In this chapter, characteristics of the sputtering Al2O3 films with PDA at temperature ranging at 750℃ to 950℃ for 30 sec are shown, respectively. Higher PDA temperature can effectively reduce leakage current. For the samples with PDA in an O2 ambient, it will induce extra interface growth to increase CET.[25][26] In a
reliability. The dielectric is shown to be a potential candidate for next generation high-k gate dielectric applications.
1
. RCA clean and LOCOS Si-SubstrateSi-Substrate
Al2O3
3
. RTA treatment N2/ 5 slm/30 sec or O2/ 5 slm/30 sec 750℃ 850 , ℃ ,950℃2
. Al2O3 deposition ( 60Å ) by Reactive Sputter4
. Top electrode formation TiN-2000 Å &Backside contact formation Al -5000Å
Al2O3
Si-Substrate
TiN
Al2O3
Si-Substrat
Fig.2-1 Process flows of experimental samples Al
-2 -1 0 1 2
750 800 850 900 950 1.5
2.0 2.5 3.0 3.5
O2-annealing N2-annealing
CET (nm)
PDA temperature (℃)
10-8
750℃_O
2 annealing 850℃_O
2 annealing 950℃_O
2 annealing
Current Density (A/cm2 )
Fig.2-4 The CET of Al2O3 samples after various post annealing temperature 750℃, 850℃,950℃
-5 -4 -3 -2 -1 0
750 800 850 900 950
1.0x10-8
O2ambient annealing N2ambient annealing
Fig.2- 7 The J at Vg= -1 V of Al2O3 samples with various PDA temperature Fig.2-6 The J-E curves of Al2O3 samples with various PDA temperature in a
N2 ambient.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
2 ambient annealing 850℃_O
2 ambient annealing 950℃_O
2 ambient annealing
0 1 2 3 4 5 6 7 8 9
2 ambient annealing 950℃_N2 ambient annealing
Fig.2- 8 The Weibull plot versus the charge to breakdown Q (C/cm2) of Al2O3 samples with various PDA temperature in an O2 ambient.
(a) As deposition
(b) O
2-800℃
750 800 850 900 950 1000
750 800 850 900 950 1000
0.55
Fig.2-11 Leakage current density,Rms values versus PDA temperature in an O2 ambient
Chapter 3
Characteristics of Al
2O
3Gate Dielectrics Using NH
3Surface Nitridation Technology
3.1 Introduction
As many reports, the direct contact of high-k materials and Si-substrate will be imperfect and have many issues. The dominance of the Si MOSFET over competing technologies has largely been attributed to the high quality of thermally grown SiO2 and the resulting Si/ SiO2 interface.[27] The Si/SiO2 interface is known to have a very low density of interface states ( Dit~2x1010 states/cm2 ) arising from unsaturated surface bonds and other electrically active imperfections.[27] Interface states lead to degradation of on-current, since carrier mobility is limited by scattering at the interface due to the strong vertical electric fields present in the channel. For maintaining the excellent transport properties at the Si interface, a possible method to suppress the interfacial layer thickness is to passivate the Si surface before Al2O3deposition. Generally, there are many methods to passivate the Si surface such as surface nitridation, nitrogen-contained ambient annealing, or nitride deposition as the bottom layer. Nitridation of the Si surface using NH3 treatment before the deposition of high-k materials has been shown to be effective in achieving the low EOT and preventing the boron penetration [28][29] However, this technique results in higher interface charges [30] which leads to higher hysteresis and
reduced channel mobility. In this chapter, we studied the effect of suppressing interfacial layer growth by NH3 surface treatments. The NH3 treatmentwould nitridize the Si surface to form a silicon nitride layer. [31]-[33] Silicon nitride is a superior barrier for H2O and oxygen, and it can suppress oxygen to diffuse into Si substrate.[34] After the NH3
treatment, a thin silicon nitride (SiNx) layer (~10Å) was deposited and measured by optical measurement system (Ellipsometer). As reports, Nitridation of the Si surface is prior to the deposition of high-k gate dielectrics and it shows the result to achieve the low EOT and increase reliability by making the interface smoother.[35] In this study, the effect of post thermal annealing was studied in the O2 and N2 ambient, respectively.
3.2 Experiment
First, the 6-inch P-type Si(100) wafers were cleaned with standand RCA clean. The samples were divided into two groups. One was without any surface treatment before Al2O3 deposition, and the other was with a NH3 surface treatment before Al2O3
deposition. The NH3-treatment is performed in high temperature furnace, 800℃ for 1 hour. After NH3-treatment, a SiNX layer(~10 Å) was deposited. Al2O3 dielectric was then deposited various thickness (30Å, 40Å, and 50Å) by reactive sputtering in the Ar/O (ratio
= 24/1) ambient. After Al2O3 deposition, a high temperature post deposition annealing was performed at 900℃ in the O2 or N2 ambient. Finally, the gate electrode formation was performed by the Physical Vapor Deposition (PVD) systems. We deposited the TiN film (2000Å ) as the top gate electrode and the thermal evaporation system to deposit the
Al film ( 5000Å ) as the backside contact. The cross-sectional view and total process flow were shown in Fig 3-1. The Capacitance-Voltage (C-V) and Current-Voltage (I-V) characteristics were measured by HP-4284 and HP-4156C systems, respectively. The capacitance equivalent thickness (CET) were extracted from C-V curve. In order to study the conduction mechanism in the Al2O3 film. The current-voltage characteristics with various temperature were measured at room temperature (RT), 50℃ , 75℃ , 100℃ , 125℃ , respectively.
3.3 Results and discussion
3.3-1 Capacitance-Voltage characteristic
Figure 3-2 shows the comparisons of the C-V curves of the samples with and without surface treatment after PDA 900℃ in an O2 ambient. Both sample shows less hysteresis and hump phenomenon. It is clear that samples with a NH3 treatment had higher capacitance at strong accumulation than samples without a NH3 treatment. The CET of Al2O3 filmis effectively reduced after NH3 nitridation since NH3 surface can suppress the growth of interfacial layer.[36] In Fig 3-2, we can also found the C-V curves shift negatively after NH3 nitridation can be observed due to the nitridation-induced fixed positive charges[37]. Since positive fixed charges in conventional NH3 nitridation film is due to N–H bonds at the interface [38]-[40]. Fig 3-3 shows the comparisons of the C-V curves of the samples with or without surface treatment after PDA 900℃in a N2 ambient.
We still can find that the CET of Al2O3 filmis effectively reduced after NH3 treatment.
However, the amount of the C-V curves shift negatively after NH3 nitridation is decreasing. The reason is due to that PDA in a N2 ambient will enhance the magnitude of fixed charge in Al2O3 film Fig 3-4 shows the variation of deposited thickness (measured by ellipsometer ) versus CET after PDA 900℃in an O2 ambient. the NH3 treatment can effectively reduce the CET despite the initial oxide thickness, which shows the excellent CET scalibility of this nitridation process. Because presence of Si3N4 layer can effectively suppress the diffusion of oxygen species into the high-k/Si substrate interface[41]. Without surface nitridation, Al and O atoms are easier to react with Si and are likely to form additional silicon dioxide and/or aluminum silicate layer with relatively lower k value.[42] As the result, the NH3 nitridation treatment can effectively suppress the interface growth to lower the CET and increases the effective dielectric constant. [43]
Fig 3-5 shows the variation of Al2O3 thickness versus capacitance equivalent thickness (CET) after PDA 900℃ in a N2 ambient. We can see almost the same result in an O2
ambient as in a N2 ambient, except in a N2 ambient with higher k value. So the PDA in a N2 ambient exhibits superior behavior than PDA in an O2 ambient.
3.3-2 Current density-Electric field characteristic
Figure 3-6 (a) and (b) show the relationship of gate leakage current versus gate bias after PDA 900℃ in the O2 and N2 ambient, respectively. The NH3 –treatment samples show lower leakage current even with thinner CET. It is postulated that the NH3-treatment can improve the quality of interface between high-k and silicon substrate and effectively
reduce the leakage current.[44] Fig. 3-7 shows the curve of Current density versus CET curves at Vg = -1V with NH3 treatment and without NH3 treatment. It is clear that NH3
treatment can effectively reduce leakage current. According to the bonding strength comparisons as follow, Si-O(8.42eV) > Si-N(4.75eV) > Si-Si(3.38eV) > Si-H(3.18eV) The Si-N bonds have larger bonding strength than the Si-Si bonds or Si-H bonds.
