0 оценок0% нашли этот документ полезным (0 голосов)
46 просмотров3 страницы
A new static technique for precise measurements of charge, surface potential, low-frequency capacitance, interface charge, and interface state density in MOS structures is presented. A comparison with the conductance method is made. For a -oriented n -type silicon wafer, the distribution of the interface state density is shown over an energy range of 0. EV within the energy gap.
A new static technique for precise measurements of charge, surface potential, low-frequency capacitance, interface charge, and interface state density in MOS structures is presented. A comparison with the conductance method is made. For a -oriented n -type silicon wafer, the distribution of the interface state density is shown over an energy range of 0. EV within the energy gap.
Авторское право:
Attribution Non-Commercial (BY-NC)
Доступные форматы
Скачайте в формате PDF, TXT или читайте онлайн в Scribd
A new static technique for precise measurements of charge, surface potential, low-frequency capacitance, interface charge, and interface state density in MOS structures is presented. A comparison with the conductance method is made. For a -oriented n -type silicon wafer, the distribution of the interface state density is shown over an energy range of 0. EV within the energy gap.
Авторское право:
Attribution Non-Commercial (BY-NC)
Доступные форматы
Скачайте в формате PDF, TXT или читайте онлайн в Scribd
Static technique for precise measurements of surface potential
and interface state density in MOS structures
K. Ziegler and E. Klausmann Institut fuer Angewandte Festkoerperphysik der Fraunhofergesellschaft, 78 Freiburg, Eckerstrasse 4, West Germany (Received 16 December 1974) A new static technique for a very accurate measurement of charge, surface potential, low-frequency capacitance, interface charge, and interface state density in MOS structures is presented. A comparison with the conductance method is made. For a <111>-oriented n -type silicon wafer, oxidized in dry oxygen and annealed in hydrogen, the distribution of the interface state density is shown over an energy range of 0.9 eV within the energy gap. PACS numbers: 73.20., 73.40.Q A well-known method for the determination of the low- frequency capacitance, the surface potential, and the interface state density of an MOS structure is the volt- age slow-ramp technique of Castagne, 1 Kerr, 2 and Kuhn. 3 A charge slow-ramp technique for the determi- nation of the MOS gate charge, the surface potential, and the interface charge has been reported by Koomen. 4 Both methods assume quasistatic conditions. In this letter, we present a measurement technique which approximates the ideal case of a static measurement to a much higher degree. Because this new method is a pOint-by-point technique, it is very appropriate for use with digital instrumentation or data -collection systems. In addition, a higher accuracy in the determination of the MOS parameters, and in the energy distribution of the interface state density over a larger part of the energy gap is attainable. In the last part of this letter, we compare the results of this method with those ob- tained by the conductance techniqueS for a silicon n- type sample. The basic circuit for the static technique is shown in Fig. 1. A voltage-independent capacitor Co is connect- ed in series with the MOS capacitor. The magnitude of Co is chosen to be comparable to that of the oxide capa- citance COX' In parallel with the MOS capacitor, a cir- cuit capacitance C c is indicated. C c is the capacitance of the sample holder and the connecting wires. A direct voltage U B is applied to the series combination of Co and C c in parallel with the MOS structure. U B is mea- sured with a digital voltmeter. After a steady state is reached in the MOS structure the voltage drop U o across Co is determined from the output of a high -input -im- pedance operational amplifier used as a voltage follower. The parameters of the MOS structure are obtained as follows: The voltage V across the MOS capacitor is V=-(UB-U o )' The charge Q m which flows onto the gate electrode is Q m = - corro - cy. (1) By varying the voltage U B the charge Qm can be obtained as a function of the voltage V. Differentiating Q m with respect to V yields the low-frequency MOS capacitance C _dQm LF - dV . Except for an additive constant, 1/Iso' the surface poten- 400 Applied Physics Letters, Vol. 26, No.7, 1 April 1975 tial 1/I s is the l<ifference between the voltage V and the voltage drop Qm/Cox across the oxide: 1/Is = V - Qm/Cox + 1/150= s + 1/1 50 ' where s '" V - Qm/Cox is the surface potential increment caused by the applied voltage V. s can be measured with high precision using the static technique. From Eq. (1) we obtain s= (1 + Cc/Cox)V+ (Co/Cox)Uo' 1/1 50 is the surface potential at the voltage V = o. 1/I so must be determined independently. The oxide capaci- tance Cox can be obtained from the MOS capacitance in strong accumulation. The flat-band voltage is the value of Vat which V - Qm/Cox= -1/150' If the static measurements of 5 are combined with high-frequency MOS capacitance measurements one can determine 1/Iso' a nd, in addition, the semiconductor dop- ing density N v . We have applied the C ; ~ analysis S to the high-frequency MOS capacitance