Transparent electronics full report
report word1.docx (Size: 219.23 KB / Downloads: 913)
Transparent electronics is an emerging science and technology field focused on producing ‘invisible’ electronic circuitry and opto-electronic devices. Applications include consumer electronics, new energy sources, and transportation; for example, automobile windshields could transmit visual information to the driver. Glass in almost any setting could also double as an electronic device, possibly improving security systems or offering transparent displays. In a similar vein, windows could be used to produce electrical power. Other civilian and military applications in this research field include realtime wearable displays. As for conventional Si/III–V-based electronics, the basic device structure is based on semiconductor junctions and transistors. However, the device building block materials, the semiconductor, the electric contacts, and the dielectric/passivation layers, must now be transparent in the visible –a true challenge! Therefore, the first scientific goal of this technology must be to discover, understand, and implement transparent high-performance electronic materials. The second goal is their implementation and evaluation in transistor and circuit structures. The third goal relates to achieving application-specific properties since transistor performance and materials property requirements vary, depending on the final product device specifications. Consequently, to enable this revolutionary technology requires bringing together expertise from various pure and applied sciences, including materials science, chemistry, physics, electrical/electronic/circuit engineering, and display science.
Transparent electronics is an emerging science and technology field focused on producing ‘invisible’ electronic circuitry and opto-electronic devices. Applications include consumer electronics, new energy sources, and transportation; for example, automobile windshields could transmit visual information to the driver. Glass in almost any setting could also double as an electronic device, possibly improving security systems or offering transparent displays. In a similar vein, windows could be used to produce electrical power. Other civilian and military applications in this research field include real-time wearable displays. As for conventional Si/III–V-based electronics, the basic device structure is based on semiconductor junctions and transistors. However, the device building block materials, the semiconductor, the electric contacts, and the dielectric/passivation layers, must now be transparent in the visible –a true challenge! Therefore, the first scientific goal of this technology must be to discover, understand, and implement transparent high-performance electronic materials. The second goal is their implementation and evaluation in transistor and circuit structures. The third goal relates to achieving application-specific properties since transistor performance and materials property requirements vary, depending on the final product device specifications. Consequently, to enable this revolutionary technology requires bringing together expertise from various pure and applied sciences, including materials science, chemistry, physics, electrical /electronic/ circuit engineering, and display science.
During the past 10 years, the classes of materials available for transparent electronics applications have grown dramatically. Historically, this area was dominated by transparent conducting oxides (oxide materials that are both electrically conductive and optically transparent) because of their wide use in antistatic coatings, touch display panels, solar cells, flat panel displays, heaters, defrosters, ‘smart windows’ and optical coatings. All these applications use transparent conductive oxides as passive electrical or optical coatings. The field of transparent conducting oxide (TCO) materials has been reviewed and many treatises on the topic are available. However, more recently there have been tremendous efforts to develop new active materials for functional transparent electronics. These new technologies will require new materials sets, in addition to the TCO component, including conducting, dielectric and semiconducting materials, as well as passive components for full device fabrication.
2. COMBINING OPTICAL TRANSPARENCY
WITH ELECTRICAL CONDUCTIVITY
Transparent conductors are neither 100% optically transparent nor metallically conductive. From the band structure point of view, the combination of the two properties in the same material is contradictory: a transparent material is an insulator which possesses completely filled valence and empty conduction bands; whereas metallic conductivity appears when the Fermi level lies within a band with a large density of states to provide high carrier concentration.
Efficient transparent conductors find their niche in a compromise between a sufficient transmission within the visible spectral range and a moderate but useful in practice electrical conductivity. This combination is achieved in several commonly used oxides – In2O3, SnO2, ZnO and CdO. In the undoped stoichiometric state, these materials are insulators with optical band gap of about 3 eV. To become a transparent conducting oxide (TCO), these TCO hosts must be degenerately doped to displace the Fermi level up into the conduction band. The key attribute of any conventional n-type TCO host is a highly dispersed single freeelectron- like conduction band (Figure 1). Degenerate doping then provides both (i) the high mobility of extra carriers (electrons) due to their small effective mass and (ii) lowoptical absorption due to the lowdensity of states in the conduction band. The high energy dispersion of the conduction band also ensures a pronounced Fermi energy displacement up above the conduction band minimum, the Burstein–Moss (BM) shift. The shift helps to broaden the optical transparency window and to keep the intense optical transitions from the valence band out of the visible range. This is critical in oxides which are not transparent throughout the entire visible spectrum, for example, in CdO where the optical (direct) band gap is 2.3 eV.
