CC BY
2
Visnyk N'l'UU KP1 Seriia Radiolekhnika tiadioaparat.obuduuannia, "2021, Iss. 85, pp. 69—74
Y^K 621.382
The Experimental Study of the Cerium Dioxide -Silicon Interface of MIS Structures
Korolevych L. M., Borisov A. V., Voronko A. O.
National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute" E-mail: korolevych. l.yubomyr&gmaU. com
The article is devoted to the actual task of studying a dielectric, which is an alternative to silicon dioxide in metal-insulator-semiconductor (MIS) structures. In metal-silicon dioxide-silicon structures, upon going to nanosize. the thickness of the dielectric film decreases so much that it becomes tunnel-transparent, and its breakdown voltage decreases. These phenomena can be eliminated by replacing silicon dioxide with a dielectric with a higher dielectric constant. These dielectrics primarily include oxides of transition and rare-earth metals. The parameters and characteristics of the MIS structure are determined by various factors, but the properties of the dielectric and the dielectric-semiconductor interface play a special role. In previous works of the authors, it was theoretically proved that cerium dioxide from a number of candidate dielectrics should have the best quality of the interface with silicon. This work is devoted to a study aimed at determining the fiat-band voltage and capacitance of MIS structures and at assessing the quality of the cerium dioxidesilicon interface. The study is carried out by the method of capacitance-voltage characteristics. For this, the high-frequency capacitance-voltage characteristics of the aluminum cerium dioxide silicon structures were measured at different temperatures. The capacity of the space charge region (SCR) in the enrichment and weak inversion modes of the near-surface layer of a semiconduct.oris considered. It is shown that the dependence of this capacitance in the (-2) degree on the voltage at the metal electrode c-2(Vg) is linear. The intersection of this line with the abscissa axis makes it possible to determine the fiat-band voltage. The slope tangent of this linear dependence makes it possible to determine the energy density of the charge at the dielectric semiconductor interface. It is shown that the charge density at the cerium dioxide silicon interface corresponds to the minimum values of the charge density at the silicon dioxide silicon interface. The absence of a shift in the capacitance-voltage characteristics of the structures under study with a change in temperature indicates the stability of the charge at the cerium dioxide - silicon interface.
Key words: MIS structure: cerium dioxide: capacitance-voltage characteristic (CV characteristic): fiat-band voltage: charge density at the dielectric-semiconductor interface
DOI: 10.20535/RADAP.2021.85.69-74
Introduction
The basis of modern micro- arid riarioelectroriics is metal-irisulator-seriiicoriductor (MIS) structures. The parameters arid characteristics of the MIS structure are determined by various factors, but the properties of the dielectric and the dielectric-semiconductor interface play a special role fl 3]. Until recently, silicon dioxide (Si02) films were used dielectric in such structures, since its physical characteristics almost completely satisfy technological and operational requirements. Among inorganic solid materials, Si02 has the largest band gap (about 9eV). is characterized by the highest resistivity (about 10230hnbcm), and the glassiness and ariiorphousriess of the films obtained by thermal oxidation in dry oxygen ensures their isotropy. However, the transition to nanotransistor electronics is faced with the problems of an ultra-thin dielectric [3]. which is associated with the low dielectric constant of
2
2
current, decreases the breakdown voltage of the dielectric. and increases the effect of high-temperature technological modes on the properties of the dielectric-semicondnctor interface.
2
a dielectric that has a higher dielectric constant and would meet the following conditions:
1. The technology for producing dielectric films must be compatible with the classical technology for manufacturing silicon MIS devices.
2. The process temperature is lower than the oxidation temperature of silicon in dry oxygen.
3. The quality of the dielectric-silicon interface is
2
Therefore, the search and study of a dielectric alternative to silicon dioxide is an urgent task.
70
Korolovvch L. M., Borisov A. V'., Voronko A. O.
1 Problem statement
The simplest approach to choosing a dielectric for MIS structures is based on the relationship between the dielectric constant of a dielectric and its band gap. which determines the height of the potential barrier at the insulator-semiconductor interface [3]. Based on this, a group of dielectrics with increased dielectric constant (high-k), suitable for silicon MIS structures, is distinguished. High-k dielectrics include transition metal oxides and rare-earth metal oxides. These materials meet the technological requirements, and their relative permittivity lies in the range e = 8 ... 30. In this regard, there is an intensive study of candidates for replacing silicon dioxide, such as Zr02 [4,5], AI2O3 [1,6,7], Y2O3 [81, Gc^Os [9,101, Hf02 [111, Ce02 [12]
and others.
The increased dielectric constant of the dielectric makes it possible to increase the thickness of the dielectric film, which automatically excludes tunneling currents through it and increases the breakdown voltage. However, in addition to the dielectric constant, the parameters and characteristics of MIS devices are also influenced by other dielectric parameters: the degree of ionicity, the crystal lattice parameter, the effective charge in the dielectric and charge at the dielectric-semiconductor interface, work function and band gap [3].
Engstrom et al. [13] proposed a choosing criterion for a dielectric based on numerical simulation of the carrier tunneling process through a dielectric. This criterion only allows to reduce the range of suitable materials: numerical calculations are complex and their result is unique for each specific case of the dielectric thickness. Thus, after changing initial conditions, the simulation result may also change, therefore, a material change will be necessary in the technological process.
In works [3,14], choosing criteria, that are focused on the choice of a certain dielectric, was proposed. In accordance with these criteria, the most suitable is a dielectric with a largest value of a figure of merit, which depends on the set of dielectric parameters. Although this criterion assumes the choice of a certain dielectric, the choice result does not guarantee reliability, due to the optimal (reference) values are parameters of a hypothetical dielectric.
In addition, the considered criteria do not take into account the quality of the dielectric-semiconductor interface and are designed exclusively for a silicon substrate. However, the properties of the dielectric-semiconductor interface have an exceptional impact on the parameters and characteristics of MIS devices. In this regard, the problem arises of developing such a criterion for choosing a dielectric that would take into account the state of this interface for any semiconductor substrate.
In [2] general choosing criterion for a dielectric for any semiconductor substrate of an MIS structure
theoretically obtained. This criterion is based on minimizing the value of the effective charge density at the dielectric-semiconductor interface. According to this criterion, the most satisfying the formulated requi-
2
electric has already found practical application in MIS structures for various purposes, the properties of the 2
The purpose of this paper is experimentally confirm the choice of a dielectric based on this criterion. For this, it is necessary to determine the fiat-band voltage and fiat-band capacitance of MIS structures
2
The study was carried out by the capacitance-voltage (C-V) characteristics method. For the study, MIS structures were fabricated by oxidation of a metal mirror (substrate temperature was 160oC, chamber pressure was 10-5Pa), on a silicon n-type substrate with volume resistivity 10 Ohm •cm. The thickness of
the dielectric films was ^300 nm, the gate area Aq
2
2 Gate dielectric capacity
To determine the fiat-band voltage and the charge density at the dielectric-semiconductor interface, it is necessary to know the gate dielectric capacitance of the MIS structure Ci [15]. This capacitance is part of the total structure capacitance Cmis, which includes the capacitance C/ connected in series with parallel connection of the capacitance of the near-surface layer of the semiconductor Csc and the charge capacity at the dielectric-semiconductor interface Cit (Fig. 1) [3].
\Ci
C
Fig. 1. Capacitive equivalent circuit of the MIS structure. C¡ - dielectric capacitance, Csc - capacitance of the near-surface layer of the semiconductor, Cit - charge capacity at the dielectric-semiconductor interface
In the case of a dielectric with a low dielectric constant, the gate capacitance is directly determined from the C-V characteristics of the MIS structure. It is equal to the capacitance in the saturation section of the C-V characteristic in the mode of enrichment of the near-surface layer of a semiconductor substrate with charge carriers. In this section of the C-V characteristic Cs is much larger than C¡ and it can be assumed that the total capacity of the structure is determined only by the gate capacitance and does not depend on the gate voltage Vq-
EKCii<;pnM<;iiTa.;ibii<; ^ocjii^>KoiiiiM moîkî po:s,ai.;iy ,aioKcn,a uopiio - KpoMiiiii M/J,H-CTpyKTvp
71
A specific feature of the C-V characteristics of the 2
2
not have a saturation region. This is due to the fact
2
comparing to Si02, and the capacitance Ci is quite large. In the enrichment mode, Ci is commensurate with the capacity Csc and therefore the gate capacity cannot be separated from the total capacity of the structure Cmis-
It should also be noted that, with a change in temperature, such MIS structures do not have a shift of the C-V characteristics along the stress axis (Fig. 2),
2
Si interface.
To determine the gate capacitance of the 2
was used. The essence of this method is to plot the dependence |dCMIS/dVG\1/2 = f(CMIS) using experimental C-V characteristics. In these coordinates, in the enrichment section of the C-V characteristic, it is a straight line. The point of intersection of this line with the abscissa gives the value of the gate capacitance. Based on the real C-V characteristics data of the 2
plotted ldCMis/dVGl1/2 = f(CMis) (Fig. 3). From which it follows that the capacitance of the structure of a particular sample is 435 pF.
400 300 200 100 0
□
cmis> pf
□ 20°C A40°C 060°C 080°C X100°C
Va, V
-1,1 0,9 2,9 4,9 6,9 8,9
Fig. 2. High-frequency C-V characteristics of Al-Ce02-Si structures for different temperatures
Fig. 3. Dependence IdCMis/dVa\1/'2 = f(CMis) for determining the gate capacitance Cj
2
3 The voltage and capacity of flat capacities:
bands
_ ___^\-1
CS _ CSC + Cit _l ç C /A j '
The gate capacitance Cj does not depend on the voltage, so in determining the flat band voltage we will Specific capacitance of the space charge region csC consider only a specific semiconductor and interface a depletion mode and weak inversion has the following
(1)
72
KopojifiiiU'i j'l. M., Bopucou O. B., BopoiibKo A. O.
dependence of the surface potential
<1
£s£oqNsub ^s - PT '
(2)
and surface potential is linearly related to gate voltage Vq:
VG = VFB +
CI - Cjt - cs CI
(3)
where Nsub is impurity concentration in a semiconductor substrate; q is elementary electric charge; £o is vacuum permittivity; £s is relative dielectric constant of a semiconductor; fp is temperature potential; Vps is flat-band voltage; cj is gate specific capacity; c*c is specific capacity of the SCR, provided that = y0; cit is surface charge specific capacity. For simplicity, let us denote the constant factor in front of by the coefficient n:
CI-Cjt -cs CI
(4)
Substituting the surface potential from ( ) to (2) we obtain the dependence of the SCR capacity on the gate voltage:
V
£s£oqNsub
(VG -VFB ) n 1 - <Pt'
In the coordinates c~2(Va) this dependence is linear, and the point of its intersection with the voltage axis makes it possible to determine the fiat-band voltage Vps-
VFB « Vg|c-2_^0 - n ■ ^T.
(6)
n =(ksc£s£oqNsub)-
(7)
Based on the experimental data of the C-V characteristics of the structure (Fig. 2) the dependence c~2(Vg) for determining the fiat-band voltage was plotted (Fig. 4). The plotted linear section of the dependence c~2(Vg), (which corresponds to the area of enrichment and weak inversion) by the method of least squares gives the value of the tangent ksc = 6 • 1015F
The concentration of the acceptor impurity Nsub = Na in the semiconductor substrate, determined from the Irwin curves [17], is l,35401scm-3.
Substituting value of ksc mid Na in ( ) we find that the coefficient n is 0,74.
Tims, according to the graph (Fig. 4) fiat-band voltage of the test sample is ^0,7V, and the capacitance corresponding to this voltage is 330 pF. Similar results were obtained for all structures of the batch in the amount of 10 samples (Table 1).
4 Charge density at the interface CeC>2 - Si
Surface charge density Dit and specific capacity cit related by the ratio cit = q ■ Dit. Then from ( ) we find:
(5)
D,
ci (l-n)-c*s
q
(8)
By the tangent of the slope of this line ksc coefficient n can be determined:
To determine the density of the surface charge, it is necessary to know c*c, the value of which is found from ( ) under the condition that = <fo = <Pt • ln(Nsub/ni). So, yo = 0,296 V, and c*sc = 2,8940-8F/cm. Substituting the data in (8) we obtain that the energy density of the charge at the Ce02-Si interface is on average 2,54011 eV-1cm-2. This valne corresponds to the charge density at the Si02-Si interface 10neV-1cm-2) of traditional 2,
oxidation in dry oxygen.
c„„ =
n
c
1
Fig. 4. Plot for determining the fiat-band voltage Vpb and capacitance Cpb Al-Ce02-Si structures. △ _ c-2(Vg); □ - experimental C-V characteristic Cm is (Va)'-,----fiat-band voltage and capacitance
EKCii<;pnM<;iiTa.;ibii<; ^ocjii^>KoiiiiM Me>Ki po:s,ai.;iy ,aioKcn,a nepiio - KpeMiiiii M/fH-crpyKTyp
73
Tafwi. 1 Parameters of studied structures
Sample № Flat-band voltage vFB,v Flat-band capacitance Cfb, pF Energy density of the charge at the dielectric-semiconductor interface Dit, eV-1cm-2
1 0,7 330 2,5 • 1 011
2 0,5 332 2,5 • 1 011
3 0,3 325 3,2 • 1 011
4 0,7 320 1,4 • 1 011
5 0,8 334 2,7 • 1 011
6 0,7 326 2,4 • 1 011
7 0,9 333 2,5 • 1 011
8 0,9 335 2,6 • 1 011
9 0,7 315 2,5 • 1 011
10 0,7 345 2,3 • 1 011
Conclusions
1. The flat-band voltage and capacitance and the charge density of the Ce02 - Si interface have been experimentally determined.
2. It is shown that the charge density at the Ce02 -Si interface corresponds to the minimum values of the
2
3. As the temperature changes, there is no shi-
2
structures. This indicates the stability of the charge 2
4. Significantly lower temperatures for obtaining Ce02 films (160°C) in comparison with thermal oxidation of silicon in dry oxygen (> 1000°C), excludes the effect of high-temperature operations on their properties.
2
2
ces will eliminate the influence of tunneling currents, while maintaining the high quality of the dielectric-semiconductor interface.
References
[1] Hlali, S., Hizem, N. and Kalboussi, A. (2016). High-fc dielectric materials for the gate oxide of a MIS capacitor: eilect of interface states on the C V characteristics. ■Journal of Computational Electronics, 15, pp. 1340 1350. doi: 10.1007/sl0825-016-0916-0.
[2] Korolevych, L. M. and Borisov, A. V. (2018). The Criterion for Choosing an Insulator for Silicon MOS Structures. Microsystems, Electronics and Acoustics, Vol. 23, Iss. 6, pp.6 12. doi: 10.20535/2523-4455.2018.23.6.141435.
[3] Kar, S. (ed.) (2013). High Permittivity Cate Dielectric Materials. 1st edn, Springer Series in Advanced Microelectronics (M1CROELECTR., volume 43). Springer Berlin Heidelberg, doi: 10.1007/978-3-642-36535-5.
[4] Zhou Y., Kojima N. and Sasaki K. (2008). Growth and dielectric properties of tetragonal Zr02 films by limited reaction sputtering. .Journal of Physics 1): Applied Physics, Vol. 41, Iss. 17, p. 175414. doi: 10.1088/00223727/41/17/175414.
[5] Liu, .1. et al. (2019). Structure and Dielectric Property of High-k ZrC>2 Films Grown by Atomic Layer Deposition Using Tetrakis (Dimethylamido) Zirconium and Ozone. Nanoscale Research Letters, Vol. 14, Iss. 1, Article number: 154. doi: 10.1186/sl 1671-019-2989-8.
[6] Wang, X. et al. (2016). Electrical properties and interfacial issues of high-fc/Si MIS capacitors characterized by the thickness of A12C>3 interlayer. , Vol. 6, Iss. 6, 065224. doi: 10.1063/1.4955001.
[7] Hlali, S. et al. (2017). Eilect of interface traps for ultra-thin high-fc gate dielectric based MIS devices on the capacitance-voltage characteristics. Microelectronics Reliability, Vol. 75, pp. 154 161. doi: 10.1016/j.microrel.2017.06.056.
[8] Rubio E. .1. et al. (2014). Electronic Structure and
23
and Interfaces on Silicon. The .Journal of Physical Chemistry C, Vol. 118, Iss. 25, pp.13644 13651. doi: 10.1021/jp502876r.
[9] Joseph A. et al. (2020). lon-lmplanted Epitaxially Grown
23
es. Journal of Electronic Materials, Vol. 49, Iss. 11, pp.6270 6275. doi: 10.1007/sll664-020-08392-4.
[10] Gribisch P. and Fissel A. (2020). Tuning of structural
23
Journal of Applied Physics, Vol. 128, Iss. 5, p. 055108. doi: 10.1063/5.0007793.
[11] Tirmali P. M. et al. (2011). Structural and electri-
2
MOS capacitors. Solid-Stale Electronics, Vol. 62, Iss. 1, pp.44 47. doi: 10.1016/j.sse.2011.04.009.
[12] Niu G. (2010). Epitaxy of crystalline oxides for functional materials integration on silicon. Ecole Centrale de Lyon, HAL Id: tel-00601689.
[13] Engstrom O. et al. (2007). Navigation aids in the search
fc
Solid-State Electronics, Vol. 51, Iss. 4, pp.622 626. doi: 10.1016/j.sso.'2007.02.021.
[14] Kar S. and Singh R. (2002). Correlation between the material constants of and a figure of meritfor the high-k gate dielectrics. In Kar S. (ed.) Physics and Technology of High-k Gate Dielectrics 1: Proceedings of International Symposium on High Dielectric Constant Materials: Materials Science. Salt Lake City, USA, pp. 13 23.
74
Королевич Л. М., Борисов О. В., Вороиько Л. О.
[151 D(i(in Л. and Champlain .1. G. (2011). High frequency capacitance-voltage technique for the extraction of interface trap density of the heterojunction capacitor: 'Italian's method revised. Applied Physics Letters, Vol. 99. Iss. 5, p. 053501. doi: 10.1063/1.3615279.
[16] Kar S. (2003). Extraction of the capacitance of ultrathin high-K gate dielectrics. IEEE 'transactions on Electron Devices, Vol. 50. Iss. 10. pp. 2112 2119. doi: 10.1109/TED.2003.817271.
[17] Irvin .1. C. (1962). Resistivity of Bulk Silicon and of Diffused Layers in Silicon. Bell System Technical Journal, Vol. 41, Iss. 2. pp. 387 410. doi: 10.1002/j.lo38-7305.1962.tb02415.x.
Експериментальне дослщження меж! роздшу дюксид цер!ю — кремнш МДН структур
Королевич Л. М., Борисов О. В., Вороиько А. О.