C HF C ac is the semi- conductor space-charge capacitance. The relations C ; ~ = - [2/( EsqN v A2) ](1/15 + kT /q), valid in depletion for n- silicon, and 1/C ac = 1/C HF -1/C ox lead to ( Cox _1)2 = _ 2(Cox/ A)2 (s + 1/I so + kT), C HF EsqN v q where A is the gate electrode area. Thus, the (Cox/C HF - 1)2 versus s curve is a straight line in depletion. The doping density N D can be obtained from the slope, and the additive constant 1/I so from the intercept on the s axis. If these values are known the interface state den- sity Nss can be obtained from Nss = (1/ Aq)[C LF (1- C LF / coxt 1 - C ac ). In depletion C sc could be taken from the high-frequency MOS capacitance. However, because in inversion the minority carriers do not follow the high- frequency ac signal, and in accumulation the interface states do follow, C sc was calculated from expressions Co Electro- meter Digitol - volt- meter FIG. 1. Basic circuit of the measurement setup. Copyright 1975 American Institute of Physics 400 Downloaded 17 Aug 2011 to 129.128.216.34. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions 1.5 N 1.0 . U. I
X 0 u 0.5 -0.8 -0.6 -O.L ..... 0 -02!IP kT -1 so q J6 s (Voltl FIG. 2. Measured (C.,./C HF -)2 vs <l>s values. given in Ref. 6. The sum of the charge Q ss in the inter- face states and the fixed charge Qf in the oxide also can be determined from Q ss + Qf = - Qm - Q.c. Q.c is the semiconductor space charge. Then, the interface state density can be obtained from the fact that Nss=' (I/Aq) x (dQss/dl/J s ), where knowledge of the low-frequency MOS capacitance is not necessary. In our static measurements, we have used a Keithley solid-state electrometer model 610c as an unity-gain operational amplifier. A Six-figure digital voltmeter (Hewlett Packard model 3462A) was used to measure the output of the electrometer. Before applying a new voltage all the capacitors were shorted for a short time in order to eliminate extraneous Signals which might arise from the off -set current of the electrom- eter' and the finite time constants of the capacitors. The voltage U B was applied slowly by use of a poten- tiometer to avoid causing instabilities in the electrom- eter. The capacitance Co, including the input capacitance of the electrometer, was measured in the following manner: Instead of the MOS capacitor, a voltage- and frequency-independent capacitor C 1 with a geometric form similar to that of a MOS capacitor was placed in the circuit. Then Co was obtained from the voltage division from the expression CO/C l = (U B - Uo)/U o C l , C e , and Cox were measured with a General Radio 1615A capacitance bridge by three terminal measurements. The high-frequency MOS capacitance was measured with a Boonton RF admittance bridge model 33B/l at 30 MHz. In order to check the accuracy of the static tech- nique' the MOS capacitor was replaced by several volt- age -independent capacitors C. For such linear capaci- tors, it follows that V - Qm/C =. O. Thus, the measured values of (V - Qm/C) indicate the error of the surface potential inc rement s' assuming that the error in the oxide capacitance is not larger than the possible error of the General Radio 1615A capacitance at 1 kHz (0.01%). The maximum error of s was 0.0002x I VI + O. 000 05 V, where the second term results from the nonlinearity of the MOS capacitance. The time between 401 Appl. Phys. Lett., Vol. 26, No.7, 1 April 1975 applying the voltage U B and measuring the voltage U o was 10 sec. Co had a value of 360 pF. The capacitor time constants were larger than 5x10 5 sec. Output of the electrometer used was equal to the input within 1 ppm. The off-set current was 1. 5xlO- l5 A. The maximum error of the low-frequency MOS capa- citance was 0.2%, occurring at extreme values of the C LF (V) curve. At more favorable positions for numeri- cal differentiating, for example, in accumulation and inverSion, the error was smaller than 0.05%. The measurement error of the doping denSity was smaller than 1%, and that of the additive constant I/J so smaller than 0.002 V. Near midgap, Nss values of a few times 10 9 V-l cm- 2 can be detected. The reported sample is a (111) -oriented n-type sili- con wafer oxidized in dry oxygen at 1140 C to a thick- ness of 1000 A, and annealed 30 min at 350C dry hydrogen after deposition of aluminum dots. The dots have a diameter of 0.5 mm. Figure 2 shows experimental values of (Cox/C HF - 1)2 versus s' A straight line is fitted to the experimental values. As long as the majority and minority carrier concentrations at the interface are negligible in com- parison to the doping density, an excellent fit is ob- tained. For the doping density the value (4.93 O. 05) XlO15 cm- 3 was measured, and for I/J so the value (0.128 0.002) V. In Fig. 3(a) curve (1) shows the distribution of the interface state density obtained by the static technique. ,---------------------,15.1011 la) ;:::;} static method aDD conductance method ........ staloc method D D D conductance method "l'E u Ib) Band Conduction Bond -to -0.8 - 0.6 - 0.4 - 0.2 _tV FIG. 3. Distribution of the interface state denSity. For curve (1) the interface is assumed to be an equipotential plane. For curve (2) surface potential fluctuations are taken into ac- count. (b) Interface state density on a logarithmic scale. K. Ziegler and E. Klausmann 401 Downloaded 17 Aug 2011 to 129.128.216.34. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions For comparison, the distribution as obtained for the same sample with the conductance method by Deuling, Klaus mann, and Goetzberger 7 is also shown. Near mid- gap, a reasonable agreement exists in accordance with the results in Ref. 8. Near the conduction band the curves differ. Possible causes for the difference are the assumption of an oxide capacitance which is too small, a nonuniform doping profile, or surface potential fluc- tuations. It is shown in Ref. 9 that surface potential fluctuations are the reason for the deviation, and that employing the equivalent circuit given by Brews 10 yields curve (2) which agrees very well with the results ob- tained by the conductance method. Figure 3(b) shows on a logarithmic scale the distribu- tion of the interface state density over an energy range of 0.9 eV within the energy gap. As in curve (2) of Fig. 3(a), the equivalent circuit of Brews 10 is used. The large range over which the interface state denSity can be determined with the new technique is remarkable. The interface state denSity increases strongly toward the band edges. Near the valence band, the denSity seems to be higher. The maximum at O. 40 e V from the valence band edge can be due to impurities ll ,12 or im- perfections, 13 e. g., excess oxygen incorporated into the interface. Energy levels at that position were mea- sured for implanted oxygen by Fahrner and Goetzberger. 14,15 The static technique described here is preferable to both the voltage slow-ramp method and the charge slow- ramp method if high accuracy, high resolution, and large measurement range are required. The static character of the new technique is useful in inversion. For fast measurements the slow-ramp method is pre- ferable. On the other hand, the surface potential can be determined more easily by the new technique and the charge slow-ramp method, because no integration is necessary. The analysis of the high -frequency MOS capacitance curve to obtain 1J!so used here is more re- liable than using the accumulation condition or the intrinsic condition as suggested by Berglund 16 and Koomen. 4 Moreover, that analysis is preferable, be- cause the doping denSity is obtained Simultaneously, and the uniformity of the doping profile right up to the inter- face can be checked easily. 17 In addition to the determi- nation of the MOS parameters mentioned above, the static technique is also useful for the determination of the semiconductor doping profile right up to the semi- conductor surface using an MIS capacitor 0 18 The authors wish to thank Professor A. Goetzberger, Professor CoR. Crowell, and Dr. K. Eisele for criti- cal reading of the manuscript and many helpful suggestions. lR. Castagne, C. R. Acad. Sci. B 267, 866 (1968). 2D. Kerr, International Conference on Properties and Uses of MIS Structures, Grenoble, 1969, p. 303 (unpublished). 3M, Kuhn, Solid-State Electron. 13, 873 (1970). 4J. Koomen, Solid-State Electron. 14, 571 (1971). 5E. H. Nicollian and A. Goetzberger, Bell Syst. Tech. J. 46, 1055 (1967). ~ . M . Brown and P. V. Gray, J. Electrochem. Soc. 115, 760 (1968). 7H. Deuling, E. Klausmann, and A. Goetzberger, Solid-State Electron. 15, 559 (1972). 8G.Declerck, R. van Overstraeten, and G. Broux, Solid- State Electron. 16, 1451 (1973). 9K. Ziegler and E. Klausmann (unpublished). 10J.R. Brews, J. Appl. Phys. 43, 3451 (1972). I1S. Kar and W. Dahlke, Appl. Phys. Lett. 18, 401 (1971). 12S. Kar and W. Dahlke, Solid-State Electron. 15, 221 (1972). 13B.E. Deal, M. Sklar, A.S. Grove, and E.H. Snow, J. Electrochem. Soc. 114, 266 (1967). 14W. Fahrner and A.Goetzberger, Appl. Phys. Lett. 21, 329 (1972) . 15W. Fahrner, J. Electrochem. Soc. 121, 784 (1974). 16C. N. Berglund, IEEE Trans. Electron. Devices ED-13, 701 (1966). 17K. Ziegler (unpublished). 18K. Ziegler, E. Klausmann, and S. Kar, Solid-State Electron. 18, 189 (1975). New high-speed bubble garnets based on large gyromagnetic ratios (high g) R. C. LeCraw, S. L. Blank, and G. P. Vella-Coleiro Bell Laboratories. Murray Hill, New Jersey 07974 (Received 4 December 1974; in final form 20 January 1975) A new approach to overcoming the problem of dynamic conversion in high-mobility bubble garnets is described based on large gyromagnetic ratios (high g factors). In a film of EUI.4,Yo.4,CaI.\Fe3.9Sio.6Ge0.5012' a g factor greater than 30 has been obtained, which increases the usable domain wall velocity before onset of dynamic conversion by more than an order of magnitude over comparable bubble garnet films with g approximately 2. Using a circuit period of 28.8 /-Lm, propagation data showing no deterioration in bias margins out to \0 7 steps were obtained on this film at 1 and 2 MHz. PACS numbers: 7S.60.F, 8S.70.G Attempts to increase the speed of LPE magnetic bub- ble garnets were limited initially by low domain wall mobilities usually -100 cm/sec Oe. When useful higher- mobility films became available, 1-3 another limitation to high-speed domain wall velocities was encountered. Experimentally, erratic propagation of individual bub- 402 Applied Physics Letters, Vol. 26, No.7, 1 April 1975 Copyright 1975 American Institute of Physics 402 Downloaded 17 Aug 2011 to 129.128.216.34. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions