zyx zyxwvutsr IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 42, NO. I , FEBRUARY 1993 39 Capacitance-Voltage Measurements on Schottky Diodes with Poor Ohmic Contacts Klaus Steiner Abstract-The evaluation of Schottky-diode capacitances or reliable free-carrier concentration profiles in doped semiconductors by admittance measurements need a time- and moneyconsuming diode structuring process since the rectifying and ohmic contacts are produced separately. The formation of both contacts simultaneously during the same process step reduces costs. However, since alloying cannot be done, the diodes exhibit a poor ohmic contact. Consequently, minority carrier influence, series resistance effects, and deep level influence lead to frequency-dependent admittances. In this work, frequencydependent admittance analysis on such diodes and a simple small-signal equivalent circuit model are used to evaluate spacecharge capacitances reflecting only the free carriers of the doped material. This method is useful for the automatic routine control of semiconductors. The minority carrier, deep level, and series resistance influence on the diode admittance is reviewed. I. INTRODUCTION APACITANCE-VOLTAGE (CV) measurements are one of the most popular electrical measurement techniques used to evaluate important device characteristics. There are several reasons for the popularity of this technique. The method is nondestructive,.and it provides material parameters of great interest such as impurity concentration profiles [ 11, [2], information on interface, bulk or surface states [3]-[6], and heterojunction properties [7], [8]. The technique can be used on fabricated devices with commercially available capacitance meters and is adaptable for automation. Normally, measurements are carried out using an impedance analyzer. A constant ac voltage is applied to the Schottky diode or the p-n junction. The capacitance is evaluated from the imaginary part of the resulting admittance, whereas the real component gives the conductance. This simply that the diode is described with a small-signal equivalent circuit model consisting of a capacitor parallel to a resistor. The capacitor stands for the space-charge capacitance, while the resistor represents the residual conductance of the diode. Deep trap, minority carrier, and series resistance influences are neglected. To obtain a depth profile, dc voltages are successively superimposed. C CV depth profiling is normally carried out on diodes or FET (field-effect transistor) gates since structuring free CV analysis using electrochemical or mercury contacts is not nondestructive [9]-[l l]. Diodes or FET's for testing are simultaneously structured when semiconductor devices are manufactured. However, in cases where the technique is used for controlling material properties onlyfor instance, measuring implanted profiles-a time- and money-consuming diode structuring process is necessary. At the least, a rectifying and an ohmic contact have to be formed. Moreover, the device under test has to be insulated from diodes located nearby. The formation of both the rectifying and ohmic contacts simultaneously during the same process step reduces costs. However, since nonalloyed metallizations are used, the diodes exhibit poor ohmic contacts. Consequently, minority carrier influence, series resistance effects, and deep level influence lead to frequency-dependent admittances [2]-[6], [ 121-[ 191. In this paper, frequency-dependent admittance analysis on Al-GaAs Schottky diodes with nonideal ohmic contacts is carried out. A simple small-signal equivalent circuit model is used to evaluate space-charge capacitances, reflecting only the free camers of the doped material. This new technique is highly applicable for the automatic routine control of semiconductors or devices. zyxwvutsrqp zyxwvutsrqp Manuscript received February 1 1 , 1992; revised June 25, 1992. The author is with the Frauhofer-Institut fur Physikalische MeBtechnik, W-7800Freiburg i. Br., Germany. IEEE Log Number 9203666. 