Fig.1: (a) Schematic electronic band structure of aTCOhost – an insulator with a band gap Eg and a dispersed parabolic conduction band which originates from interactions between metal s and oxygen p states. (b) and © Schematic band structure and density of states of a TCO, where a degenerate doping displaces the Fermi level (EF) via a Burstein-Moss shift, EBM, making the system conducting. The shift gives rise to inter-band optical transitions from the valence band,
3. Electronic Properties of Conventional TCO
Conventional n-type TCO hosts (In2O3, SnO2, CdO and ZnO) share similar chemical, structural and electronic properties. Exclusively oxides of the post transition metals with (n-1)d10ns2 electronic configurations, they have densely packed structures with four- or six-coordinate metal ions. Strong interactions between the oxygen 2p and metal ns orbitals give rise to electronic band
structures qualitatively similar for all these oxides (Figures 1 and 2): the bonding and nonbonding O 2p states form the valence band while the conduction band arises from the antibonding Ms–Op interactions. The empty p states of the metal ion form the following band at a higher energy. The partial density of states plots (Figure 2), reveal that the oxygen 2p and metal ns states make similar contributions to the conduction band. This provides a three dimensional Ms–Op network for charge transport once extra carriers fill the band.
Fig.2: Electronic band structure and partial density of states of TCO hosts, In2O3, SnO2, ZnO and CdO, as obtained within the screened-exchange local-density approximation. In the density of states plots, the thick, dashed and thin lines represent metal s, metal p and oxygen p states, respectively. The plots should be compared with the schematic band structure shown in Figure 1(a).
4. Carrier Generation in Conventional TCO Hosts
The optical and transport properties of a conventional TCO are governed by theefficiency and the specifics of the carrier generation mechanism employed. Evenin the most favorable situation, i.e. when the effects of dopant solubility, clustering,secondary phase formation and charge compensation can be avoided, largeconcentrations of electron donors (substitutional dopants and/or native pointdefects) not only promote the charge scattering but also may significantly alter theelectronic band structure of the host oxide, leading to a nonrigid band shift of theFermi level. A detailed band structure analysis of the doped oxides helps toelucidate the role of different factors involved.
4.1 SUBSTITUTIONAL DOPING
Substitutional doping with aliovalent ions is the most widely used approach togenerate free carriers in TCO hosts. Compared with native point defects, it allowsa better control over the resulting optical and transport properties as well as betterenvironmental stability of the TCO films. Traditionally, same-period, next-rowelements, e.g, Sn4+ for In3+ and In3+ for Cd2+, are thought to provide bettercompatibility and, thus, less disturbance in the host crystal and electronicstructure.However, other dopants may prove beneficial for optimizing theproperties for a specific application. For example, transparent conducting ZnOfilms have been prepared by doping with Group III (Al, Ga, In and B), Group IV (Si,Ge, Ti, Zr and Hf) and a Group VII element (F substituted at an oxygen site),giving rise to a wide range of electrical conductivities.Here we will give a detailed consideration to rocksalt CdO, where the high crystalsymmetry and the densely packed structure ensures the most uniform chargedensity distribution via the isotropicMs–Op network. Compared with morecomplex In2O3 or SnO2, one can expect fewer ionized and neutral scatteringcenters and, hence, longer relaxation times. At the same time, introduction ofdopants into the densely packed structure may significantly influence the Cds–2phybridization and, therefore, alter the structural, electronic and optical propertiesof the host. A systematic comparison of CdO doped with In, Ga, Sc or Y, whoseionic radius and electronic configuration differ from those of the host cation, hasrevealed thati) Substitutional dopants with smaller ionic radii compared with that of Cdshrink the lattice. The shrinkage, however, is not as large as expectedfrom the Vegard’s law weighted average of the six-coordinated X3+ andCd2+ ionic radii. Moreover, in the case of X=In or Y, the latticeparameter is similar or even slightly greater than that of CdO (Table 2).One of the possible explanations is that the doping-induced shrinkageis compensated by an expansion mechanism which originates fromthe antibonding character of the conduction band formed from Cd 5sand O 2p states. The antibonding mechanism is dominant in In or Ydoped CdO, while Sc or Ga have sufficiently smaller ionic radii toweaken Ms–Op hybridization and, thus, to compress the lattice.