30 in. Hg) and heated at 200 °C for 1 h. These films were then placed in a tube furnace (Thermcraft, Inc.) and sintered in flowing O2 (40 mL/min) for 1 h at 300 °C. Method 3 was employed to prepare crystalline films comprising ∼100% R-MoO3. This method followed the same steps as those employed in method 1; however, the sintering temperature was raised to 450 °C. All films prepared from any of the three methods described above were stored in a desiccator until use. CP-AFM Characterization. Topographic and conductivity maps of molybdenum oxide films were obtained using a Digital Instruments Bioscope atomic force microscope interfaced with a Nanoscope IV controller. For these measurements, the atomic force microscope scan head was fitted with an ultralow current sensing preamplifier, which operates in the 60 fA-120 pA range (TUNA module, Digital Instruments). CP-AFM studies were performed in contact mode with Pt-coated, silicon cantilevers (MikroMasch, Ultrasharp cantilevers, length 125 µm, resonance frequency ∼160 kHz, tip radius 250 °C induces a gradual phase transition, through water loss and film densification, to produce microcrystalline domains of R-MoO3 ranging in size from a few micrometers up to 1015 µm in length. However, sintering at intermediate temperatures around 250 °C produces disordered, polymorphous films where microcrystalline R-MoO3 domains coexist with a nanocrystalline phase that is amorphous to X-rays. Figure 1b shows the simultaneously acquired tunneling current image taken under a low applied potential bias (+0.5 V). Clearly, the current distribution is nonuniform, where the brighter areas depicted correspond to regions that exhibit higher tunneling current responses due to more facile conduction of electrons between the tip and the sample. We observe that the current contrast is a function of the applied bias voltage and increases in tunneling currents are seen when higher bias voltages are employed. At the low +0.5 V bias applied for films shown in Figure 1, we estimate that that ∼26.7% of the film area is highly conductive (>50 pA tunneling currents) while ∼73.3% behaves as though it was nearly electronically insulating. Low bias voltages are applied in all characterization studies to emphasize differences

Figure 1. Simultaneously acquired (a) topography and (b) current images (75 × 75 µm2) of a polymorphous MoO3 thin film. Three points of interest corresponding to a microcrystalline region (point A), a nanocrystalline region (point B), and a mixed phase region (point C) are highlighted. Shown in part c is the composite image where the current map is overlaid on the topography.

in measured currents between the distinct structural phases. As we demonstrate below, the current response can be used to exclusively identify specific MoO3 polymorphs. Figure 1c shows a composite image created from the combination of the topography (Figure 1a) and current (Figure 1b) images. Visibly, the areas that exhibit higher current correlate well with the nanocrystalline domains. From a purely structural standpoint, this observation seems counter-intuitive as higher tunneling currents are

3524

Langmuir, Vol. 21, No. 8, 2005

Figure 2. Localized I-V spectra obtained at structurally distinct areas comprising (a) a microcrystalline domain (open circles) and a nanocrystalline area (open triangles). (b) Magnification of the I-V curve obtained at the nanocrystalline region.

typically seen at crystalline grain boundaries where fast electronic diffusion often dominates the conduction process over bulk transport routes.18 To better assess the spatially distinct electronic properties of these materials, we performed localized, pointcontact current-voltage (I-V) measurements at specific structural domains corresponding to the microcrystalline regions and to those associated with the nanocrystalline areas. Figure 2 shows the I-V response acquired at a microcrystalline domain (open circles) presumably associated with R-MoO3, because this is the known thermodynamically stable phase. These microcrystalline regions exhibit n-type semiconducting behavior with turnon voltages seen for positive sample biases at about +1.5 V and turn-on negative sample biases at about -3.5 V. In comparison, the I-V response measured at a nanocrystalline region (open triangles) is characterized by lower turn-on voltages of about +0.20 V for positive sample biases and -0.30 V for negative sample biases. The I-V behavior for these nanocrystalline regions is more easily seen in Figure 2b. Both microcrystalline and nanocrystalline domains exhibit asymmetric semiconducting I-V responses where larger currents are measured at positive sample bias voltages than at negative voltages. The observed asymmetric I-V responses are most likely due to factors associated with differences between sample composition and atomic force microscope tip material and (18) McLachlan, D. S.; Blaszkiewicz, M.; Newnham, R. E. J. Am. Ceram. Soc. 1990, 73, 2187.

McEvoy and Stevenson

the geometric properties of the tip and sample,19,20 rather than band bending21 effects associated with the ambient characterization environment. This finding is consistent with the reports of Lucas and co-workers,19 where asymmetric I-V responses are seen when electron tunneling between two dissimilar materials is different in the forward bias (positive sample bias) direction than in reverse bias (negative sample bias). Geometric differences between the tip and the sample (conical vs planar) are also known to influence the shape of the tunneling barrier because the electric field is highest at the curved atomic force microscope tip. As a result, the tunneling barrier tends to bend inward for positive sample bias and bend outward for negative sample bias and effectively produces a lower tunneling barrier in the positive bias direction. It is also highly possible that thermally induced electron population broadening is a contributing factor which in effect acts to smear out the electron distribution in the tip states and thereby increases the current measured in the positive bias direction. We note that attempts to fit the experimental I-V curves to a Fowler-Nordheim tunneling current model were unsuccessful due to the lack of finding a suitable reference system.22 Regardless, the significantly lower turn-on voltages exhibited by the nanocrystalline domains implies that the tunneling barrier is smaller as a result of differences in the molybdenum:oxygen coordination environment and oxygen stoichiometry because these characteristics inherently influence molybdenum oxide electronic properties.23,24 In an effort to more fully correlate the observed nonuniform current response with compositional differences, we employed Raman microprobe spectroscopy utilizing its excellent mapping ability at high spatial resolution (∼1 µm diameter sampling size). By creating reference registry marks on the MoO3 films, we were able to obtain localized Raman spectra at specific areas previously imaged by CP-AFM (such as those labeled in Figure 1). This approach allowed us to directly correlate structural, electronic, and chemical properties measured at identical regions in the sample. Figure 3 shows Raman signature spectra acquired at three unique morphological areas from the same film that was first interrogated by CP-AFM (correspondingly labeled as points A-C in Figures 1 and 3). Unique Raman spectra are observed at each area, where the spectra acquired at point A, corresponding to the microcrystalline domain displayed Figure 1, are consistent with that of R-MoO3. Spectra taken at point B associated with the nanocrystalline morphology are characteristic of β-MoO3.25-27 Spectra obtained at point C exhibited spectral features indicative of the coexistence of intermixed nanocrystalline phases comprising both R-MoO3 and β-MoO3. The signature Raman spectra for each polymorph are corroborated by Raman spectra (19) Lucas, A. A.; Cutler, P. H.; Feuchtwang, T. E.; Tsong, T. T.; Sullivan, T. E.; Yuk, Y.; Nguyen, H.; Silverman, P. J. J. Vac. Sci. Technol., A 1988, 6, 461. (20) Sestovic, D.; Sunjic, M. Solid State Commun. 1996, 98, 375. (21) Stroscio, J. A.; Feenstra, R. M.; Fein, A. P. Phys. Rev. Lett. 1986, 57, 2579. (22) Problems associated with the asymmetric tip/sample junction prohibited fitting the spectra using Fowler-Nordheim tunneling theory. Unfortunately, we were unsuccessful at estimating localized electrical properties such as the metal-oxide potential barrier height. (23) Robin, M. B.; Day, P. In Advances in Inorganic Chemistry and Radiochemistry; Emeleus, H. J., Sharpe, A. G., Eds.; Academic Press: New York, 1967; p 247. (24) Schlenker, C., Ed. Low-Dimensional Electronic Properties of Molybdenum Bronzes and Oxides; Kluwer: Dordrecht, 1989. (25) McCarron, E. M., III. J. Chem. Soc., Chem. Commun. 1986, 4, 336. (26) Carcia, P. F.; McCarron, E. M., III. Thin Solid Films 1987, 155, 53.

MoO3 Structural, Electronic, Chemical Properties

Figure 3. Raman microprobe spectra acquired at the points of interest noted in Figure 1: (a) point A, (b) point B, and (c) point C, which are indicative of R-MoO3, β-MoO3, and intermixed R-/β-MoO3 phases, respectively. The insets in parts a and b are pictorial representations of the crystal structures for the R-MoO3 and β-MoO3 polymorphs.

previously reported by Julien and Nazri28 and by McCarron and Carcia25,26 of MoO3 materials prepared by various routes. We present the specific spectral assignments for the different polymorphs in Table 1 along with a comparative summary of Raman spectral assignments reported in the literature. A detailed discussion on the spectral assignments for both R-MoO3 and β-MoO3 is presented as Supporting Information. Pictorial representations of the crystal structures for both R-MoO3 and β-MoO3 polymorphs29 are also shown in Figure 3. The unit cell dimensions and atomic parameters used to reconstruct the crystal structures were taken from (27) Early reports by Magne´li also described the formation of a similarly structured oxide phase, Mo8O23, that they termed β-MoO3. To reduce any confusion, it should be noted here that our results indicate the presence of β-MoO3 as reported by McCarron and Carcia (refs 25 and 26) and not the presence of Mo8O23. For references on the crystal structure of Mo8O23 we direct the reader to the following: Magne´li, A. Acta Chem. Scand. 1948, 2, 501. (28) Nazri, G. A.; Julien, C. Solid State Ionics 1992, 53-56, 376.

Langmuir, Vol. 21, No. 8, 2005 3525

Andersson and Magneli30 (R-MoO3) and McCarron et al.31 (β-MoO3). Both structures were based on Pbnm symmetry. Orthorhombic R-MoO3 adopts a unique layered structure in which linked, distorted MoO6 octahedra share both corners and edges. As a consequence, in each octahedron one oxygen is unshared, while two oxygen atoms share common octahedra and three other oxygens are partly shared and common to three octahedra. In contrast, β-MoO3 has a distorted monoclinic structure, similar to the monoclinic ReO3 structure, formed by corner sharing MoO6 octahedra, where all oxygen atoms are multiply coordinated except for those on the surface. These structures were created assuming a stoichiometric oxygen: molybdenum ratio; however, this is not always a valid assumption for transition metal oxides. Mestl et al.32 have reported that the degree of nonstoichiometry (oxygen:molybdenum ratio) can be estimated from the ratio of the intensity of the bending modes located at 285 and 295 cm-1 provided that the ratio is between 2.94 and 3.00, because for ratios smaller than 2.94, the two peaks are indistinguishable. As seen for the data presented above for our films, these two Raman bands are not clearly resolved for domains consisting of R-MoO3, β-MoO3, or mixed R-/β-MoO3, suggesting that the O:Mo ratio is smaller than 2.94. This observation is consistent with our previous characterization studies14,16 where the chemical composition of freshly deposited, unsintered molybdenum oxide films had an O:Mo ratio of ∼2.67. As a whole, we broadly estimate, using the XPS data presented below, that the O:Mo ratio for films sintered at 250 °C falls between 2.7 and 2.8 as heat treatment in air acts to reduce the defect concentration by incorporating oxygen into the lattice.32 To establish associations between measured tunneling currents and differences in O:Mo stoichiometry, the electronic (VB) structure of our MoO3 films was examined using XPS. Colton and co-workers33,34 have previously used XPS to identify the presence of a defect band located at slightly lower binding energies than the O(2p) band in substoichiometric MoO3, WO3, and V2O5. Accordingly, we also obtained VB photoelectron spectra for the O(2p) band for two different films with one comprising ∼100% R-MoO3 (prepared via method 3) and the other containing ∼90% β-MoO3 (prepared via method 2), as determined by Raman spectroscopy. Figure 4 presents O(2p) VB spectra for both R-MoO3 and β-MoO3 films. The figure shows broad principle VBs for both polymorphs extending from ∼3 to 11 eV. An additional small peak immediately below the Fermi level at ∼1.5 eV is present for the β-MoO3 polymorph (see inset in Figure 4b). In accordance with previous reports,33,34 this peak was assigned to the existence of an appreciable defect band due to anion vacancies (missing oxygen atoms), which act as electron accumulation sites. Interestingly, the observed peak near the Fermi edge in VB spectra is clearly related to the magnitude of the turnon voltage observed in the I-V response of β-MoO3 (see inset in Figure 4b).35 No similar peak close to the Fermi level is seen for R-MoO3 (see inset in Figure 4a). This (29) Crystal structures were drawn using Shape Software’s, ATOMS crystal drawing program. Dowty, E. ATOMS crystal drawing program, version 5.1; Shape Software: Kingsport, TN, 2000 (www.shapesoftware.com). (30) Andersson, G.; Magneli, A. Acta Chem. Scand. 1950, 4, 793. (31) Parise, J. B.; McCarron, E. M., III; Von Dreele, R.; Goldstone, J. A. J. Solid State Chem. 1991, 93, 193. (32) Dieterle, M.; Weinberg, G.; Mestl, G. Phys. Chem. Chem. Phys. 2002, 4, 812. (33) Rabalais, J. W.; Colton, R. J.; Guzman, A. M. Chem. Phys. Lett. 1974, 29, 131. (34) Colton, R. J.; Guzman, A. M.; Rabalais, J. W. J. Appl. Phys. 1978, 49, 409.

3526

Langmuir, Vol. 21, No. 8, 2005

McEvoy and Stevenson

Table 1. Raman Frequencies and Assignments for r-MoO3 and β-MoO3 R-MoO3 (Nazri/Julien28)

β-MoO3 (McCarron25)

point A (this work) 1009vw 995s

993sb 904m 850vs 817vs

903vw 850vs 820vs

776s 664w 468w

776vs 666w 458wbr

414w 391vw 349m 335m 282m

point B (this work)

289m

313w 284vw

245w 215w 196w

194wbr 176wsh

156m 127w

198vw

assignment28,a

1010vw 998m

Ag: ν(OMo)

850m 818m 776w 663w 463wbr

MosO stretch B1g: ν(OMo2) MosO stretch B3g: ν(OMo3) B1g: ν(OMo3)

337w

OsMosO deformation OsMosO deformation Ag: δ(OMo3) OsMosO deformation B3g: δ(OMo)

418w 393vw 353m 337m

310w 283w 237w

point C (this work)

283m 244w

202vwbr

160vw

Ag: δ(OMo2) B2g: τ(OdModO twist)c Ag: δ(O2Mo2)n B3g: translational rigid MoO4 chain modec

130w 114m

B2g: lattice modes 91sh

80s

Ag: MosMo mode 76s

a Assignments for β-MoO are from Mattes, R.; Schroder, F. Z. Naturforsch., B: Chem. Phys. 1969, 24, 1095, and Anderson, A. Spectrosc. 3 Lett. 1976, 9, 809. b The terms vw, w, m, s, and vs correspond to very weak, weak, medium, strong, and very strong Raman shifts, while br and sh correspond to peaks that are broad or appear as a shoulder. c Assignments are from ref 32.

indicates that β-MoO3 most likely contains a higher amount of oxygen defects than R-MoO3. Due to sensitivity issues and large sampling spot size (ca. 0.4 mm) associated with the XPS analysis, we are unable to quantitatively determine the exact O:Mo stoichiometry for distinct micrometer-sized MoO3 domains. However, bulk XPS analysis of films comprising R-MoO3 and β-MoO3 consistently exhibit O:Mo ratios ranging from 2.7 to 2.8. Schematically, the band structure of MoO3 can be represented by construction of a simple energy level diagram (Figure 4c). Following the constructs of Goodenough,36 the VB for MoO3 consists of O(2p) states, while Mo(3d) and Mo(4s) states dominate the conduction band (CB). For stoichiometrically pure R-MoO3 and β-MoO3 the band gaps are estimated to be ∼3.1 and ∼2.9 eV, respectively, whereas, sub-stoichiometric MoO3-x compositions have a narrowed band gap and the localization of electrons at oxygen vacancies produce a defect band lying ∼0.5-1.0 eV below the CB.37 Collectively, the Raman and XPS results strongly suggest that the β-MoO3 polymorph is more electronically conductive due to missing lattice oxygen and not solely dominated by differences in the molybdenum-oxygen coordination environment. This interpretation is consistent with reports of Gillet and coworkers38 on tungsten oxide films where they found that substoichiometric WO3-x was significantly more conductive due to the formation of a donor level within the band gap. (35) In the I-V experiment, when the sample bias voltage is varied between -1.0 and 0 V, the current measured corresponds to the physical situation where the electrons tunnel from the occupied states of the sample into the empty states of the tip. Therefore, the -0.3 eV turn-on voltage observed corresponds to a band of occupied states lying about 0.3 eV below the Fermi level. (36) Goodenough, J. B. In Progress in Solid State Chemistry; Reiss, H., Ed.; Pregamon: London, 1971; Vol. 5, p 145. (37) Crouch-Baker, S.; Dickens, P. G. Solid State Ionics 1989, 32/33, 219. (38) Gillet, M.; Lemire, C.; Gillet, E.; Aguir, K. Surf. Sci. 2003, 532535, 519.

The knowledge gained from these studies has allowed us to systematically tune the amount of β-MoO3 contained within MoO3 films through modification of the postdeposition sintering parameters (i.e., raising/lowering the sintering temperature or increasing/decreasing the time in a vacuum and oxygen atmospheres). Following an analogous procedure outlined initially by McCarron25 for spray-deposited MoO3 films, we have prepared electrochemically deposited films consisting of up to 95% β-MoO3 as determined by Raman and CP-AFM measurements. In particular, we find that CP-AFM is exceptionally useful for directly evaluating resultant film compositions and distributions even when the material appears nearly topographically homogeneous. For example, representative CP-AFM images are shown in Figure 5 for a MoO3 film prepared via a modified sintering approach (method 2).39 In contrast to polymorphous films prepared by sintering in air at 250 °C (Figure 1), no distinguishing morphological features are seen (e.g., microcrystallites) and the films are significantly more uniform and smoother (RMS roughness ) 4.3 nm). Although only slight morphological differences are observed, clear distinction between R- and β-MoO3 polymorphs can be made from the current image (Figure 5b). In the upper and lower right portions of this image, the measured current is appreciable (>40 pA) and nearly uniform suggesting the presence of the conductive nanocrystalline β-MoO3 phase. Raman microprobe measurements confirm that these regions indeed comprise β-MoO3 (data not shown). In the remainder of the current image, one can see the outline of several fan-shaped microcrystallites with lower conductivity. These regions are not completely insulating, as measurable amounts of current are observed to flow at the edges of the grain boundaries that make up these (39) The area shown in these images contains ∼30-40% β-MoO3. This region of the film was chosen to highlight the subtle differences in topography and at the same time show that the R- and β-MoO3 phases can be easily distinguished using CP-AFM. Larger scan areas of the same film show that β-MoO3 is present in amounts equaling ∼75%.

MoO3 Structural, Electronic, Chemical Properties

Figure 4. VB X-ray photoelectron spectra of the O(2p) band for a film comprising (a) ∼100% R-MoO3 and (b) ∼90% β-MoO3. Insets: Magnifications of parts a and b showing spectra from 0 to 2.5 eV. (c) Pictorial representation of the energy level diagram for MoO3. The VB, CB, and defect band are denoted.

crystallites. Closer inspection of this image reveals that the small particles within these regions are arranged in aligned, layered structures, characteristic of R-MoO3. Raman microprobe studies of these crystallites substantiate the existence of R-MoO3 within these fan-shaped domains (data not shown). Historically, others investigating these materials inconclusively established relationships between electrochromic and electroinsertion behaviors and specific phases because the characterization techniques used only evaluated their ensemble-averaged properties. As detailed in a previous report,14 cyclic voltammograms (CV) obtained at MoO3 electrodes in 1 M LiClO4/propylene carbonate display multiple Li+ insertion/de-insertion peaks, which is suggestive of energetically distinct, domain-specific reactivity. To more directly evaluate structural and chemical inhomogeneities we developed a new spectroelectrochemical imaging methodology,15 to probe localized electrochemical reaction kinetics and monitor electrochemically induced phase changes. Because the chargetransfer reaction involves a simultaneous visible change in the optical density of the material, the electrochromic response is used to monitor the lithium insertion behavior. The electrochromic response results from intervalence charge-transfer optical transitions between MoVI and MoV sites formed during the electrochemical reduction and

Langmuir, Vol. 21, No. 8, 2005 3527

Figure 5. Simultaneously acquired (a) topography and (b) current images (12.5 × 12.5 µm2) of a MoO3 thin film prepared by sintering an as-deposited film in a vacuum at 200 °C for 1 h, followed by sintering in flowing O2 (40 mL/min) at 300 °C for 1 h. The images were obtained using a scan rate of 0.3 Hz and a sample bias of +0.5 V.

injection of a charge-compensating cation into the lattice (MoVIO3 transparent + xe- + xLi+ h LixMoVI1-xMoVxO3 blue).5 To better understand the relationships between structural and electronic properties and lithium insertion/ coloration behavior, a random area of a polymorphous MoO3 film was imaged using CP-AFM and then the same area was subsequently studied using spectroelectrochemical microscopy. The results of such an experiment are shown in Figure 6. Representative CP-AFM images of a polymorphous MoO3 thin film prepared as described above (method 1) of the topography (Figure 6a) and current (Figure 6b) are shown. Similar topographic features displaying microcrystalline and nanocrystalline domains are observed. The current image is also nonuniform with the nanocrystalline regions exhibiting the highest current response. Figure 6c shows an optical microscopy image of the identical area of the film that was first interrogated by CP-AFM and then mounted in the electrochemical cell and poised at an oxidizing potential (+0.4 V vs Ag/AgCl, de-inserted state) prior to lithium insertion. Images for pristine polymorphous MoO3 films prior to insertion are transparent and consist with randomly oriented grains of dispersed size, orientation, and crystallinity (Figure 6c). Following application of a potential step from +0.4 V to a subsequent reducing potential (-0.8 V, insertion state) time-lapsed transmitted light images (λ ) 630 ( 60 nm)

3528

Langmuir, Vol. 21, No. 8, 2005

McEvoy and Stevenson

degrees of Li+ insertion reactivity. Normally, when the equivalent spectroelectrochemical experiment is conducted using a conventional UV-vis spectrometer, it is assumed that the material responds optically and electrochemically in a homogeneous manner. Clearly, this is not the case for thin films prepared at moderate sintering temperatures (250 °C for 3 h). This may help explain the large variation in reported behaviors for MoO3 prepared by various methods and procedures. The films shown here are not only heterogeneous structurally but also exhibit inhomogeneous ion/charge-transfer reactivity toward lithium insertion. Through spatially resolved elucidation of electrochemical reactivity at individual domains of different crystallinity and chemical composition, we hope to gain a more fundamental understanding of the relationships between surface morphology, chemical composition, and ion/charge-transfer reactivity. A more detailed description of this spectroelectrochemical imaging approach and results obtained using this new methodology to study the coloration/insertion dynamics in polymorphous MoO3 is presented in part II of this study.40

Figure 6. Simultaneously acquired (a) topography and (b) current images (75 × 75 µm2; current scale, 0.0 fA to 30.0 pA) of a polymorphous MoO3 thin film. The scan rate was 0.3 Hz, and the sample bias was +1.5 V. The chronoabsorptometric imaging experiment carried out on the same area of the film shown in parts a and b while immersed in a propylene carbonate solution containing 1 M LiClO4. The images shown are for the film in (c) an oxidized state (+0.4 V vs Ag/AgCl, 0 s) and in (d) a reduced state (-0.8 V, 20 s).

were collected as a function of time. For better visualization, a movie of this imaging experiment can be viewed by accessing the Supporting Information. Images of reduced films (Figure 6d) exhibit variegated behavior indicative of inhomogeneous coloration/insertion at phasesegregated domains. CP-AFM and Raman microprobe spectroscopy experiments (data not shown) performed prior to these optical imaging studies indicate that regions that appear the most transmissive (i.e., those areas that are the least colored) consist primarily of nanocrystalline monoclinic β-MoO3, whereas the domains that are least transmissive (i.e., those that undergo the largest coloration) comprise predominantly orthorhombic R-MoO3. Regions characterized by an intermediate degree of coloration are characteristic of intermixed R- and β-MoO3 phases. The correlation of the domain-specific chemical composition with localized electrochromic behavior enables us to directly estimate the relative contributions of each identified phase by simply thresholding pixel intensities of the images at desired coloration levels. For example, the estimated electroactive areas for coloration/ insertion within each of these distinguished phases in Figure 6 were estimated to be 17.4, 15.8, and 66.8% for R-MoO3, β-MoO3, and intermixed R/β-MoO3 domains, respectively. It is clear from the data presented here that the degree of coloration/insertion is dependent upon the localized chemical composition. The various gray scale intensities observed in distinct regions are indicative of different degrees of coloration and consequently suggest differing

Conclusions In this work CP-AFM, Raman microprobe spectroscopy, and XPS have been used to study variations of electronic conductivity and chemical composition in polymorphous MoO3. CP-AFM images reveal that the electronic conductivity is nonuniformly distributed throughout the MoO3 film. Complementary Raman microprobe spectroscopy measurements indicate that domains that exhibit higher levels of electronic conductivity comprise β-MoO3, while domains that display lower electronic conductivity consist of R-MoO3. Comparison of the X-ray photoelectron spectra of both polymorphs indicates that a small defect band exists in β-MoO3 indicative of the presence of oxygen deficiencies. In part II of this study, we describe the use of spectroelectrochemical microscopy to quantitatively analyze domain-specific coloration/insertion behavior at these polymorphous MoO3 electrodes. The measurement of localized diffusion coefficients, ionic conductivities, and insertion capacities in domains identified by CP-AFM and Raman microprobe spectroscopy as comprising R-MoO3, β-MoO3, and mixed R-/β-MoO3 will be discussed. Acknowledgment. The National Science Foundation (CHE-0134884) and the Robert A. Welch Foundation (Grant F-1529) are gratefully acknowledged for financial support of this work. We also thank Tim Hossain and Lynette Ballast at Advanced Micro Devices, Inc. (Austin, TX), for their assistance in performing the Raman microprobe experiments. Supporting Information Available: Text describing the Raman spectral assignments for both R-MoO3 and β-MoO3 presented in Table 1 (three pages, PDF). A visualization of the integrated spectroelectrochemical imaging experiment that correlates structural, electrical, and electrocoloration properties can be viewed as a movie (in Macromedia Flash or Apple QuickTime format). This material is available free of charge via the Internet at http://pubs.acs.org. LA047276V (40) McEvoy, T. M.; Stevenson, K. J. Langmuir 2005, 21, 3529.