11. SAMPLEPREPARATION The schematic cross section and a top view of the investigated device are illustrated in Fig. 1 . The structuring consists of one photolithography step and an A1 deposition followed by a liftoff. The active region below the A1 contacts is Si-implanted into a semi-insulating (s.i.) GaAs substrate. The Si implants are accelerated at 50 keV to a total dose of 4 . lo'* cmP2. The diode diameter is 0.5 mm. Both the rectifying and ohmic contacts are structured simultaneously during the same process step. The A1 metallization remains nonalloyed. The difference between both 'Ontact' is the electrically active area* The area Of the ohmic contacts is much larger than the Schottky-contact area. In this paper, the diode admittance has been evaluated using a Hewlett-Packard HP(4275) LCR meter. The precision of the measurement has been checked using test capacitors. The ac voltage was always smaller than kT, and remained constant throughout the investigations. zyxwvuts 0018-9456/93$03.00 0 1993 IEEE T 40 zyxwvut zyxwv zyxwvutsrqponmlkjihgfed zyxwvutsrqponmlk zyxwvut zyx IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 42, NO. 1, FEBRUARY 1993 AI - contacts \\A space chargeregion -t-- s.i. GaAs Krq -=I+ R b S L C c Fig. 2. Small-signal equivalent circuit models of Schottky diodes. AI - contacts and the measured conductance G ' zyxwvutsrqpo Fig. 1 . Schematic cross section and top view of the investigated diode. 111. FREQUENCY-DEPENDENT SCHOTTKY-DIODE ADMITTANCE Unfortunately, CV measurements on semiconductor films are not free from limitations. Neglecting phase shift by cables, gate drain and source drain stray capacitances, or peripheral space-charge capacitances outside the active region [ 1 11, [ 181, [20], [2 13, the avalanche or the Zener mechanism degrades CY data evaluated on highly doped semiconductor films [ 11, [2]. In less doped material, compensated material, or near pinchoff, the free-carrier concentration can become comparable to the deep level concentration. In this case, charged and recharged deep levels contribute considerably to the measured capacitance [3], 1191, ~ 2 1 . When the measuring frequency of the applied voltage becomes comparable to the inverse relaxation time of a deep level, the gate capacitance is frequency dependent and is lowered with increasing frequency. At much higher or lower frequencies, the gate capacitance is frequency independent and remains constant [3], [ 191. In the above discussion, it was assumed that the dc bias voltage adjustment does not influence equilibrium conditions. However, in practical measurement conditions, very slow deep levels cannot reach steady-state conditions in one measurement cycle. If this period is comparable to the relaxation time of such slow states, the capacitance becomes dependent on the dc voltage adjustment time. This effect can be used to detect slow states [3], [19]. In semiconductor film profiling, series resistance effects can cause serious errors [ 151-[ 181; in particular, series resistance effects near pinchoff must be considered in FET-like structures with lateral contact configurations. The measured capacitances and conductances become frequency dependent. Wiley and Miller [15] have shown that such effects on the diode admittance can be described by a simple small-signal equivalent circuit model consisting of a capacitor C in series with a resistor Rs (cf. Fig. 2). Thus, the measured capacitance C ' obeys [w, C' = C 1 The equations show that series resistance effects do not freeze out with increasing frequency. This is in contrast to deep level effects. Series resistance effects are intensified with increasing frequency; i.e., the measured capacitance is decreased while the conductance is increased. Thus, suitable frequency-dependent admittance measurements can be used to distinguish between series and deep level effects or to calculate effective series resistances in FET-like structures [16]-[18]. Forward-biased Schottky diodes can exhibit inductive reactances or excess capacitances [ 11 , [2], [4], [61, [121, [ 131. For both, the diode admittance is strongly frequency dependent. The physical reason for these effects might be interface states or minority carrier injection at the Schottky contact. At forward biases, interface states can be charged or recharged since the Fermi level crosses their energetic level in the forbidden gap. Charged interface states create a dipole layer with the metallizatiqn of the Schottky contact. This dipole modifies the Schottky barrier height. Consequently, the charging and recharging of such trap levels during a measurement cycle periodically change the Schottky barrier height. Finally, the modified measurement current gives a capacitive contribution to the diode admittance [2], [4]-[6]. However, there are no contributions when the charging of the trap levels is dominated by a tunneling current independent of the measurement frequency. Stored minority carriers in the neutral region of the diode outside the space-charge region lead to both diffusion capacitances and inductive contributions to the Schottky-diode admittance. The minority carriers can be injected at forward-biased metal semiconductor contacts [l], [2], [12], [13]. The minority carriers are stored in the neutral region of the diode and change the neutrality condition. This effect is called conductivity modulation. Since the nonalloyed poor ohmic contacts lead to a high series resistance, the total current density will be limited. Thus, the hole storage time, and finally the Schottky-diode recovery time, will be extended. Therefore, minority carrier zyxwvutsr + (wRsC)2 (') STEINER: CV MEASUREMENTS ON SCHOTTKY DIODES zyx zyxwvutsrqpo 41 storage effects might be observable in this paper. The diffusion capacitances are freezing out with an increase of the frequency since the minority carrier current is diffusion limited. With an increase of the frequency, the minority carrier current becomes out of phase to the applied ac voltage. Since the current follows the voltage, the contributions become inductive. These contributions can dominate the whole diode admittance. Finally, this gives an inductive diode reactance. Consequently, the measured “capacitance” is negative. In practical measurements, the space-charge capacitance, the diffusion capacitance, the deep level effects, the series resistance influence, and the inductive components of the minority carrier current are superimposing [6]. It might be partly possible to distinguish these effects using temperature- or frequency-dependent measurements [ 121, [ 131, [ 181. However, it is possible that the admittance is additionally distorted by p-buffer layers or ohmic contacts which can influence the minority carrier injection [2], [12], [13]. Moreover, the deep level influence is dependent on the technology [6], [12]. Therefore, each sample has to be discussed in detail separately. Ideal ohmic contacts are assumed to exhibit IV characteristics with a low contact resistance relative to the bulk4 material. In particular, no contributions to the diode reactance are expected. In our case here, nonalloyed ohmic contacts have been used. Thus, frequency-dependent contributions to the diode admittance can be expected [12][ 141. Since the ‘‘ohmic contacts” are forward-biased minority carrier injection and interface traps have to be taken into account. Furthermore, the nonideal ohmic contacts will lead to a high diode series resistance. An inductance and a series resistance in a small-signal equivalent circuit model represent both the minority carrier and series resistance influence on the diode admittance. The devices are in series to the space-charge capacitance of the Schottky contact. A diffusion capacitance does not need to be recognized since it is effectively in series to the much smaller space-charge capacitance. Moreover, the residual space-charge capacitance of the forward-biased ohmic contact does not need to be taken into accodnt since the diffusion capacitance dominates the total capacitance of the nonideal ohmic contact. Finally, the admittance of a Schottky diode with a poor ohmic contact might be describable with a three-component small-signal equivalent circuit model consisting of an inductance L, a capacitance C , and a series resistance Rs. It is illustrated in Fig. 2. If the impedance meter is interpreting the measured admittance as a resistor in parallel to the space-charge capacitance as mentioned above, the measured capacitance obeys Fig. 3 shows the measured capacitance and conductance values versus the dc bias voltage of the device of Fig. 1 . Both sets of curves exhibit frequency-dependent effects. The measured capacitances decrease, while the conductance values increase with increasing the frequency. At 10 MHz, the diode reactance becomes inductive in a certain voltage range. Both the capacitance and conductance behaviors were simulated with formulas (3) and (4). The results are plotted as dashed lines in Fig. 3. The data input for the simulation is illustrated in Fig. 4. The capacitance, inductance, and series resistance values are plotted against the dc bias voltage. Between 200 kHz and 4 MHz, the admittance behavior can be described as quite excellent. The simple small-signal equivalent circuit model is able to describe the frequency-dependent admittance behavior in this frequency range. At 10 MHz, the data fit is becoming poor. This might be due to the fact that the minority carrier influence is now dominant over the series resistance influence. However, the tendency towards negative “capacitances” can be seen. Fig. 4 shows nearly constant inductances at higher diode voltages, reflecting constant conductivity modulation by minority carriers. Finally, this leads to the assumption that the minority carrier charge in the neutral bulk region is cohstant. By decreasing the bias voltage, the capacitance and the inductance decrease. The reduction of the inductance might be due to the reduction of the stored minority carriers at the edge of the space-charge region near pinchoff. Since near pinchoff the minorities see the band bending upwards, they drift towards the substrate. Thus, by lowering the diode reverse voltage, the stored minority carrier charge at the edge of the spacecharge region decreases. Finally, the inductive contribution to the diode admittance becomes smaller. Only stored minority carriers at the periphery of the space-charge region far away from the substrate contribute to the diode admittance. Since the peripheral space-charge region area is small compared to the total space-charge region area, the inductance becomes negligible at lower voltages. As expected, the diode series resistance increases, and the space-charge capacitance decreases rapidly near pinchoff [ 161-[ 181. However, the inductance and capacitance decrease, and the series resistance increase is mainly due to the lateral contact configuration. This is not the case for a conducting substrate with reverse side ohmic contacts. Fig. 5 shows two CV-carrier concentration profiles. The upper curve has been calculated using the standard CV technique [ l l ] and the capacitance data of Fig. 4. The lower profile has been deduced using a Bio-Rad electrochemical CV-profile plotter [SI, [ 101. Both measurements have been carried out on wafers prepared under the same implantation conditions. At higher carrier concentrations, both measurements lead to nearly the same carrier concentration profiles. At lower carrier concentrations, some discrepancies can be seen. However, since these differences are quite small, the comparison of both methods verifies the space-charge capacitance calculations proposed in this paper. zyxwvuts C‘ = zyxwvuts zyxwv zyxw zyxw C(1 - w2LC) (1 - o ~ L C ) ~( w R ~ C ) ~ + (3) 02RsC2 (1 - w ~ L C ) ~ (wR~C)~‘ (4) and the measured conductance G’ = T + 42 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQ zyxwvutsrqponmlkjihg zyxwv I I I IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 42, NO. I , FEBRUARY 1993 zyxwvutsrqpo 300K 300 0 I I I -2 I -1 1 0 L zyxwvutsrqponm 2 1 0 Voltage V (V) Voltage V (V) Fig. 4. Calculated space-charge capacitance C, series resistance R,, and inductive component L of the diode in Fig. 1 above the applied voltage. (a) 4 calculation this work 16d . E ! zyxwvu zyxwvuts 15- I. CV-profile plotter I. 3 c 14- 0,O 0,l 0,2 0,3 Depth 0,4 0.5 0,6 0,7 x (Irm) Fig. 5 . CV-camer concentration profiles. Upper curve: calculation using standard CV technique [ll] and capacitance calculation of Fig. 4, lower curve: profile evaluated using an electrochemical CV-profile plotter. 1 -1 -2 0 Voltage V (VI 0) Fig. 3. Measured and simulated frequency-dependent admittances of the diode in Fig. 1 above the applied voltage, measurement (-), simulation (----). (a) Capacitance. (b) Conductance. zyxw It should be pointed out that a simple small-signal equivalent circuit model was used to describe the observed frequency-dependent admittance. For automatic routine control, only two frequency-dependent measurements must be made to evaluate the three unknowns, i.e., the space-charge capacitance C, the inductive contributions L, and the series resistance Rs. Furthermore, the physical meaning of the unknowns is clear. The simple structuring and evaluation method drastically reduces the testing time and costs. This underlines the usage of the developed method. STEINER: CV MEASUREMENTS ON SCHOTTKY DIODES IV. CONCLUSION The frequency-dependent admittance of Al-GaAs Schottky diodes with nonideal ohmic contacts has been discussed using a simple three-component small-signal equivalent