(ii) Weaker Cd5s–O2p hybridization associated with strong structuralrelaxation around dopant with a smaller ionic radius results in asmaller optical band gap (Table 2). Doping with Ga whose ionic radiusis significantly smaller than that of Cd, reduces the optical band gap(to 2.53 eV) so that it becomes smaller than the one in undoped CdO(2.82 eV) – despite the doping-introduced BM shift of 2.3 eV. Thesmallest optical band gap in Ga-doped CdO as compared with In, Yand Sc cases was observed experimentally.Fig.4: Contour plots of the charge density distribution in In, Y, Sc and Ga-dopedCdO illustrate considerable electron localization around Sc and Ga ions ascompared with In and Y cases where the charge density is more uniform. Theplots are calculated in the xy plane within the 2kT energy window near the Fermilevel. The grey scale increases with charge; the same scale is used for all plots.Atoms within one unit cell are labeled.(iii) In and Y dopants preserve the uniform charge density distribution whileSc and Ga lead to significant electron localization around the dopant(Figure 4). The difference originates from the mismatch of theelectronic configuration of the dopants and the energy location of thedopant empty p or d states with respect to the Fermi level. The Sc 3dstates and Ga 4p states are energetically compatible with theconduction 5s states of Cd, while theY4d and In 5p are located higherin energy. As a result, the contributions from the Sc d or Ga p orbitalsbecome significant near the Fermi level: the Sc d orbital contribution isdominant (85% of the Sc total) and the Ga p and s orbitals givecomparable contributions (60% and 40%, respectively). Theanisotropic Sc d or Ga p orbitals form strong directional bonds with theorbitals of the nearest oxygen atoms resulting in significant chargelocalization which is clearly seen from the charge density distributionplots (Figure 4). (iv) The electron localization in Sc and Ga doped CdO results in a narrowerconduction band and, hence, a reduction of the electron velocity ascompared with In or Y (Table 2). Moreover, due to the high anisotropyof the Sc d or Ga p orbitals, a significantly reduced velocity is found inthe (Sc d orbitals) or (Ga p orbitals) directions so that anisotropictransport properties are expected.(v) The electron binding in Sc and Ga-doped CdO also leads to larger (inenergy) optical transitions from the Fermi level (Ec in Figure 1), incontrast to the In and Y cases where the charge delocalizationdeminishes the second (hybridization) gap.(vi) Finally, we note that even in the In, Yand F cases where the dopant ionicradius and electronic configuration are similar to that of Cd or O, theoptical properties are worse than expected from the rigid band shift(CdO + e-) (Table 2). However, the calculated electron velocity andthe density of states for In, Y and F-doped CdO are similar to thoseobtained from the rigid-band model (Table 2). Both factors contributeto the conductivity s, given by the expressiono that the relaxation time will play the dominat role in determiningthe final carrier transport. [In Equation (1.2) e is the electron charge,is the volume of the Brillouin zone, k is the wave vector, is the bandindex, v is the electron group velocity and EF is the Fermi energy.]Assuming that t is similar for all X3+ -doped systems, estimates of theFermi electron velocity and the density of states at the Fermi levelresult in the trend In>Y>Sc>Ga, which is in agreement withexperimental observations of the conductivity.
4.1 OXYGEN REDUCTION
Removal of an oxygen atom from a metal oxide leaves two extra electrons in thecrystal. Whether one or both of these electrons become free carriers or remainlocalized at the vacancy site correlates with the oxide free energy of formation. Inlight metal oxides, such as CaO or Al2O3, where the formation energy is high,oxygen vacancies create deep charge localized states within the electronic bandgap known as color or F centers. A relatively low formation energy of theconventional TCOs favors large oxygen deficiencies even under equilibriumgrowth conditions, giving rise to the free-carrier densities of forIn2O3 and ZnO.Electronic band structure investigations of oxygen deficient oxides showed thatthe oxygen defect (in notation the superscript stands for effectivepositive charge) corresponds to a non-conducting state associated with the fillingof the lowest single conduction band by the two vacancy-induced electrons. Only ifthe vacancy is excited, e.g. via a photoexcitation , or partially compensatedto ,does the single conduction band become half-occupied and conductingbehavior may occur.In oxygen deficient TCOs, the conduction band wave function resembles the onein the corresponding hosts, i.e. it is derived from the M s and O p states (Figure 1).A relatively uniform charge density distribution suggests that the vacancy-inducedelectrons are delocalized. However, a more thorough analysis of reduced In2O3reveals that the metal atoms nearest to the oxygen defect give about two timeslarger contributions than the rest of the In atoms in the cell. As a result, there is anotable build-up of the charge density near the vacancy site. Importantly, the Inatoms nearest the vacancy exhibit a reduction of the s-orbital contribution: therelative orbital contributions from the In s, p and d states are 81%, 8% and 11%,respectivly, in contrast to 97% s-orbital contributions from other In atoms in thecell. The high anisotropy of the p and d orbitals favors stronger covalent (directional) bonds between the In atoms which surround the defect and theiroxygen neighbors. These In–O pairs trap about 31% of the total charge density atthe bottom of the conduction band. Similar behavior is found for other TCOs: inoxygen deficient CdO and ZnO, 18% and 39%, respectively, of the total chargedensity belong to the nearest (cation) and next nearest (oxygen) neighbors of theoxygen vacancy.The presence of oxygen vacancies leads to significant changes in the electronicband structure of a TCO host. To illustrate the typical behavior, we compare theresults obtained for oxygen deficient and Sn-doped In2O3 (Table 3 and Figure 5):Table3: Properties of oxygen-deficient and Sn-doped In2O3 as obtained fromelectronic band structure calculations within local density approximation. Valuesfor undoped stoichiometric In2O3 found from a rigid band shift are given forcomparison. The electron concentration is for all systems. Theplasma frequency is calculated from Equation (1.3)(i) Strong structural relaxation around the vacancy reduces thedistance between the In and O atoms nearest to the defect to2.12 (on average). This leads to an increased In–O distancesfor the atoms located further from the defect and, hence, to aweaker Ins–Op hybridization. As a result, the fundamental bandgap and the optical transitions from the valence band (Ev) aresignificantly reduced in oxygen-deficient In2O3 as comparedwith Sn-doped oxide. (ii) Owing to the stronger binding between the In and O atoms nearestto the defect, the lowest single conduction state occupied by thevacancy-induced electrons is split from the rest of the conductionband by a second gap. In marked contrast, the second gap isabsent in the substitutionally doped oxide. This is a manifestationof a more uniform spatial charge density distribution, i.e. thecharge delocalization. Note, the second gap previously reportedfor Sn-doped In2O3 vanishes upon structural relaxation aroundSn ions [Figure 5(b)].(iii) The increased charge density in the vicinity of the oxygen vacancyand the related narrowing of the conduction band give rise to thereduced electron velocity (Table 3). At the same time, the densityof states near the Fermi level increases. Since both factorscontribute to the conductivity [Equation (1.2)], the difference inthe charge transport of the oxygen-deficient and Sn-dopedIn2O3 will be determined primarily by the relaxation time in thesame equation. Qualitatively, the stronger structural relaxationwith the atomic displacements around the oxygen vacancy beingtwice as large as those around Sn ions, implies a strongercharge scattering in oxygendeficient oxide. In addition, a shorterelectron relaxation time in this case should be expected due tothe Coulomb attraction of the free carriers to associated with itshigher formation energy compared with that of , which is theground-state defect . Moreover, due to the strong preference ofthe extra electrons to bind with to form , the charge transportwill be adversely affected since the latter defect corresponds to anonconducting state (a completely filled single conduction band).(iv) Due to the narrower conduction band in the oxygen-deficient oxide,the plasma frequency is expected to be significantly smaller thanthat in Sn-doped material. The plasma oscillations affect theoptical properties: the electromagnetic waves offrequency below(and wavelength above) are reflected due to theelectronscreening. The plasma frequency is given by theexpression:where e is the electron charge, is the volume of the Brillouinzone, k is thewave vector, is the band index, is the electrongroup velocity and EF is the Fermi energy. Our estimates forin the oxygen-reduced and Sn-doped In2O3 as well as the oneobtained from the rigid band model are given in Table 3.In summary, compared with substitutional doping, oxygen reduction of a TCO hostmay result in higher carrier densities but would limit the electron mobility due toshorter relaxation times and considerable charge trapping near the vacancy site.Also, a weaker Ms–Op hybridization due to stronger structural relaxation aroundthe vacancy significantly reduces the optical transparency window.There may be other native point defects that give rise to a conducting behavior ina TCO. For example, it was shown that interstitial Sn ions in SnO2 have lowformation energies and produce donor levels inside the conduction band of thismaterial. In this case, significant structural rearrangement associated with theformation of Sn(II)O bonds as in SnO is expected to have an even stronger effecton the properties of the oxide host and to increase electron scattering.The above considerations demonstrate the advantages of employing substitutionaldoping as a primary carrier generation mechanism in conventional TCO hosts.However, notwithstanding the above limitations, we believe that varying thedegree of nonstoichiometry may serve as a versatile tool for optimizing a TCO’soverall performance.
5. TRANSPARENT ELECTRONICS DEVICES
In order to produce a transparent-electronics-based system, appropriatematerials must be selected, synthesized, processed, and integrated together inorder to fabricate a variety of different types of devices. In turn, these devicesmust be chosen, designed, fabricated, and interconnected in order to constructcircuits, each of which has to be designed, simulated, and built in such a waythat they appropriately function when combined together with other circuit andancillary non-circuit subsystems. Thus, this product flow path involves materials→ devices → circuits → systems, with each level of the flow more than likelyinvolving multi-feedback iterations of selection, design, simulation, fabrication,integration, characterization, and optimization.From this perspective, devices constitute a second level of the product flow path.The multiplicity, performance, cost, manufacturability, and reliability of availabledevice types will dictate the commercial product space in which transparentelectronics technology will be able to compete. Thus, an assessment of thedevice toolset available to transparent electronics is of fundamental interest, andis the central theme of this chapter.Passive, linear devices - resistors, capacitors, and inductors – comprise the firsttopic discussed. Passive devices are usually not perceived to be as glamorousas active devices, but they can be enabling from a circuit system perspective,and they are also the simplest device types from an operational point-of-view.Together, these two factors provide the rationale for considering this topicinitially.Next, two-terminal electronic devices - pn junctions, Schottky barriers,heterojunctions, and metal-insulator-semiconductor (MIS) capacitors - constitutethe second major topic. The motivation for this topical ordering is againassociated with their relative operational complexity, rather than their utility. Thethird and final major topic addressed is transistors. This is the most importantmatter considered in this chapter. Most of this discussion focuses on TTFTs,since they are perceived to be the most useful type of transistor for transparentelectronics. Additionally, a very brief overview of alternative transistor types -static-induction transistors, vertical TFTs, hot electron transistors, and nanowiretransistors - is included. This is motivated by recognizing the desirability ofachieving higher operating frequencies than are likely obtainable using TTFTswith minimum gate lengths greater than ~2-10 μm, a probable lower-limitdimensional constraint for many types of low-cost, large-area applications.Alternative transistors such as these offer possible routes for reaching higheroperating frequencies, in the context of transparent electronics.
6. PASSIVE, LINEAR DEVICES
A passive device absorbs energy, in contrast to an active device, which iscapable of controlling the flow of energy (Spencer and Ghausi 2003). A lineardevice is distinguished by the fact that its input-output characteristics aredescribable using a linear mathematical relationship. The three passive, lineardevices of interest are resistors, capacitors, and inductors.
An ideal resistor is a device whose current-voltage characteristics are linear,described by Ohm’s Law, and which dissipates power if a voltage exists acrossit. The two foundational ideal resistor device equations are indicated by the firsttwo entries in Table 4.A real resistor may not be perfectly linear, i.e., precisely obey Ohm’s Law, andmay also possess some undesirable capacitive or inductive parasiticcharacteristics. Transparent thin-film resistors (TTFRs) are expected to operateat relatively low frequencies, so that parasitic inductance is not anticipated to berelevant. Additionally, TTFRs will most likely be fabricated on insulatingsubstrates, so that parasitic capacitance should be minimal. Finally, if properlydesigned, a TTFR is expected to exhibit linear or very near-linear behavior.Thus, in most respects, we expect a TTFR to be adequately modeled as an idealresistor.An equation for the resistance of a TTFR is given by the third entry in Table 4.The resistance depends on a material property, namely the resistivity of theTTFR layer, and the geometry, which is assumed to be rectangular for thesituation considered in Table 4. Given this geometrical constraint, if the currentpath length is assumed to be equal to the crosssectional width, i.e., if L = W, andif a sheet resistance is defined as RS = ρ/t, then the resistance depends simplyon RS and the number of resistor squares in the TTFR layout.Table 4: A summary of resistor device equations.Figure 5 a shows a plan-view of a straight-line or linear TTFR. The TTFR pathconsists of 16 squares, with two larger end squares, generically illustrating onepossible resistor termination scheme. Many variations on resistor termination arepossible. A more complicated, meandering TTFR layout with 36 path squares isgiven in Figure 5b. This TTFR occupies approximately the same area as thestraight-line TTFR of Fig. 5 a, but possesses a larger resistance because of thelarger number of squares in its path and also due to the existence of bends inthe meander structure, which increase the resistance (Glaser and Subak-Sharpe1979; Elshabini-Riad and Barlow 1998).Fig 5: (a) Plan-view of a straight-line or linear transparent thin-film resistor(TTFR) and (b) a meander TTFR. © Cross-sectional view of a TTFR in whichcontacts and passivation are also indicated.Figure 5 c offers a cross-sectional view of a TTFR in which the resistor path andends have the same thickness, which need not always be the case. Contactsfrom the resistor ends to other locations on the substrate are indicated, as is apassivation layer. If the TTFR layer is heavily doped, e.g., ITO, the passivationlayer’s role is merely to provide physical and chemical protection. However, it ispossible that a passivation layer may actually play an active role in establishingthe TTFR resistance if a highly insulating layer is used in the resistor path and itsconductance is established by creation of a surface accumulation layer due tothe presence of the passivation layer. In this case, the resistance would notscale with the TTFR thickness, but would be controlled by the surfaceaccumulation charge due to interface properties and charge within thepassivation layer.TTFR sheet resistances of ~10-105 Ω/□ should be possible, using doped TCOssuch as ITO or undoped TTFT channel layers such as ZTO. Thus, a wide rangeof TTFR resistance is possible.Two thin-film resistor concerns are the desire to have a small temperaturecoefficient of resistance (TCR) and resistor tolerance. Near-zero temperaturecoefficient of resistance values have been demonstrated for antimony- dopedSnO2, with an appropriate doping concentration (Maissel and Glang 1970).TTFR resistance tolerance is expected to be similar to that of conventional thinfilmresistors, ±10%, unless resistor trimming is performed, in which case atolerance of approximately ±0.1% is possible (Glaser and Subak-Sharpe 1979;Elshabini-Riad and Barlow 1998). It is not clear whether resistor trimming will bepractical in a transparent electronics technology, given its anticipated low-coststructure. Smooth surfaces are highly desirable for TTFR applications,suggesting that amorphous layers would be preferred.