Therefore, we preferred to use the stronger Si-N bonds to replace Si-H bonds, subsequently resist the oxygen diffusion and reduce defect-generation.[45] We also suspect that surface nitridation may effectively reduce the concentration of oxygen vacancies during PDA due to its capability to suppress oxygen diffusion.[42] Therefore, the leakage current will be reduced large. However, as the requirement of the gate dielectric thickness shrinks to below 2nm, the excessive direct tunneling current dominates the gate leakage characteristics and limits scalability of the conventional SiO2
gate dielectrics. As the result, NH3 treatment process become potentially to reduce the gate leakage current to meet the request for deep sub-micron CMOS applications.
3.3-3 Time Dependent Dielectric Breakdown
In this study, the dielectric reliability was also investigated with the behavior of time dependent dielectric breakdown (TDDB). The gate dielectric failure occurred when the significant gate leakage was observed. The magnitude of charge to breakdown (QBD) can be calculated in TDDB measure. Fig 3-8(a) and (b) show the weibull plots of the charge to breakdown (QBD) of Al2O3 samples with nearly the same CET after PDA 900 in ℃
theO2 and N2 ambient , respectively. The samples with NH3 treatment shows more charge to breakdown (QBD). Thus, we suspect that surface nitridation may effectively reduce the concentration of oxygen vacancies during PDA due to its capability to suppress oxygen diffusion.[42] On the other hand, the stronger Si-N bonds bring a stronger interface layer.[35] When part of voltage drop across the Si3N4 interfacial layer in samples, the voltage drop across the Al2O3 film is lower than expectance, and the influence of the electric field stress is not as severe as the un-NH3 treatment Al2O3 samples. As the result, it will enhance the QBD for NH3 treated Al2O3 samples. Besides, the slope of the Weibull distribution is an important factor in reliability calculations, where it is used for scaling to total oxide area on chip and low percentiles. A low Weibull slope results in a strong reduction of the QBD. A value of slope of 0.9~1.5 is very low for a gate dielectric layer.
For thermally grown SiO2 with a physical thickness of 9 nm, is expected slope = 12 for intrinsic breakdown.[46] Similarly low values below 2 (dotted line) are measured for the other ALCVD high-k layers grown at IMEC as well as for layers reported by other research groups, as citation in Fig.3-9. Breakdown is determined by sputter-induced defects causing weak spots in the Al2O3 film. The percolation model is not applicable and the slope values are always low.[47] The sputtered Al2O3 film seems to produce a lot of initial defects during a PVD process. Therefore, other deposition technology should be considered such as MOCVD and ALCVD..
3.3-4 Conduction Mechanism
improve its electrical and dielectric properties. Comparing the conduction mechanism reported, the current transport mechanism in Al2O3 is still unclear and seems to be strongly process dependent.[48] There may be different conduction mechanisms in the insulator thin film. Typically, two possible effects are in the metal-insulator interface, one is Schottky effect ,the other is Frenkel-Poole effect. The Schottky-Richardson emission generated by the thermionic effect is caused by the electron transport across the potential energy barrier via field-assisted lowering at a metal-insulator interface. The leakage current equation is:
where βs =(e3/4πε0ε)12,A* effective Richardson constant,φs the contact potential barrier. We can find the slope of the leakage current equation.
The Frenkel- Poole ( F-P ) emission is due to field-enhanced thermal excitation of trapped electrons in the insulator into the conduction band. The leakage current equation is:
12 0 3
FP = (e /πε ε)
β ,e the electronic charge,ε0 the permittivity of free space,ε the high
frequency relative dielectric constant, T absolute temperature ,E the applied electric filed, KB the Boltzmann constant,φPF the contact potential barrier. We can find the slope of the leakage current equation.
From the equations as shown above, leakage current behaviors of insulate films can be investigated further on the leakage current density (J)-electric field (E) characteristics such as ln J vs. E1/2 plots. The plot of the nature log of leakage current density versus the square root of the applied electric field was observed. It is found that the leakage current density is linearly related to square root of the applied electric field. The linear variations of the current correspond either to Schottky emission or to Frenkel-Poole conduction mechanism. For trap states with coulomb potentials, the expression is virtually identical to that of the Schottky emission. The barrier height, however, is the depth of the trap potential well, and the quantity β is larger than in the case of Schottky emission by a FP
factor of 2. Distinction between the two processes can be done by comparing the theoretical value of β with the experimental one obtained by calculating the slope of the curve ln J-E1/2. The dielectric constant of Al 2O3 is 7.63 at PDA O2 ambient and 8.29 at PDA N2 ambient extracted by Fig 3-4 and-5, the theory β values are 4.40×10-23 for Frenkel-Poole and 2.20×10-23 for Schottky after PDA in an O ambient and the theory β
T
values are 4.22×10-23 for Frenkel-Poole and 2.11×10-23 for Schottky after PDA in a N2 ambient. Table 3-1 (a) and (b) shown the experimental β and Schottky barrier high of Al2O3 samples with NH3 and without NH3 treatment after PDA 900℃in an O2 ambient and after PDA 900℃in a N2 ambient ,respectively .Fig 3-10 (a) and (b) show the conduction mechanism fitting of Al2O3 samples with NH3 and w/o NH3 treatment after PDA 900℃in an O2 ambient and after PDA 900℃ in a N2 ambient, respectively. We find the conduction mechanism in Al2O3 thin film is dominated by Schottky conduction.
However, Schottky conduction depends strongly on the barrier between metal and insulator and has the inclination to occur for insulators with fewer defects and a more perfect metal-insulator interface.[49] Samples after NH3 treatment will have higher barrier high and it is clear that the NH3 treatment is very effective for improving the interface properties like the barrier height and Al2O3 film to reduce the leakage current.[50]
3.4 Summery
In this chapter, characteristics of the fabricated Al2O3 samples with NH3 and without NH3 treatment after PDA 900℃ in an O2 ambient and at PDA 900℃ in a N2 ambient are present, respectively. With surface NH3 treatment, the CET of Al2O3 film can be reduced due to the suppression of interfacial layer growth. In samples with a NH3 pre-treatment had lower leakage current and better dielectric reliability. Therefore, the NH3 pre-treatment is a potential technique to improve the performance in high-k gate
dielectric applications.
‘
1
. RCA clean and LOCOS Si-Substrate2
. Without treatmentSi-Substrate Si-Substrate
5
. Top electrode formation TiN -2000 Å pattern contact& Backside contact formation Al -5000Å contact
Al2O3
Fig.3-1 Process flows of experimental samples Si-Substrate
SiNX
SiNX
Al Al
-2 -1 0 1 2 without surface treatment after PDA 900℃in an O2 ambient
Fig. 3-3: C-V curve of Al2O3 sample with surface treatment compare to Al2O3 sample without surface treatment after PDA 900℃in a N ambient
3.0 3.5 4.0 4.5 5.0 5.5 6.0
w/o NH3treatment k=6.56 interface layer thickness=1.4nm NH3treatment k=7.63 interface layer thickness=1.1nm
3.0 3.5 4.0 4.5 5.0 5.5 6.0
3treatment k=7.97 interface layer thickness=1.2nm NH3treatment k=8.29 interface layer thickness=0.9nm
CET(nm)
Optical thickness(nm)
Fig. 3-5: CET versus Optical thickness of Al2O3 sample with surface treatment compare to Al2O3 sample without surface treatment after PDA 900℃in a N2 ambient Fig. 3-4: CET versus Optical thickness of Al2O3 sample with surface treatment
compare to Al2O3 sample without surface treatment after PDA 900℃in an O2 ambient
-4 -2 0
Fig. 3-6-(a): The gate leakage current density (J) versus electric field (E) curves for NH3 nitrided and w/o NH3 nitrided after PDA 900℃in an O2 ambient
Fig. 3-6-(b): The gate leakage current density (J) versus electric field (E) curves for NH3 nitrided and w/o NH3 nitrided after PDA 900℃in a N2ambient
1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2
w/o NH3 treatment slope=1.67927 w/o NH3 treatment slope=1.21684 NH3 treatment slope=1.1456 NH3 treatment slope=0.76316
Fig. 3-7 The Current density at Vg = -1Vversus EOT curves for NH3 treatment and
Fig. 3-8 –(a)The weibull plot shows charge to breakdown for Al2O3 samples with NH3
and w/o NH3 treatment at PDA after PDA 900℃in an O2 ambient
Fig. 3-7 Current density at Vg = -1Vversus CET curves for NH3 treatment and without NH3 treatment.
2 4 6 8 10 12 14 16 18 20 22 24 26 28 -2.5
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0
CET=1.95nm CET=2.26nm
CET=1.69nm CET=1.95nm
w/o NH3 treatment slope=1.1456 w/o NH3 treatment slope=0.44363 NH3 treatment slope=0.92516 NH3 treatment slope=0.13322
ln(-ln(1-F))
QBD(C/cm2)
Fig. 3-8 –(b)The weibull plot shows charge to breakdown for Al2O3 samples with NH3
and w/o NH3 treatment at PDA after PDA 900℃in a N2 ambient
Fig. 3-9. Measured Weibull slopes are plotted versus the physical layer thickness, and a clear gap between the slopes for SiO2 and the high-k layers can be observed.[20]
w/o NH3 treatment With NH3 treatment
βexp Schottky barrier high βexp Schottky barrier high
25℃ 1.88*10-23 0.71ev 2.25*10-23 0.97ev
βexp Schottky barrier high βexp Schottky barrier high
25℃ 1.93*10-23 0.73 ev 1.98*10-23 0.83ev
CET=3.14nm schottky effective barrier high=0.73ev
CET=2.66nm schottky effective barrier high=1.126ev Schottky emission dominated
Table.3-1(a) The experimental β and Schottky barrier high of Al2O3 samples with NH3 and w/o NH3 treatment after PDA 900℃in a O2 ambient
Table.3-1(a) The experimental β and Schottky barrier high of Al2O3 samples with NH3 and w/o NH3 treatment after PDA 900℃in a N2 ambient
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
CET=2.35nm schottky effective barrier high=0.72ev
CET=2.01nm schottky effective barrier high=0.96ev
E1/2(MV/cm)1/2
Chapter 4
Conclusions and Suggestions for Future Work
4.1 Conclusions
In this thesis, characteristics and reliability of Al2O3 gate dielectrics with pre-deposition NH3 treatment and rapid thermal annealing in O2 and N2 ambient have been investigated. Several important phenomena were observed and summarized as follows.PDA can effectively reduce leakage current. For the samples with PDA at O2 ambient, the interfacial layer increases with the higher PDA temperature. It is obviously that O2 penetration will induce the increasing of the interfacial layer at Al2O3/Si-substrate and higher annealing temperature will speed up interface layer growth rate. For samples with PDA in a N2 ambient, the void–defect of sputtered Al2O3 film will be eliminated. The as deposited Al2O3 film shows larger roughness, and after PDA it will become smoother. However, with higher PDA temperature, surface rough will slightly increase. The surface NH3 treatment, can lower the CET value, and reduce the leakage current. In additional, the dielectric reliability was enhanced with NH3 treatment. The low Weibull slope of Al2O3 film may result from sputter-induced defects causing weak spots. Significant process improvements are necessary to enhance dielectric quality ,such as MOCVD, ALCVD. The conduction mechanism in Al O film was investigated by the various temperature measurement
and fitting. The conduction mechanism in Al2O3 thin film is dominated by Schottky conduction which occured for insulators with fewer defects and a more perfect metal-insulator interface. Samples after NH3 treatment will have higher barrier high to reduce the leakage current.
4.2 Suggestions for Future Work
From recent reports, the major potential show-stoppers for high-k gate dielectrics are considered to be (a) interfacial layer thickness and quality, (b) film morphology after the whole thermal process, (c) reliability issues. Base on the above results, Al2O3 gate dielectrics with pre-deposition NH3 treatment and rapid thermal annealing can have very notable improvement. But still several works are worthy to do in the future and are recommended here.
(1). MOSFET devices fabrication with the above results：
The issues in the integration of Al2O3, and it’s performances can be investigated with the device structures. Mobility is the first concern for considering device performance.
Interfacial layer thickness and quality are related to mobility degradation and scaling limit.
(2). More potential interfacial layer investigation：
It is difficult to make a balance between the low leakage current tunneling and the low EOT. The quality of the NH3 treatment interfacial layer still could be improved.
a key to answer this question.
(3). More potential surface treatment investigation：
For the NH3 surface treatment, the excellent improvement of properties are observed. However, the large traps are still existence in NH3 surface treatment samples. Other more potential interfacial treatments maybe can be developed to minimize the defects. For example：N2O gas treatments.
(4). More potential deposition method investigation：
For reactive sputtering in Ar/O ambient to deposit Al2O3 film, it can not avoid with sputter damages. These damages will become leakage path and can not be accepted to nanometer CMOS fabrication. Other potential manufacturing system like MOCVD maybe can be developed to deposit high quality Al2O3 film.