Робота присвячепа актуальпш задач! досл1джеп-пя д!електрика альтернативного дюксиду кромппо в структурах метал-д1електрик-пап!впров1дпик (МДН). У структурах метал-дюксид кремшю-кремпш. при переход! до папорозм!р!в. товщипа д!електричпо1 плвки змепшуеться пасильки, що стае тупелыю-прозорою i зпижуеться i"i папруга пробою. Виключити ц! явища можпа замшою дюксиду кромппо д!електриком з б!лын впсокою д!електричпою пропикшстю. До таких д!еле-ктришв в поршу чергу в!дпосяться оксиди перех!дпих i р!дкоземелышх метал!в. Параметри i характеристики МДН структури визпачаються р!зпими факторами, але особливу роль в!д!грають властивост! д!електрика i меж! д!електрик-пап!впров!дпик. У попередшх роботах автор!в теоретично доведено, що дюксид церпо з ряду д!електрик!в-претепдепт!в повипеп мати пайкращу як!сть меж! роздглу з кремшем. Дапа робота присвячепа досл!джепшо спрямовапому па визпачеппя папруги i емпост! плоских зон МДН структур та па оцшку яко-ст! меж! роздглу д!оксид церпо кромшй. Досл!джеппя проводиться методом вольт-фарадпих характеристик. Для цього були вим!ряш високочастотп! вольт-фарадп! характеристики структур алюмшш д!оксид церпо кромшй за р!зпих температур. Розгляпуто емшсть облает! просторового заряду (ОПЗ) в режим! збагачеппя ! слабко! iimepcii приповерхпевого шару пап!впров!дпика. Показано, залежшеть ц!е! емпост! в ( 2) ступепя в!д папруги на металевому електрод! с- (Va) мае лишний характер. Перетип ц!е! л!ш1 з в!ссю абсцис дае можли-в!сть визпачити папругу плоских зон. а тапгепс кута i'l пахилу епергетичпу пдльшеть заряду па меж! под!-лу д!електрик-пап!впров!дпик. Показано, що щгльшеть заряду па меж! дюксид церпо кремп!й в!дпов!дае м!-шмалышм значениям щ!лыюст! заряду па меж! дюксид кромппо кромшй. В!дсутп!сть зеуву вольт-фарадпих характеристик досл!джува1шх структур при зм!п! тем-ператури св!дчить про стабгльшсть заряду па кордоп! дюксид церпо кромшй.
Клюновг слова: МДН структура: дюксид церио: вольт-фарадпа характеристика (ВФХ): папруга плоских зон: щ!льшсть заряду па меж! под!лу д!електрик-пап!впров!д1шк
Экспериментальное исследование границы раздела диоксид церия — кремний МДП структур
Королевич Л. Н., Борисов А. В., Вороиько А. А.
Работа посвящена актуальной задаче исследования диэлектрика альтернативного диоксиду кремния в структурах метал-диэлектрик-полупроводник (МДП). В структурах метал-диоксид кремпия-кремпий. при переходе к папоразмерам. толщина диэлектрической пленки уменьшается настолько, что становится туппелыю-прозрачпой и снижается ее напряжение пробоя. Исключить эти явления можно заменой диоксида кремния диэлектриком с более высокой диэлектрической проницаемостью. К таким диэлектрикам в первую очередь относятся окислы переходных и редкоземельных металлов. Параметры и характеристики МДП структуры определяются различными факторами, по особую роль играют свойства диэлектрика и границы раздела диэлектрик-полупроводник. В предыдущих работах авторов теоретически доказано, что диоксид церия из ряда диэлектриков-претендентов должен иметь наилучшее качество границы раздела с кремшем. Эта работа посвящена исследованию направленному па определение напряжения и емкости плоских зон МДП структур и па оценку качества границы раздела диоксид церия кремний. Исследование проводится методом вольт-фарадпых характеристик (ВФХ). Для этого были измерены высокочастотные вольт-фарадпые характеристики структур алюминий диоксид церия кремний при разных температурах. Рассмотрено емкость области пространственного заряда (ОПЗ) в режиме обогащения и слабой инверсии приповерхностного слоя полупроводника. Показано, что зависимость этой емкости в ( 2) степени от напряжения па металлическом электроде с-2(Ус) имеет линейный характер. Пересечение этой липли с осыо абсцисс дает возможность определить напряжение плоских зон. а тапгепс угла ее наклона энергетическую плотность заряда па границе раздела диэлектрик-полупроводник. Показано, что плотность заряда па границе диоксид церия кремний соответствует минимальным значениям плотности заряда па границе диоксид кремпия кремний. Отсутствие сдвига ВФХ исследуемых структур при изменении температуры свидетельствует о стабильности заряда па границе диоксид церия кремний.
Ключевые слова: МДП структура: диоксид церия: вольт-фарадпая характеристика (ВФХ): напряжение плоских зон: плотность заряда па границе раздела диэлектрик-полупроводник