An ideal capacitor is an electric field energy storage device possessing linearcurrent-voltage derivative (i-dv/dt) characteristics. Defining ideal capacitorequations are collected in the first three entries of Table 5.A plan-view and a cross-sectional view of a transparent thin-film capacitor(TTFC) are given in Fig. 6. In order for this device to be completely transparent,all of the layers should be transparent. Most insulators are transparent, so thatthis constraint mainly applies to the contact layers, which will most likely behighly conducting TCOs such as ITO. Alternative TTFC structures, e.g.,interdigitated capacitors (Glaser and Subak-Sharpe 1979; Elshabini-Riad andBarlow 1998), are possible in addition to the simple TTFC structure shown inFig.6.Table 5:A summary of capacitor device equations.The primary TTFC design equation is included as the fourth entry in Table 5. TheTTFC capacitance is established by the capacitance density, i.e., thecapacitance per unit area, of the basic capacitance stack, i.e., εS/dI, and thegeometric area of the capacitor layout. Usually a large capacitance density isdesired, in order to minimize the size of a capacitor. Therefore, a thin insulatorwith a high dielectric constant (sometimes referred to as a ‘high-k dielectric’,where ‘k’ denotes the relative dielectric constant) is best. However, a high-kdielectric typically has a smaller bandgap, which usually results in a lowbreakdown electric field. Although reducing the insulator thickness alsoincreases the capacitance density, a minimum thickness is required to avoidpinholes and other types of defects which degrade the breakdown field andwhich yield more insulator leakage. Optimal TTFC and TTFT gate insulators willFig.6: (a) Plan-view and (b) cross-sectional view of a transparent thin-filmcapacitor (TTFC).have maximal dielectric constant-breakdown field products (Ono 1995). For amore detailed discussion of insulators and thin-film capacitors the interestedreader is referred to the following references (Maissel and Glang 1970; Glaserand Subak-Sharpe 1979; Elshabini-Riad and Barlow 1998; Ono 1995; Kao2004).Real TTFCs are expected, for most purposes, to be accurately modeled as idealcapacitors. Although TTFC performance may be degraded by inductive andresistive parasitic effects, neither of these are expected to be severe given thatthese devices are expected to be used at relatively low frequencies and that lowleakagethin-film insulators are already employed in non-transparentapplications. From process integration, manufacturability, and reliabilityconsiderations, TTFC contacts and insulators should ideally be amorphous.
An ideal inductor is a magnetic field energy storage device possessing linearvoltage-current derivative (v-di/dt) characteristics. Important ideal inductorequations are collected in the first two entries of Table 6.In contrast to a TTFR and a TTFC, a transparent thin-film inductor (TTFI) andrelated transparent magnetically-coupled devices are expected to behave in anon-ideal manner. Two main reasons underlie this expectation. First, because ofthe relatively poor conductance of TCOs compared to metals, TTFIs will possessa significant amount of parasitic resistance. Second, efficient magnetic fieldcoupling is strongly facilitated by the use of a magnetically-permeable insulator.However, we are not aware of a transparent, magnetically-permeable insulatormaterial. Thus, realizing high performance TTFIs and related magneticallycoupleddevices is expected to be a challenging task.The last two entries included in Table 6 are useful for understanding certainaspects of TTFI non-idealities. The quality factor, Q, is basically an inductorperformance figure-of-merit. A larger Q is better. Thus, since the parasiticresistance of a TTFI is expected to be large, as a consequence of employing aTCO instead of a metal, high-Q TTFI’s are not expected. The last entry in Table6 indicates that obtaining a large inductance, L, requires the inductor to cover alarge area and to possess a large number of turns. The large-area requirementis not necessarily problematic, since ‘real estate’ if often ‘free’ in transparentelectronics. However, needing to have a large number of turns is likely to causetrouble, since having a large number of turns will increase the inductor parasiticseries resistance, and probably also the inductor parasitic capacitance.Table 6: A summary of inductor device equations.These TTFI challenges are disappointing since a TTFI and its magneticallycoupleddevice variants are potentially application-enabling, performanceenhancingcomponents. Inductors are useful in resonant circuits and filters. Theycan also function as power supply chokes (limiting current fluctuations), energystorage devices for switched-mode power supplies, etc. Furthermore,magnetically-coupled inductors may be used for the construction of transformersfor power and signal conditioning. Finally, a somewhat-related application is thatof an antenna to transmit or receive rf signals. In this regard, we note that atransparent microwave antenna constructed using ITO has been reported(Outaleb et al. 2000).Even though a TTFI with good performance appears to be difficult to construct,the benefits of magnetic field coupling devices appear to offer enoughenticements to warrant further investigation.
7. TRANSPARENT THIN-FILM TRANSISTORS(TTFTs)
TTFTs constitute the heart of transparent electronics. The first two sectionsfocus on ideal and non-ideal behavior of n-channel TTFTs. Next, n-channelTTFT stability is considered. Finally, issues related to alternative devicestructures – double-gate TTFTs and the realization of p-channel TTFTs - arediscussed.Fig.7: Two possible transparent thin-film transistor (TTFT) device structures,(a) astaggered, bottom-gate, and (b) a coplanar, top-gate.Figure 7. illustrates two of four possible TTFT device structures. The first one,considered in Fig. 7a, is denoted as a staggered, bottom-gate since source-drainand gate contacts are located at the top and bottom of the device, respectively.Figure 7b shows a coplanar, top-gate structure in which the source-drain and thegate are all positioned on the top side of the TTFT. The remaining two TTFTdevice structures, not shown, are the staggered, top-gate and coplanar, bottomgateconfigurations. Although in a conventional TFT, the source, drain, and gatecontact materials would be metals, a highly conductive TCO, such as ITO, isused in a TTFT. Additionally, while the channel layer of a conventional TFTemploys a narrow band gap, opaque semiconductor, a highly insulating, wideband gap transparent semiconductor is used in a TTFT.
Ideal operation of an n-channel TTFT is described as follows.
With the source grounded, a positive voltage is applied to the drain in order toattract electrons from the source to the drain. The amount of drain current, ID,which flows from the source to the drain depends upon whether or not anelectron accumulation layer exists at the channel-insulator interface. No draincurrent flows if the gate-source voltage, VGS, is less than the turn-on voltage,VON, since electrons are not present in the channel. This situation is indicated inFig. 8a. Thus, it is the low carrier concentration nature of the channel whichholds off the flow of drain current between the source and the drain for VGS’sbelow VON.If a gate voltage, VGS, greater than VON is applied while a positive voltage existsat the drain, i.e., if VDS is positive, drain current flows, as indicated in Fig. 8b.Physically, electrons are injected from the source into the gate-bias-inducedelectron accumulation layer channel, are transported across this low-resistanceaccumulation layer, and are extracted at the drain contact. This electrontransport process corresponds to the flow of drain current in the oppositedirection, from the drain to the source. The magnitude of drain current flowingdepends on the gate overvoltage, i.e., on VGS-VON, which determines theaccumulation layer sheet charge density, and also on the magnitude of the drainvoltage, which establishes the electric-field aided drift condition along thechannel, from the source to the drain. For small VDS’s compared to the gateovervoltage (i.e., for VDS<<VGS-VON), drain current flow in the channel isdescribable using Ohm’s law, i.e., ID = VDS/RC(VGS), where RC(VGS) is theresistance of the channel, indicating that electron transport across the channelmay be modeled as simply resistive. In fact, the channel resistance is indicatedin Ohm’s law in the functional form RC(VGS) since the channel resistancedepends on the accumulation layer sheet charge density, which is controlledby VGS.As the magnitude of this applied positive drain voltage increases so that VDS isno longer negligible compared to the overvoltage (VGS-VON), ID is no longerOhmic with respect to VDS, but, rather, becomes sublinear and eventuallysaturates when channel pinch-off occurs at VDS≡VDSAT=VGS - VON.Pinch-off may be understood quantitatively by recognizing that the electronaccumulation layer sheet charge density is given by Qn(y)=CG[V(y)- VON], whereV(y) is the channel-insulator interfacial voltage drop along the channel from thesource to the drain, with y=0 and y=L corresponding to distances along thechannel at the edge of the source and the edge of the drain, respectively. At thedrain edge of the channel, Qn(L)=CG[VGS-VDS-VON]. Thus, the accumulation layersheet charge density is equal to zero at y=L, and the channel is thereforedepleted or ‘pinched off’, when the term in the square brackets goes to zero,leading to the pinch-off relationship as specified by the first entry in Table 7. Thisnonlinear, pre-pinch-off to saturated, post-pinch-off situation is sketched in Fig.8c, attempts to capture both the drain voltage-induced elimination of the electronaccumulation layer channel near the drain-end of the channel, and the inherently2- dimensional nature of the TTFT electric field near the drain in this regime ofdevice operation.Fig.8: Ideal n-channel transparent thin-film transistor (TTFT) operation. (a) Cutoff.Zero drain current (ID=0) occurs in cut-off, which is defined by VGS<VON, andcorresponds to a situation in which no electron accumulation layer exists at thechannel-gate insulator interface. (b) Linear, pre-pinch-off. ID is described byOhm’s law [ID=VDS/RC(VGS)] at low VDS’s [VDS<<VGS-VON], corresponding to theformation of a uniform electron accumulation layer at the channelgate insulatorinterface from the source to the drain. © Nonlinear, pre-pinch-off and, postpinch-off, saturation. ID becomes sublinear with respect to VDS and thensaturates when VDS≡VDSAT=VGS-VON because of the depletion or ‘pinch-off’ of theelectron accumulation layer at the channel-gate insulator interface near thedrain. VDSAT defines the boundary between pre-pinch-off and post-pinch-off orsaturation.Table 7: A summary of ideal, square-law theory TTFT device equations.A quantitative formulation of ideal TTFT operation is given according to squarelawtheory (Borkan and Weimer 1963; Tickle 1969; Hong et al. 2007), assummarized in Table 7. Note that the linear or Ohmic regime, as indicated in Fig.8b, corresponds to the pre-pinch-off limit in which VDS<<VGS-VON so thatID≈(W/L)μCG(VGS-VON)VDS, which means that RC(VGS)=(W/L)μCG(VGS-VON). Idealaspects of square-law theory include the absence of subthreshold current (i.e.,the subthreshold swing, S=0) and hard saturation of the drain currentcharacteristics (i.e., in saturation, dID/dVDS = 0).
As the oxide semiconductors are wide band gap materials, transparent TFTs canbe easily realized by the combination of transparent electrodes and insulators.Transparency is one of the most significant features of TAOS TFTs. As the bandgap of a-Si is 1.7 eV and that of crystalline-Si is 1.1 eV, ‘transparent electronics’cannot be realized in Si technology. In TAOS TFTs, features of high mobility orlow process temperature have attracted a lot of attention. However, transparencyhas been underestimated or even neglected in the research and development ofTAOSs. Few examples of actual applications have been reported exploiting thetransparency of TAOSs until now [25, 26]. Transparent circuits will haveunprecedented applications in flat panel displays and other electronic devices,such as seethrough display or novel display structures. Here, practical examplestaking advantage of the transparency of TAOS TFTs are: Reversible Display,‘Front Drive’ Structure for Color Electronic Paper, Color MicroencapsulatedElectrophoretic Display, Novel Display Structure – Front Drive Structure. Indiumoxide nanowire mesh as well as indium oxide thin films were used to detectdifferent chemicals, includingCWA simulants.
9. FUTURE SCOPE
It should be apparent from the discussion that although much progress has beenmade in developing new materials and devices for high performance transparentsolar cells, there is still plenty of opportunity to study and improve deviceperformance and fabrication techniques compared with the nontransparent solarcell devices. In particular, the stability of transparency solar cells has not beenstudied yet. Solution-processable transparent PSCs have become a promisingemerging technology for tandem solar cell application to increase energyconversion efficiency. The transparency of solar cells at a specific light band willalso lead to newapplications such as solar windows. The field of energyharvesting is gaining momentum by the increases in gasoline price andenvironment pollution caused by traditional techniques. Continuedbreakthroughs in materials and device performance, accelerate and establishindustrial applications. It is likely that new scientific discoveries and technologicaladvances will continue to crossfertilize each other for the foreseeable future.
10.CONCLUSION AND REMARKS
Oxides represent a relatively newclass of semiconductor materials applied toactive devices, such as TFTs. The combination of high field effect mobility andlowprocessing temperature for oxide semiconductors makes them attractive forhigh performance electronics on flexible plastic substrates. The marriage of tworapidly evolving areas of research, OLEDs and transparent electronics, enablesthe realization of novel transparent OLED displays. This appealing class of seethroughdevices will have great impact on the human–machine interaction in thenear future. EC device technology for the built environment may emerge as oneof the keys to combating the effects of global warming, and this novel technologymay also serve as an example of the business opportunities arising from thechallenges caused by climate changes The transparency of solar cells at aspecific light band will also lead to newapplications such as solar windows. Thefield of energy harvesting is gaining momentum by the increases in gasolineprice and environment pollution caused by traditional techniques.
Transparent Electronics ’, Springer publications, J.F.Wager, D. A. Keszler, R. E.Presley.‘Transparent electronics: from synthesis to applications’, Wiley publications:Antonio Facchetti, Tobin J. Marks.www.wikipedia.orgwww.ieee.orgwww.sciencemag.orgwww.wiley.com