Electrical characterization of Au/Pd/n-GaN/Pd/Au device structure in the radio frequency range by simulation study

Synthetic Metals 161 (2011) 1434– 1440 Contents lists available at ScienceDirect Synthetic Metals j o ur nal homep ag e: www.elsev ier .com Electric /Au frequen A. Sertap luo a Mugla Univer b Mugla Univer a r t i c l Article history: Received 5 Feb Received in re Accepted 11 M Available onlin PACS: 73.40.Sx 73.40.Qv 73.30.+y 73.40.Ns 84.37.+q 61.80.−x Keywords: Schottky barri C–V and G–V c Interface state Capacitance an g opt tion, r, me oper ol to rk cu evice tra by simulation. All current–voltage characteristics exhibited good rectification behavior. The forward and reverse bias capacitance–voltage (C–V) characteristics of the Au/Pd/n-GaN/Pd/Au device were simu- lated at 2 MHz probing frequency. Nyquist plots of simulated device at 295 K are shown that two-barrier heights can be observable above critical threshold bias point. Frequency depended inverse dielectric loss tangent spectra of Au/Pd/n-GaN/Pd/Au device at different DC bias voltages also shows significant two peaks. This indicates that our simulation is extremely useful for determination of potential barrier 1. Introdu Recently tronic, opto because of high absorp erties for e researches voltaic effic greater tha gap and de superior hi peratures fo GaN is direc at the band be absorbed growth tec ∗ Correspon Department, 4 fax: +90 252 2 E-mail add 0379-6779/$ – doi:10.1016/j. er haracteristics s omaly numbers. © 2011 Elsevier B.V. All rights reserved. ction , GaN have begun to gain attention for use in elec- electronic, lighting, display and data storage industries low effective mass of carriers, high mobility values, tion coefficients, and good electroluminescence prop- lectroluminescence device application [1–8]. Recent offer that in order to put into practice terrestrial photo- iencies of greater than 50%, materials with band gaps n 2.4 eV must be used [9]. GaN have 3.39 eV energy monstrate favorable photovoltaic properties such as, gh energy radiation resistance, tolerance to high tem- r space based PV applications [1–8]. The band gap of t and very strong absorption of approximately 105 cm−1 edge allows a large fraction of the incident light to in a few hundred nanometers of material [1–8]. GaN hnology serves high device-quality crystalline struc- ding author at: Mugla University, Faculty of Sciences, Physics, 8170 Kotekli, Mugla, Turkey. Tel.: +90 252 211 1674; 23 8656. ress: [email protected] (A.S. Kavasoglu). tures and fabricates optoelectronic devices, which supports its potential in high-efficiency photovoltaic. There has been a great many of research on experimen- tal and theoretical studies of the metal–GaN-metal type devices [10–13]. Such studies enables a better understanding the influ- ence of metal–semiconductor junction on device operation. The metal–semiconductor–metal (MSM) device structure is considered back-to-back connected two Schottky contacts [10]. The perfor- mance and stability of these devices are especially dependent on the formation of two-barrier heights at both side of the semicon- ductor. It is well known that the interface state densities, Schottky barrier height as well as ideality factor are influenced from the metal–semiconductor interface [14–19]. Finding the optimal n type GaN fabrication has occupied the researches for years [1,3,5]. Until this day in n type GaN device con- figuration the issues of impedance and admittance spectroscopy results realization remain challenges. While detailed analysis and discussion of I–V–T result can be found nearly in every device related journal and textbook [1,3,5,8] the issue of admittance spec- troscopy result is practically absent. We could not attend any study about the impedance spectroscopy result and rectification proper- ties of metal–GaN–metal device configuration in the literature. For see front matter © 2011 Elsevier B.V. All rights reserved. synthmet.2011.05.018 al characterization of Au/Pd/n-GaN/Pd cy range by simulation study Kavasoglua,b,∗, Nese Kavasoglua,b, Gorkem Oylum sity, Research Laboratory Center, TR-48000 Kotekli/Mugla, Turkey sity, Faculty of Sciences, Physics Department, TR-48000 Kotekli/Mugla, Turkey e i n f o ruary 2011 vised form 8 May 2011 ay 2011 e 12 June 2011 a b s t r a c t GaN based devices are highly promisin tions in photovoltaic energy conversa based devices such as Schottky barrie ricated and characterized so far. A pr result is crucial since it is a powerful to this study, temperature dependent da based metal–semiconductor–metal d / locate /synmet device structure in the radio glua,b