circuit model. Both the “ohmic” and the rectifying contacts of this diode are simultaneously produced during the same process step. This simplifies the formation procedure and saves testing time and costs. However, the nonalloyed ohmic contacts lead to a high series resistance, minority camer injection, and deep level influence. All of these effects give a frequency-dependent diode admittance. Frequency -dependent admittance analysis in a certain frequency range using the three-component equivalent circuit model leads to the space-charge capacitance of the diode reflecting only the free majority camers. The method is highly suitable for the automatic routine control of semiconductor material properties, diode, or gate capacitances. ACKNOWLEDGMENT The author would like to thank Prof. Dr. E. Wagner for encouragement, and B. Halford for refining his English. 43 zyx tion of interface preparation conditions,” Appl. Phys. Lett., vol. 58, pp. 155-157, Jan. 1991. [7] H. Kroemer, W.-Y. Chien, J. S. Hams, J . Edwall, and D. D. Edwall, “Measurement of isotype heterojunction barriers by C-V profiling,” Appl. Phys. Lett., vol. 36, pp. 295-297, Feb. 1980. [8] E. H. Rhoderick, “Interpretation of CIV characteristics for heterojunctions and high-low junctions,” Electron. Lett., vol. 20, pp. 868869, Oct. 1984. [9] T. Ambridge and M. M. Faktor, “An automatic carrier concentration profile plotter using an electrochemical technique,” J. Appl. Electrochem., vol. 5 , pp. 319-328, Apr. 1975. [lo] P. Blood, “CV-profiling using electrolyte barriers,” Semiconductor Sci. Technol., vol. 1, pp. 7-27, 1986. [ l l ] D. K. Schroder, Semiconductor Material and Device Characterization. New York: Wiley, 1990. [12] J. Wemer, A. F. J . Levi, P. T. Tung, M. Anzlowar, and M. Pinto, “Origin of excess capacitances at intimate Schottky contacts,” Phys. Rev. Lett., vol. 60, pp. 53-56, 1988. [13] K. Steiner, N. Uchitomi, and N. Toyoda, “Inductive reactances and excess capacitances at WN,/n-GaAs Schottky contacts,” J. Vac. Sei. Technol. B , vol. 8, pp. 1113-1116, Sept./Nov. 1990. [14] S. M. Sze, D. J. Coleman, J. R. Loya, and A. Loya, “Current transport in metal-semiconductor-metal (MSM) structures,” Solid-State Electron., vol. 14, pp. 1209-1218, 1971. [15] J. D. Wiley and G. L. Miller, “Series resistance effects in semiconductor CV profiling,” IEEE Trans. Electron Devices, vol. ED-22, pp. 265-272, May-1975. J . D. Wiley, “C-Vprofiling of GaAs FET films,” IEEE Trans. Electron Devices, vol. ED-25, pp. 1317-1324, Nov. 1978. J. Engemann, “A. C. admittance studies of Schottky diodes using a vector-analyzer-system,” Solid-State Electron., vol. 24, pp. 467-474, 1981. K. Steiner, N. Uchitomi, and N. Toyoda, “Mobility profiles in submicron WN,-BPLDD-GaAs MESFETs,” Japan J. Appl. Phys., vol. 29, pp. 489-494, Mar. 1990. J.-J. Shiau, A. L. Fahrenbruch, and R. H. Bube, “Interpretation of capacitance versus voltage measurements in the presence of a high density of deep levels,” J. Appl. Phys., vol. 59, no. 8, pp. 28792884, 1986. W. W. Lin and P. C. Chan, “On the measurement of parasitic capacitances of device with more than two external terminals using an LCR meter,” IEEE Trans. Electron Devices, vol. 38, pp. 2573-2575, Nov. 1991. K. Steiner, H. Mikami, N. Uchitomi, and N. Toyoda, “Mobility profiles in short and narrow GaAs MESFET channels,” IEEE Trans. Electron Devices, vol. 38, pp. 23-27, Jan. 1991. J. M. Golio, R. J. Trew, G.N. Maracas, and H. Lefevre, “A modeling technique for characterizing ion-implanted material using C-V and DLTS data,” Solid-State Electron., vol. 27, pp. 367-373, Apr. 1984. zyxwvutsrqponm zyxwvutsrq REFERENCES S. M. Sze, Physics of Semiconductor Devices, (2nd ed.). New York: Wiley, 1981. E. H. Rhoderick and R. W. Williams, Metal Semiconductor Contacts (2nd ed.). Oxford: Clarendon, 1988. J.-J. Shiau and R. H.Bube, “General considerations for interpreting junction capacitance in complex systems,” Solid-State Electron., vol. 29, pp. 1153-1160, NOV.1986. P. Muret, “The influence of interface states upon the admittance of metal semiconductor diodes,” Semiconductor Sei. Technol., vol. 3, pp. 321-338, 1988. J. Wemer, K. Ploog, and H.J. Queisser, “Interface states at Schottky contacts: A new admittance technique,” Phys. Rev. Lett., vol. 57, pp. 1080-1082, 1986. P. Muret, D. Elguennouni, M. Missous, and E. H.Rhoderick, “Admittance of Al/GaAs Schottky contacts under forward bias as a func-