oelectronic devices for many years due to their useful applica- fiber optic communication and atmosphere monitoring. GaN tal–semiconductor–metal and p-i-n structure have been fab- understanding of the impedance or admittance spectroscopy calculate the physical and electronic parameters of a device. In rrent–voltage (I–V) and dark impedance spectra of n type GaN have been studied with current–voltage and impedance spec- A.S. Kavasoglu et al. / Synthetic Metals 161 (2011) 1434– 1440 1435 Fig. 1. The cor where ˚B01 a built-in poten that reason tification pr height (BH) data and im barrier heig In this a istics of th with curren ware. Durin is interpre geneities [2 metal/semi 2. Theory In this s structure w calculation carried out Schottky st interface w iteratively, The dev tky diodes have been theory. The device from contains tw that of a me clearly dem lated devic that the n-t tact regime exhibit rect tact this con is deposited a Schottky on the wafe tact usually resistance b mechanism as regions o the least b [14–17]. Th serve as good ohmic contacts. This forces the semiconductor donor density (ND) as the only alternative to engineer contacts. It is well known that, the barrier height width does actually depend strongly on the donor density. For high donor densities, the barrier height is suf or th sem ns ca icon Au/P unde the d e ide bias ates. bia ene re a ight en th t the n un two tivel simu ented clea hott ion le ts ca alled rder rt w pres pend has nce, t–vo V1 + ˛1 = neg responding energy band diagram at thermal equilibrium (0 V dc bias) nd ˚B02 are the electron barrier heights, and Vbi1 and Vbi2, are the tials for contacts. One side of the GaN is heavily doped. , investigation of the impedance spectroscopy and rec- operties is required to display the interface and barrier profiles from the simulated non-ideal current–voltage pedance spectroscopy result in the presence of two- ht. rticle, wide range temperature dependent character- e n type GaN based MSM device have been studied t–voltage and impedance spectra by simulation soft- g the simulation process, Au/Pd/n-GaN/Pd/Au device ted based on the existence of the lateral inhomo- 0–28] of the both barrier height prevailing at the conductor interfaces. tudy using our computer program for a MSM device ith a meaningful potential profile, rigorous numerical of I–V–T, impedance and admittance characteristics was by assuming the back to back connected Au/Pd/GaN ructure. For this, the potential profile of the Schottky as simulated by solving the equivalent circuit equation using a simulation program developed by our group. ice characteristics of a MSM (essentially two Schot- connected back to back) are well known [10] and studied on the basis of the Schottky barrier device principal difference of metal–semiconductor–metal width band f metal– electro the sem The values tics of and th applied trap st applied The structu rier he betwe level a given i for the respec of the is pres It is two Sc condit contac other c In o we sta in the rent de circuit resista curren V(I) = where itive or metal semiconductor device is that a MSM device o depletion regions and the analysis is very similar to tal–semiconductor junction [14–19]. This issue will be onstrated by the Schottky barrier illustrations. Simu- e structure has been illustrated in Fig. 1. One may see ype Schottky diode consists of the two different con- s. One of the contacts calls as front contact it must ifying junction behavior and the other one is back con- tact also calls as ohmic contact. For example, if a metal on the semiconductor (n-type GaN) wafer front to form diode for device fabrication, the same metal deposited r back also forms a Schottky diode [14–17]. Back con- calls as an ohmic contact and generally has much lower ecause carrier injection can be performed by tunneling (field emission). Ohmic contacts are frequently defined f high tunneling rates. Electrons in the metal encounter arrier to their flow into or out of the semiconductor is implies that highly field emission (FE) regions should situation. V rent, V1 and D1) and dio Boltzmann’ n2 the diod equation fo reverse satu I01,2 = AA∗T where A is t tive Richard and ˚B01,2 form the fo ˚B01,2 = kB q ficiently narrow at or near the bottom of the conduction e electrons to tunnel directly, known as FE [14–17]. For iconductor contacts with narrow barrier height widths, n tunnel from the metal to the semiconductor and from ductor to the metal [14–17]. d/n-GaN/Pd/Au device is simulated at different DC bias r the dark condition. Capacitance–voltage characteris- evice provide the determination of charge distribution ntification of the presence of interface states, due to an contributes to carrier capture and emission of interface During the simulation AC frequency is held constant and s voltage is varied in C–V simulation. rgy band diagram for an Au/Pd/n-GaN/Pd/Au device t thermal equilibrium is shown in Fig. 1. The bar- of a Schottky diode, defined as the energy difference e Fermi level in metal and the conduction band energy metal/n-GaN interface, is usually denoted by ˚B0, and its of volt. Where ˚B01 and ˚B02 are the barrier heights contacts and Vbi1 and Vbi2, are the built-in potentials, y. The energy band diagram of Schottky region (diode 1) lated device at Tunneling Field Effect (TFE) conditions in Fig. 1. r that Au/Pd/n-GaN/Pd/Au device structure is basically ky barriers connected back to back [10,15]. Under the ft hand side of the device positive polarity, one of the lled the contact D1 (diode 1) is forward biased and the the contact D2 (diode 2) is reverse biased. to make the simulation of Au/Pd/n-GaN/Pd/Au device, ith the Shockley equation. For Schottky barrier diodes ence of an interface state and other effects, the cur- s on the bias voltage [14–30]. If the device’s equivalent series connected two diodes containing a parasitic thermionic emission (TE) theory predicts that the ltage characteristic becomes [14–19,31], V2 + VRs = (±1) {∣∣∣ 1 ˛1 �n ( I I01 + 1 )∣∣∣ + ∣∣∣ 1 ˛2 �n (−I I02 + 1 )∣∣∣+ ∣∣IRs∣∣} , (1) q n1kBT , ˛2 = qn2kBT and Rs is the parasitic resistance. Pos- ative sign (±) in Eq. (1) denotes forward and reverse bias is the total voltage across the device. Where, I is the cur- V2 are the voltage drop across diode 1 (denoted with de 2 (denoted with D2), q the electronic charge, kB the s constant, T the absolute temperature in Kelvin, n1 and e factors that describes departure from the ideal diode r reverse bias as well as forward bias. I01 and I02 are the ration currents are given by: 2e− q ˚Bo1,2 kBT , (2) he effective device contact area (7.5 mm2), A* the effec- son constant and is set to 26.9 A/cm2 K2 for n-type GaN the apparent barrier heights at zero-bias, which can be llowing equation [8,32]: T �n ( AA∗T2 I01,2 ) . (3) 1436 A.S. Kavasoglu et al. / Synthetic Metals 161 (2011) 1434– 1440 The using general combination properties of natural logarithm function allow us to rewrite: ˚B01,2 = kBT q { �n ( AA∗ ) + 2�n T } . (4) As can b can be writ ˚B01,2 = 2k q The first perature co of the first above can b mogeneitie from the th The later are simulat ˚B01(T) = 2 ˚B02(T) = ˚ The para mental stud parameters volt and vol to the unit b If the cu the local en reduction o ity factors a n1,2 = Eoo1, kBT where Eoo1 pendent of characterist Eoo2 = q·h¯2 √ constant, h the effectiv tric permitt semiconduc effective m Expected d can be calcu correspond ues into Eoo approximat ing energy That’s why as 8 × 1020 semiconduc tive density In another higher the d EC–EF. This the valence ues of NC2 a interface, w ductor, call (SBH) is the MS junctio tion of MSM junction is sion theory energy equ metal–semiconductor interface, contribute to the current transport [15,17]. The magnitude of the SBH represents the mismatch in the energy position of the majority carrier band edge of the n type GaN and the metal Fermi level across the MS interface. For lightly doped f the on (T rier. a 6.6 rent hott e of t avily (scr) ith n sem ]. Th heig ˚B in E kB T q NC1, ands = 8 × tance tion en c [34] prox cons (CD0) us e SPI tance (V1,2 (V1, e CD0 = A √ volt A (7 ndu ctric ial o , q th med .6 × en ta e Fer roo cove ing t de ei(ωt+ volt ill ca I01,2 e seen below equation, temperature dependency of BH ten as a sum of two terms: B T �n T + T [ kB q �n ( AA∗ I01,2 )] . (5) term generally does not change too much with the tem- mpare to second term, so the temperature dependency term can be ignored. Therefore the equation is given e simplified as ˚B01,2 = ˇ + �T. In case of lateral inho- s of the barrier height (BH), the value of ˇ and � deviate eoretical values. al inhomogeneities of the BH in the investigated device ed as below equations: .723 × 10−2 + (2.46 × 10−3 × T), (6) B01 − 0.13. (7) meters in Eq. (6) has been obtained from real experi- y [33] in the recent publication (using Ref. [33]). The ‘2.723 × 10−2’ and ‘2.46 × 10−3’ refer to the unit being t/K, respectively. In addition, the parameter ‘0.13′ refers eing volt. rrent transport is controlled by the TFE theory due to hancement of electric field, which can also yield a local f the barrier height, the relationship between the ideal- nd temperature can be expressed by [14–28] 2 coth ( Eoo1,2 kBT ) , (8) and Eoo2 are the characteristic tunneling energies inde- temperature. Eoo1 and Eoo2 are chosen as Eoo1 = 50 meV, ic energy Eoo2 (is set to 388 meV) is defined as [18] ND2/ ∈ m∗ where q is the electronic charge, h¯ is Planck’s , divided by 2�, ND2 is the donor concentration, m* is e mass of the tunneling electron, and ∈ is the dielec- ivity (9.6 ∈0 = 9.6 × 8.85 10−12 F/m) of the n type GaN tor [1,3,5]. It is generally accepted that the tunneling ass has a value of 0.19 × mo for n type GaN [1,3,5]. onor concentration value of back contact region (ND2) lated by plugging the Eoo2 (is set to 388 meV, this energy s approximately to FE) and m* (is set to 0.19 × mo) val- 2 = q·h¯2 √ ND2/ ∈ m∗ formulation. Then ND2 value yields ely as ND2 ≥ 2 × 1020 cm−3. The characteristic tunnel- (for pure field emission) depends on doping density. we have chosen the back contact doping density (ND2) cm−3. Heavily doped (ND2 = 8 × 1020 cm−3) n-type GaN tor’s backside automatically leads to increase effec- of states values (NC2) in the conduction band-edge. words NC2 values must be larger then ND2 values. The onor concentration, the smaller the energy difference leads to Fermi level will move closer to the top of band. Therefore we have chosen the numerical val- s a 2 × 1021 cm−3 (NC2 > ND2). The nature of the second hich is formed between the top contact and semicon- s as a rectifying junction. The Schottky barrier height rectifying barrier for electrical conduction across the n and, therefore, is of vital importance to the opera- device. The current across a metal–semiconductor mainly due to majority carriers. The thermionic emis- postulates that only energetic carriers, which have an al to or larger than the conduction band energy at the sides o emissi the bar ND1 as a diffe GaN Sc the on The he region tacts w to the [14–19 barrier Vbi1,2 = as seen Vn1,2 = where tion b as NC1 capaci simula has be model tion ap Model itance (m), th used in capaci CD1,2 =( 1− Her CD0 1,2 The Where semico ∈ diele potent (kBT/q) is assu ND1 = 6 has be [8], th around values Dur a MSM I(t) = I0 shift in bias w GaN (ND1) the current flows as a result of thermionic E) shown in Fig. 1 with electrons thermally excited over Therefore, we have chosen the numerical the values of × 1016 cm−3 which was already has been obtained in experimental investigation in the literature for n-type ky device [8]. During the simulation, we have assumed he contacts (D2) is heavily doped (ND2 = 8 × 1020 cm−3). doped side of n type GaN has narrow space-charge width w (w ∝ ND−1/2). For metal–semiconductor con- arrow scr widths, electrons can tunnel from the metal iconductor and from the semiconductor to the metal e built-in potential value for each diode is related to the ht by the relationship: 01,2 − Vn1,2, (9) qs. (9) and (10) �n ( NC1,2 ND1,2 ) , (10) C2 are the effective density of states in the conduc- for both diode structure. NC1 and NC2 are chosen 1019 cm−3, NC2 = 2 × 1021 cm−3. The depletion region of both sides has been taken as CD1 and CD2. During the process, the new depletion region capacitance model onsidered. This new model called as Paul Van Halen , which eliminates the singularity found in the deple- imation model and is applicable for any applied bias. iders only three input parameters: the zero bias capac- , the built-in potential and junction-grading coefficient liminating the arbitrary fitting parameters frequently CE simulation software. According to Paul Van Halen, of an investigated device structure is taken as: ) CD0 1,2 2/Vbi1,2) + ((4.7m + 0.7) VT/Vbi1,2)e((V1,2−Vbi1,2)/2VT ) )m . (11) 1,2 is given as: q ∈ ND1,2 2Vbi1,2 . (12) age drop across the diode 1 and diode 2 is given as V1,2. .5 mm2) is the junction area of the device, ND1,D2 the ctor donor concentration on both sides of the device, permittivity (n-type GaN), Vbi1,2 the device built-in n both sides of the device, VT the thermal voltage e electronic charge. The impurity profile of both sides as abrupt profile (m = 0.5) and ND1 was set to value as 1016 cm−3. It should be noted that the order of this value ken real experimental results it can be found elsewhere mi level is about 0.18 eV below the conduction band m temperature, thus the selected donor concentration rs ideal Schottky barrier band bending. he simulation applying voltage and the current through vice under test are defined as V(t) = V0ei(ωt+˚U) and ˚I). Where ω is the frequency, ˚U and ˚I are the phase age and current, respectively. Small modulations of AC use a change in barrier height, and the position of the A.S. Kavasoglu et al. / Synthetic Metals 161 (2011) 1434– 1440 1437 Fig. 2. Propos of the MSM de Fermi level the impeda Z = V(t) I(t) = The qua shift. eiϕ is r splits into t Z = Z ′ + iZ ′′ with Z′, Z′′ and reactan described v is that man vantage is often diffic that is very physical str We may of equivale ulation resu picture on with its cap This seems circuit, the cated equat Rs and CD1,D Z ′m = [ Rs + Z ′′m = − [ 1 + The dyn (CD) are m dependence RD 1,2(V1,2) The gene Y = 1 Z = R where Y is nary unit, a Siemens. Si Kramers–K tain utterly denc frequ g bo ctanc a1ω 1Rsω s(CD1 D1 + D1 + CD1C 2 D1R 2 D 2 D2CD a 1Rsω s wo rgy d ve (th rely lize E semi m ω , ma ic re ions. ity a pacit ed equivalent circuit model, used to represent the electrical properties vice. thus changing the characteristics of the device. Then nce measures [35,36]: V0ei(ωt+˚U) I0ei(ωt+˚I) = V0 I0 ei(˚U−˚I) = V0 I0 eiϕ = ∣∣Z∣∣ eiϕ. (13) ntity Z is the impedance, while ϕ stands for the phase ewritten as eiϕ = cos(ϕ) + i sin(ϕ), so that the impedance wo parts: = R + i X, (14) , R and X being real part, imaginary part, resistance ce, respectively [30,35,36]. Very often, the data can be ery well with equivalent circuits. The advantage of them y things can be described in this way, but the disad- that the physical meaning of the found parameters is ult to interpret. One example of an equivalent circuit illustrative and that still has a strong link with the actual ucture of the device is the following: interpret the impedance of the MSM devices in terms nt circuit model. In order to accurately obtain the sim- lt, the equivalent circuit of Fig. 2 is proposed. In the the left, the device is thought of as consisting of a D1 acitance and resistance placed in series with the D2. a very reasonable assumption. Even in such a simple simulated capacitance and conductance follow compli- ions. The Z′ and Z′′ can be expressed in terms of RD1,D2, 2 such as: RD1 1 + (ωRD1CD1)2 + RD2 1 + (ωRD2CD2)2 ] , (15) R2D1ωCD1 (ωRD1CD1) 2 + R 2 D2ωCD2 1 + (ωRD2CD2)2 ] . (16) amic resistance (R ) and depletion region capacitance depen signal siderin condu Gm = a where a1 = R a2 = (R a3 = R a4 = 2 a5 = R a6 = R Bm = a It i to ene negati to a pu genera metal Cm = B One dynam condit resistiv and ca D odulated by DC bias voltage. Therefore, the voltage of the RD is given by [15,37]: = 1 dI/dV1,2 = (n1,2kBT/q) I0e±q ∣∣V1,2∣∣/n1,2kBT . (17) ral equation defining admittance is given by: 1 + iX = Gm + i Bm, (18) the admittance, G is the conductance, i is the imagi- nd B is the susceptance and Y, G, B are measured in nce the real and imaginary part of Y are related by the roning relations [35,36], both G and B spectrums con- the resembling information. In order to investigate the test by app the imagina Impedan the parallel capacitance the C–V and frequency f tangent (ta equal to th [35,36]. If we cal tan ı = a1ω Fig. 3. Impedance analyzer equivalent circuit. e of imaginary admittance and real admittance on AC ency, we must use the more detailed calculation, con- th real and imaginary components. Therefore, simulated e (Gm) and susceptance (Bm) can be written as: 4 + a2ω2 + a3 4 + a4ω2 + a23 , (19) CD2RD1RD2) 2, (19a) Rs)R2D2C 2 D2 + (RD2 + Rs)R2D1C2D1, (19b) RD2 + Rs, (19c) D2R 2 D1R 2 D2 + R2D2C2D2(RD1 + Rs)2 + R2D1C2D1(RD2 + Rs)2, (19d) 2CD1CD2(CD1 + CD2), (19e) 2 + R2D1CD1, (19f) 5ω3 + a6ω 4 + a4ω2 + a23 . (20) rth noting that, since the conductance corresponds issipation in the sample, conductance must be non- e limiting case of a sample without losses corresponds imaginary admittance). At this point, it is convenient to q. (20) by introducing the simulated capacitance of the conductor metal device, Cm: (21) y propose that conductance determines the junction sistance (when Rs = 0) and can be varied by growth The series resistance depends on the bulk material nd on the contact resistances. Most impedance analyzer ance meters measure the capacitance of the diode under lying a constant amplitude AC voltage, and displaying ry component of the resulting AC current. ce analyzer surmises the device to be represented by circuits in Fig. 3, where Cm is the assumed as measured . For interface states, we expect a peaked response in G–V plots and flat C–ω and G–ω plots up to a certain rom where they will fall off rapidly. A maximum in loss- n ı = G/B) occurs when the radial frequency ω is almost e reciprocal relaxation time (�) of the interface states culate loss tangent it will yields as: 4 + a2ω2 + a3 a5ω3 + a6ω . (22) 1438 A.S. Kavasoglu et al. / Synthetic Metals 161 (2011) 1434– 1440 Fig. 4. The I–V characteristics of Au/Pd/n-GaN/Pd/Au device as a function of tem- perature. 3. Results Simulate at various t As can b logarithmic it departs si due to the tance, inter resistance i forward-bia is consisten [8,12,33,38 The tem simulated Fig. 5. Duri parameters ered param It is seen considered of capacitan of 10−11 Fa from low fo Fig. 5. Typica conventional a Fig. 6. Plots of room tempera the temper hape axim e ma bias tance ias v ndu ead is n ectio ,38,4 auss met oade ratur chie elect tions is h dep for 2 evic ent and discussions d I–V characteristic of the investigated device structure emperatures ranging 100–300 K are given in Fig. 4. e seen in this figure, I–V curves are linear on a semi- scale at intermediate forward bias voltage region, but gnificantly from linearity at high forward bias voltages effects of some simulated factors such as series resis- facial insulator layer, and interface states. The series s significant at high injection region (1 ≥ V ≥ 0.8) of the s I–V characteristics. It is seen that obtained I–V curves t with previously published paper in the literatures –42]. perature dependence of the capacitance at 2 MHz was capacitance at various temperatures and is seen in ng the simulation, process Eq. (21) is used and, some were taken into consideration. Theoretically, consid- eters are given in theoretical part. that obtained C–V is not consistent with traditionally capacitance curves [32]. It is seen in Fig. 5 that the values ce for each temperature are adequately low of the order rad. An exponential increase is seen in the capacitance rward bias towards the high forward bias region for spike s The m and th higher capaci ward b semico tance l If there the inj [35,36 have G tic sym gets br tempe To a of the simula voltage quency dc bias gated d depend l C–V plot of the Au/Pd/n-GaN/Pd/Au device structure simulated at c signal frequency and different probing temperature. contributio capacitance curves. The with increa of the RC t four distinc low frequen again rema sharply tow as saturatio tial barriers regions. Th This inflect from interfa [35,36]. Nyquist with real im are shown i structure. T well-displa C–f of investigated Au/Pd/n-GaN/Pd/Au device at various voltages for ture. ature range between 100 and 160 K. In addition, some peaks appear in the C–V curves for every temperature. um value of the capacitance varies with temperature ximum position of the capacitance shifts towards to voltages with increasing temperature. It is known that of the Schottky diodes increases with increasing for- oltage. Presence of interface layer between metal and ctor, interface states at interfacial layer and series resis- to deterioration in capacitance voltage characteristics. o interface layer, excess capacitance may be caused by n of the minority carriers in to the bulk semiconductor 3,44]. In addition, the peaks in the capacitance curves ian nature. The Gaussian peak looks like a characteris- ric “bell curve” shape that width of the “bell” quickly r and intensity of the “bell” decreases with increasing e. ve a better understanding of the interfacial mechanisms rical characteristics [15,17,30,35,36] of the device C–f were performed. For C–f simulations, the applied dc eld constant and the ac frequency is varied. The fre- endence of the device capacitances at different forward 95 K is shown in Fig. 6. The capacitance of the investi- e increases with decreasing frequency due to the time response of two barrier height or interface states. The n of the double diode or interface states to the device is clearly reflected in the humps of the capacitance se steps in the capacitance shift to higher frequencies sing bias voltage according to the voltage dependence ime constant of two junctions. In Fig. 6, curves have t regions. For example, capacitance remains constant at cies, decreases sharply at moderate frequency values, ins constant at high frequencies and again decreases ards to 13 MHz. The first and third regions are called n regions, which refers potential barriers (two poten- ). The second and fourth regions are called as inflection e inflection region is also called dispersion region [35]. ion in C versus f plot is a non-trivial feature arising ce state that could be possibly detected in experiments plots of the variation of imaginary impedance Im(Z) pedance Re(Z) under different biasing voltages at 295 K n Fig. 7 for the investigated Au/Pd/n-GaN/Pd/Au device he Nyquist plots consist of two parts. First part is the yed semicircle. The shape of the semicircle in the sec- A.S. Kavasoglu et al. / Synthetic Metals 161 (2011) 1434– 1440 1439 Fig. 7. Z′ versus Z′′ plots of the AC impedances of a simulated Au/Pd/n-GaN/Pd/Au device at the different DC forward bias value and constant temperature. ond part is getting well displayed with decreasing bias voltage. With increasing bias, the left-hand semicircle radius (low frequen- cies) becomes smaller. The double barrier in the device (or double RC-model) leads to two semicircles that are characterized by two- time constant [35]. The equivalent ac-circuit model is given before in Fig. 2. Cu RD1, RD2 an two-time co model and d inates othe the two-tim it difficult t lated at 0.2 apparently hand side s lower bias spectrum ( The second tiveness on and leads to in Fig. 7. A element co Fig. 8. Variati Au/Pd/n-GaN/ [15–17,29,30,35,36,38,43,44], but as well other equivalent circuit models could be responsible for the tendency of the semicircle to flatten out at low frequencies. The variation of the inverse loss tangent (tan ı)-1 with respect to frequency o voltages is s curves disp frequencies different dc yield a sing bias level. T designed as the dc bias occurred m visible. The (tan ı)-1 fre ac conducti 4. Conclus The curr device hav 100–300 K. cated diode The curren behavior a tance were in t quen show emp ward l tran ould dep s for unct ecrea rrier de o ed in riatio unde rrier disp rface here rves consist of two RC sub-circuits (parallel resistors: d capacitors: CD1, CD2) in series with a resistor Rs. The nstants impedance spectrum stems from the double RC isplays the ratio of the two time constants. If, one dom- r, only one semicircle will be visible. In the worst case, e constants lead to a distorted semicircle, which makes o separate the sub-circuits. The spectra in Fig. 7 simu- –1 V are all characterized by two semicircular shapes composed by two time constants. The radius of the left- emicircle increases with decreasing forward bias. For voltages, a distortion at the low frequency side of the right end of the spectrum) starts to be disappearing. barrier in the Au/Pd/n-GaN/Pd/Au device lifts effec- the shape of the spectrum at lower forward biases two-time constant impedance spectrum as displayed s described in different publications, the second RC- uld be due to the grain boundary or an interface state capaci device circuit ing fre device ation t the for menta and sh quency dc bia tance f with d two ba ble dio reflect the va Re(Z) two ba curves or inte ture, w on of (tan �)-1 for various applied dc bias voltage of investigated Pd/Au device structure at room temperature. Bode plot o double barr References [1] R.F. Davis Characte [2] Hung-We Materials [3] H. Morko Optical a [4] M. Senth [5] M.K. Cha Transisto lishing, 2 [6] B. Deb, A. and Phys [7] V. Rajago [8] B. Akkal, Physics 8 [9] A. De Vos Universit f Au/Pd/n-GaN/Pd/Au device structure at different bias hown in Fig. 8. As shown in Fig. 8, the (tan ı)-1 frequency lay two peaks values but remain constant (zero) at high . Intensities of the peaks in the (tan ı)-1 curves vary at bias levels. Inverse loss tangent versus frequency plots le peak at about 20 kHz as depicted in Fig. 8 for 0.2 V dc hus, in this voltage level the equivalent circuit can be a single RC network in series with the resistor (Rs). As level is increased, the frequency at which (tan ı)-1max oved to higher frequency values and second peak gets se suggest that second barrier effect is displayed in the quency curves of the Au/Pd/n-GaN/Pd/Au device for the on especially at higher forward bias level (V > 0.2). ion ent–voltage characteristics of the Au/Pd/n-GaN/Pd/Au e been simulated in the temperature range of The current–voltage (I–V) characteristics of fabri- displayed a strongly temperature-dependent behavior. t–voltage characteristics showed good rectification t all temperatures. The forward and reverse bias –voltage characteristics of the Au/Pd/n-GaN/Pd/Au simulated with the assistance of suggested equivalent he temperature range of 100–300 K at 2 MHz prob- cy. The C–V characteristics of the Au/Pd/n-GaN/Pd/Au that capacitance is quite sensitive to the device oper- erature. The peaks in the C–V curves are observed at dc bias values and probably associated to the funda- sition of electrons in the two interfaces. Similar peaks ers are observed by other researchers [43,44]. The fre- endence of the device capacitance at different forward 295 K has simulated with the help of derived admit- ion. The capacitance of the investigated device increases sing frequency due to the time dependent response of height or interface states. The contribution of the dou- r interface states to the device capacitance is clearly the humps of the capacitance curves. Nyquist plots of n of imaginary impedance Im(Z) with real impedance r different biasing voltages at 295 K are shown that height can be clearly observed. Also, (tan ı)-1frequency lays two peaks which represent two potential barrier s. Our simulations reveal that for MSM device struc- back and front contacts may induce two maxima at r two-semi circles at Nyquist plot is a good indicator of ier nature of the device. , M.S. Shur, GaN-based Materials and Devices: Growth, Fabrication, rization & Performance, World Scientific Publishing Company, 2004. n Huang, C.C. Kao, J.T. Chu, W.D. Liang, H.C. Kuo, S.C. Wang, C.C. Yu, Chemistry and Physics 99 (2006) 414. c, Handbook of Nitride Semiconductors and Devices, vol. 3: GaN-based nd Electronic Devices, Wiley-VCH, 2008. il Kumar, J. Kumar, Materials Chemistry and Physics 77 (2002) 341. ttopadhyay, Device modeling of AlGaN/GaN High Electron Mobility rs (HEMTs): An Analytical Approach, LAP LAMBERT Academic Pub- 010. Ganguly, S. Chaudhuri, B.R. Chakraborti, A.K. Pal, Materials Chemistry ics 74 (2002) 282. pal Reddy, Materials Chemistry and Physics 93 (2005) 286. Z. Benamara, H. Abid, A. Talbi, B. Gruzza, Materials Chemistry and 5 (2004) 27. , Endoreversible Thermodynamics of Solar Energy Conversion, Oxford y Press, Oxford, 1992. 1440 A.S. Kavasoglu et al. / Synthetic Metals 161 (2011) 1434– 1440 [10] Z. Hassan, Y.C. Lee, F.K. Yam, M.J. Abdullah, K. Ibrahim, M.E. Kordesch, Materials Chemistry and Physics 84 (2004) 369. [11] Y.Z. Chiou, Semiconductor Science and Technology 23 (2008) 125007. [12] K. C¸ ınar, N. Yıldırım, C. Cos¸ kun, A. Turut, Journal of Applied Physics 106 (2009) 073717. [13] Y.-Z. Chiou, J.-R. Chiou, Y.-K. Su, S.-J. Chang, B.-R. Huang, C.-S. Chang, Y.-C. Lin, Materials Chemistry and Physics 80 (2003) 201. [14] E.H. Rhoderick, Journal of Physics D: Applied Physics 3 (1970) 1153. [15] S. Sze, Physics of Semiconductor Devices, second ed., Wiley, New York, 1981. [16] E.H. Rhoderick, R.H. Williams, Metal-Semiconductor Contacts, second ed., Clarendon Press, Oxford, 1988. [17] D.K. Schroder, Semiconductor Material and Device Characterization, Wiley, New York, 1990. [18] F.A. Padovani, Semiconductors and Semimetals 7 (1971) 75. [19] R.T. Tung, Physical Review B 45 (1992) 13509. [20] A. Kassis, M. Saad, Solar Energy Materials and Solar Cells 94 (2010) 2108. [21] Zs.J. Horváth, L. Dózsa, O.H. Krafcsik, T. Mohácsy, Gy. Vida, Applied Surface Science 234 (2004) 67. [22] M. Pattabi, S. Krishnan, G. Sanjeev, Solar Energy Materials and Solar Cells 91 (2007) 1521. [23] S. Chand, J. Kumar, Semiconductor Science and Technology 10 (1995) 1680. [24] Sandeep Kumar, Y.S. Katharria, V. Baranwal, Y. Batra, D. Kanjilal, Applied Surface Science 254 (2008) 3277. [25] Y.P. Song, R.L. Van Meirhaeghe, W.H. Laflère, F. Cardo, Solid-State Electronics 29 (1986) 633. [26] H. Werner Jürgen, H. Güttler Herbert, Journal of Applied Physics 69 (1991) 1522. [27] P. Chattopadhyay, A.N. Daw, Solid-State Electronics 29 (1986) 555. [28] P. Cova, A. Singh, Solid-State Electronics 33 (1990) 11. [29] N.A. Poklonski, N.I. Gorbachuk, S.V. Shpakovski, V.A. Filipenia, S.B. Lastovskii, V.A. Skuratov, A. Wieck, V.P. Markevich, Microelectronics Reliability 50 (2010) 813. [30] T.B. Salan, M. Amairi, Z. Sassi, H. Morel, Sensor Letters 7 (2009) 707. [31] Nese Kavasoglu, A. Sertap Kavasoglu, Sener Oktik, Current Applied Physics 9 (2009) 833. [32] M. Mamor, Journal of Physics Condensed Matter 21 (2009) 335802. [33] M. Ravinandan, P. Koteswara Rao, V. Rajagopal Reddy, Semiconductor Science and Technology 24 (2009) 035004. [34] P. Van Halen, M.H. Habib, Proceedings of the Tenth Biennial Univer- sity/Government/Industry Microelectronics Symposium, Research Triangle Park, NC, USA, 18 May 1993 through 19 May 1993, Code 19109. [35] A.K. Jonscher, Dielectric Relaxation in Solids, Chelsea Dielectrics Press, London, 1983. [36] E.H. Nicollian, A. Goetzberger, Applied Physics Letters 10 (1967) 60. [37] A. Sertap Kavasoglu, Nese Kavasoglu, Sener Oktik, Solid-State Electronics 52 (2008) 990. [38] Nezir Yıldırım, Kadir Ejderha, Abdulmecit Turut, Journal of Applied Physics 108 (2010) 114506. [39] Z. Tekeli, S¸ . Altındal, M. C¸ akmak, S. Özc¸ elik, D. C¸ alıs¸ kan, E. Özbay, Journal of Applied Physics 102 (2007) 054510. [40] S. Dogan, S. Duman, B. Gurbulak, S. Tuzemen, H. Morkoc, Physica E 41 (2009) 646. [41] S. Demirezen, S. Altındal, Current Applied Physics 10 (2010) 1188. [42] Mehmet A. Ebeoglu, Physica B 403 (2008) 61. [43] S. Demirezen, S. Altındal, Physica B 405 (2010) 1130. [44] Z. Tekeli, S. Altındal, M. C¸ akmak, S. Özc¸ elik, E. Özbay, Microelectronic Engineer- ing 85 (2008) 2316. Electrical characterization of Au/Pd/n-GaN/Pd/Au device structure in the radio frequency range by simulation study 1 Introduction 2 Theory 3 Results and discussions 4 Conclusion References