Summary

According to SIA roadmap, the oxide thickness smaller than 20Å is necessary for deep sub-quarter micron devices. However, pure SiO2 can not meet the requirement due to the large tunneling current. In our study, N atomic concentration is shown to depend on the initial oxide thickness, i.e., with decreasing oxide thickness from 63Å to 22Å, the N concentration increases from 2.12 to 4.45 at.% in the interface, which was desired for the improvement of dielectric reliability in the

ultrathin region. NO-annealing can achieve better SILC immunity for both constant voltage and constant current stress. Moreover, NO-annealing also improves interface smoothness and results in tighter TDDB distribution. Even after process optimization in the future, NO-annealing can be used to improve device performance more apparent, as predicting in mind. Although boron penetration is lack in our project, many studies have been shown the improvement of NO-annealing on pMOSFET performance. As a result, NO-annealed nitrided oxides can improve dielectric reliability and are suitable to replace traditional SiO₂ at 0.13µm and beyond.

Table 2.1 Comparison of thickness extraction from various methods of NO- and N2-annealed GOX.

DTIV QMCV EOT

Rudolph

-1V -1.5V @ -2V

TEM Quantox

@ -2V NO-30Å 30.6Å 25.7Å 25.4Å 33.9Å 33.7Å 33.7Å 42.0Å

NO-22Å 21.5Å 26.1Å 25.3Å 24.9Å 25.8Å 28.2Å 32.0Å

NO-19Å 19.7Å 24.6Å 23.3Å 22.6Å 23.8Å 24.7Å 27.5Å N₂-30Å 30.7Å 28.1Å 29.6Å 36.0Å 35.7Å 37.9Å 44.3Å

N₂-22Å 22.7Å 26.8Å 25.5Å 26.3Å 27.3Å 29.8Å 33.6Å

N2-19Å 18.9Å 25.3Å 24.2Å 25.2Å 26.3Å 27.4Å 32.4Å

Table 2.2 Flat-band voltages and interface state densities of NO- and N2-annealed GOX.

V_FB

T_OX NO-GOX (V) N2-GOX (V) ∆V_FB (mV)

30Å -1.065 -1.052 -13

22Å -1.085 -1.055 -30

19Å -1.104 -1.06 -44

Dit

TOX

NO-GOX (V) N2-GOX (V) ∆Dit (cm^-2 eV^-1) 30Å 6.31×10¹¹ 5.47×10¹¹ 0.84×10¹¹ 22Å 7.06×10¹¹ 4.88×10¹¹ 2.18×10¹¹ 19Å 8.66×10¹¹ 4.59×10¹¹ 4.07×10¹¹

-2.0 -1.5 -1.0 -0.5 0.0 0.5

Fig. 2.1 The C-V curves of the 19Å NO-annealed GOX. Inset shows corrected C-V curves by two-frequency method.

20 30 40 50 60 70 80

Electrical Thickness ( A )

TEM Physical Thickness ( A ) EOT

QMCV

Fig. 2.2 Extracted dielectric constants of NO- and N2-annealed GOX. NO-annealed GOX has higher dielectric constant than N2-annealed GOX.

N2-annealed GOX κ = 4.0

NO-annealed GOX κ = 4.2

15

Oxide Thickness ( A )

QMCV @ -2V TEM

DTIV @ -1V DTIV @ -1.5V

Fig. 2.3 Extracted oxide thickness variations between DTIV and QMCV of NO- and N2-annealed GOX. Both DTIV and QMCV thickness show highly agreement with TEM thickness.

Leakage Current Density ( A / cm2 )

Electric Field ( MV / cm ) N2-annealed

NO-annealed

Fig. 2.4 Direct tunneling current density of 19Å and 22Å NO- and N2-annealed GOX.

NO-annealing does not deteriorate leakage current.

-2.0 -1.5 -1.0 -0.5 0.0 0.5 0.0

0.2 0.4 0.6 0.8 1.0

Normalized Capacitance

Gate Voltage ( V )

N2-annealed NO-annealed

(a)

-1.2 -1.1 -1.0 -0.9 -0.8

0.2 0.3 0.4 0.5 0.6 0.7

NO-annealed N₂-annealed

Normalized Capacitance

Gate Voltage ( V )

19A 22A

(b)

Fig. 2.5 (a) Normalized C-V curves of 19Å and 22Å NO- and N2-annealed GOX. (b) Magnify (a) for clearly comparison between NO- and N2-annealed GOX.

NO-annealing exhibits negative flat-band voltage shift and slightly increases interface state density.

0 20 40 60 80 100 120 140 160

Nitrogen atomic concentration ( at. % )

Depth ( A )

N Atomic Concentration ( at. % )

Initial Oxide Thickness

1e21 2e21 3e21 4e21

N Concentration ( cm-3 )

(b)

Fig. 2.6 (a) XPS nitrogen depth profiles (b) nitrogen peak concentrations of NO-annealed nitrided oxides varied with initial oxide thickness. NO-annealing piles-up nitrogen near the interface and more N incorporates as oxide thickness decreasing.

0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0 1E-9

1E-8 1E-7 1E-6 1E-5 1E-4 1E-3 0.01 0.1

CVS, -3.8V, 100sec

N2-annealed GOX NO-annealed GOX

Leakage Current Density ( A / cm2 )

Gate Voltage ( V )

Fig. 2.7 SILC characteristics of 19Å NO- and N2-annealed GOX at -3.8V CVS for 100sec. Maximum current increment after CVS occurs near flat band region.

0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0

Leakage Current Density ( A / cm2 )

Gate Voltage ( V )

Stressing Time ( sec ) (b)

Fig. 2.8 (a) SILC characteristics (b) transient hole-trapping behavior of 22Å NO- and N2-annealed GOX at -4V CVS. NO-annealing can suppress both hole-trapping and maximum trap generation rate occurred near flat-band region during constant voltage stress.

0 50 100 150 200 250 300

Maximum Trap Generation Rate ( % )

Time ( sec ) Constant Current Stress -20 mA / cm²

Maximum Trap Generation Rate ( % )

Injected Charges ( C / cm² ) (b)

Fig. 2.9 Maximum trap generation rate of 22Å N2- and NO-annealed GOX stressed at (a) -4V constant voltage stress (b) -20 mA/cm² constant current stress. NO-annealing can reduce hole-trap generation rate during both voltage and current stress.

1 10 100 1000 10000

TDDB Weibull plots ( CVS -4.3V )

ln ( - ln ( 1-F ) ) CCS. NO-annealed nitrided oxides exhibits higher stress immunity than N2-annealed oxides.

Fig. 2.11 The HRTEM cross-sectional images of 22Å (a) N2- and (b) NO-annealed GOX. NO-annealing has smoother interface than N2-annealing.

-4.0 -4.2 -4.4 -4.6

2.0 2.5 3.0

3.5 CVS

22A GOX 19A GOX

Open : N₂-annealed GOX Close : NO-annealed GOX

Weibull Slope

Gate Voltage ( V )

Fig. 2.12 Plots of weibull slope of 19Å and 22Å N2- and NO-annealed GOX.

NO-annealing clearly enhances Weibull slope.

n⁺-poly

27.3Å ± 2.3Å

p-Sub (a) N2-annealed GOX

5 nm

25.8Å ± 1.7Å n⁺-poly

p-Sub (b) NO-annealed GOX

5 nm

-1.0 -1.5 -2.0 -2.5 -3.0 -3.5 -4.0 -4.5 -5.0 10⁰

10¹ 10² 10³ 10⁴ 10⁵ 10⁶ 10⁷ 10⁸ 10⁹ 10¹⁰

-2.82V -2.56V

-2.18V

22A GOX

19A GOX Open : N

2-annealed GOX Close : NO-annealed GOX

Lifetime ( sec )

Gate Voltage ( V )

Fig. 2.13 10-year lifetime projection of 19Å and 22Å N2- and NO-annealed GOX.

Inset also indicates voltage acceleration factors.

19Å -2.96

γ N₂

22Å -2.88

-3.27 NO

-3.16

CHAPTER 3

Characteristics and Reliabilities of Sub-3nm Nitrided Oxides with Nitrogen-Implanted Silicon Substrates

3.1 Introduction

Recently, many efforts including oxygen, argon and nitrogen implantation prior to thermal oxidation, were used to form multiple gate oxide thickness in one thermal cycle to meet system-on-a-chip (SOC) requirement [73]-[78]. Oxygen implantation can improve the oxide quality [73], but it seems unsuitable to apply in the deep-submicron generation since oxygen implantation will enhance oxidation rate and limit the minimum oxide thickness. There are a few reports on argon implantation, however, it is able to provide more thinner oxides than the oxygen implantation [74].

One of the concerns about argon implantation is that the implantation damage caused by heavy mass argon ions may need longer thermal cycle to annealing gate oxides. On the other hand, the nitrogen-implanted Si substrate (NIS) prior to gate dielectrics growth seems the most proper technology to meet the requirement of deep submicron devices [55], [75]-[78]. Since the nitrogen implantation can retard oxidation rate and enhance the control ability of thinner oxide thickness, it is much more suitable to be implemented in SOC technologies. Moreover, nitrogen will pile up at the oxide/substrate interface and suppress boron penetration [78]-[80], which was not observed in both oxygen and argon implantation techniques. In this chapter, the

characteristics and reliabilities of NIS nitrided oxides are studied. NO-annealing will pile-up nitrogen near the interface, while NIS will introduce uniformly distribution nitrogen in the dielectric bulk. With optimized implantation dosage and post-oxidation annealing temperature, NIS nitrided oxides could effectively suppress trap generation and improve both time-to-breakdown (t_BD) and charge-to-breakdown (Q_BD), also suitable to reduce process steps of the SOC technology.

3.2 Experiment Details

All experimental split conditions of nitrided oxides are detailed in Table 3.1.

LOCOS isolated MOS capacitors were fabricated on p-type (100) silicon wafers.

After forming LOCOS isolation, nitrogen was implanted into the Si substrates at the split dosage of 1×10¹³, 1×10¹⁴ and 1×10¹⁵ cm^-2 with 10keV. In addition, some wafers without nitrogen implantation were used for comparison. Wafers were then cleaned and HF dipped before oxidation. The gate dielectrics were grown at 750^oC followed by either NO or N₂ annealing at 850^oC for an hour. Then 1500Å polysilicon was deposited with in-situ doped phosphorus of 2.5×10²⁰ cm^-3. Dopants were then activated at 950^oC for 30sec. After gate electrodes patterned and contact holes etched, aluminum metallization was done followed by sintering at 450^oC in N2 ambient.

Square or circular capacitors of different areas, ranging from 2.5×10^-5 to 1×10^-2 cm², with LOCOS isolation are used to evaluate the gate oxide integrity. The physical gate oxide thickness was determined by spectroscopic ellipsometer and compared with high-resolution transmission electron microscopy (HRTEM). The equivalent oxide thickness (EOT) was extracted by fitting the measured high-frequency

capacitance-voltage (C-V) data from Hewlett-Packard (HP) 4284 LCR meter under an accumulation condition with quantum mechanical correction. The tunneling leakage current density-electric field (J-E) and the reliability characteristics of MOS capacitors were measured by semiconductor parameter analyzer HP4145A. Nitrogen depth profiles and compositions were analyzed by secondary ion mass spectroscopy (SIMS) and X-ray photoelectron spectroscopy (XPS). The micro-roughness of the wafer surface and the interface between nitrided oxides/silicon were detected by atomic force microscopy (AFM).

3.3 Results and Discussions

Recently, the technique of nitrogen-implanted into Si substrate has been paid more and more attention on nitrided oxide growth and SOC application [56], [75]-[78].

NIS has the advantage in growing multiple gate oxide thickness at one thermal cycle and is suitably applied to SOC technology. However, the most concerning issue for NIS is the surface damage during nitrogen implantation. Thus, post-implantation thermal cycle must be very careful in order to fully anneal out the damage to guarantee the oxide quality.

3.3.1 Basic Characteristics of Sub-3nm NIS Nitrided Oxides

The C-V curves of 22Å N₂- and NO-annealed NIS nitrided oxides are shown in Fig. 3.1(a) and (b), respectively. For low implantation dose, 1×10¹³ and 1×10¹⁴ cm^-2, no obvious difference in C-V curves has been seen. However, significant V_FB recovery and D_it degradation observes with heavy implanted nitrided oxide, i.e. 1×10¹⁵ cm^-2,

with respect to sample without NIS. Since N is a donor-type impurity, the substrate acceptor concentration will decrease by the implanted N atoms before oxidation, and results in smaller V_FB [75]. Table 3.2 lists the flat-band voltages and interface state densities of NO- and N2-annealing NIS nitrided oxides.

Figure 3.2(a) compares oxide thickness as a function of nitrogen implantation dose into the Si substrate. The oxidation rate drops continuously with increasing nitrogen dose [74]-[80], [81], [82]. In our study, slightly enhanced oxidation rate with NIS smaller than 1×10¹⁴ cm^-2 is ascribed to insufficient annealing out the damage by ion bombardment. Significant oxidation rate suppression only has been observed with NIS larger than 1×10¹⁴ cm^-2. Accordingly, the oxidation rate is trade-off by the residual damage annealing and oxidant diffusion rate constraint, resulting in the increased oxidation rate firstly and then decreased. With a heavy implant dose of 1×10¹⁵ cm^-2, the growth rate can be reduced by 15%, which is smaller than that reported in Ref. 76, probably due to lower oxidation temperature (750^oC) in our study.

The difference between TEM and QMCV thickness is larger for sample with 1×10¹⁵ cm^-2 NIS than 1×10¹³ and 1×10¹⁴ cm^-2 NIS, which can be ascribed to the heavier N concentration induced larger dielectric constant for sample with 1×10¹⁵ cm^-2 NIS.

Dielectric constant (κ) as a function of nitrogen implantation dosage into the Si substrate is shown in Fig. 3.2(b). The κ-value of samples without NIS is extracted by extra-plotting of QMCV and TEM thickness (shown in Fig. 2.2), while the κ-value of samples with NIS is extracted directly from the QMCV and TEM thickness. κ-value increase has only been observed in the sample with the heavy implant. Post-oxidation NO-annealing can further increases κ values. Thickness and κ variation for NIS nitrided oxides is summarized in Table 3.3. As the nitrogen dose increases, dielectric constant will slightly increase from 3.9 of pure SiO₂ to 4.6 of sample with

NO-annealing and 1×10¹⁵ cm^-2 NIS. NO-annealed GOX still has higher dielectric constant than N2-annealed GOX due to higher nitrogen concentration.

Figure 3.3 compares the leakage current density for 22Å N2-annealed and NO-annealed NIS nitrided oxides at negative gate bias polarity. When NIS is less than 1×10¹⁴ cm^-2, the leakage current increases slightly due to the residual implantation damages. Nevertheless, as NIS dosage increases to 1×10¹⁵ cm^-2, the leakage current will increase more obvious since the oxidation rate will be suppressed by the heavy nitrogen-passivated surface and result in thinner oxide thickness. In short, post oxidation NO-annealing will introduce N into the gate oxides/substrate interface, increase V_FB shift and D_it. On the contrary, while nitrogen is donor impurity to Si, partial implanted-nitrogen may retain on Si substrate without forming Si3≡N bonding and counter-doping substrate concentration. Consequently, NIS before oxidation will reduce substrate doping, and slightly recover VFB shift.

3.3.2 Reliability Characteristics of Sub-3nm NIS Nitrided Oxides

Figure 3.4(a) displays the XPS depth profiles of NO-annealed nitrided oxide with 4×10¹⁵ cm^-2 NIS dosage. Not alike to Fig. 2.6(a), pre-oxidation NIS will incorporate N into dielectric bulk rather than pile-up at interface. Figure 3.4(b) compares nitrogen distribution profiles with and without NIS. Without NIS, dielectric bulk is devoid of nitrogen. After post-oxidation NO annealing NIS nitrided oxides, more N will diffuse to interface, form Si3≡N bonding. Moreover, sample with 1×10¹⁵ cm^-2 NIS not only exhibits higher peak concentration at interface, but also has tighter N distribution than samples with lighter NIS, which is beneficial to obtain an uniformly distributed reliabilities. Figure 3.5 indicates the leakage current density of 22Å N - and NO-annealed NIS nitrided oxides with and without nitrogen implantation

under constant voltage stress (CVS) at -4V for 100sec. The major difference between fresh and stressed curves occurs near flat-band conditions, as described in Chapter 2.

Comparing to gate oxides without NIS, apparent SILC increment for low and medium dose NIS is exhibited. However, there is almost no leakage current increase for heavy NIS during voltage stress. The reasons for dosage-dependent SILC immunity will be explained below.

Figure 3.6(a) compares transient trapping behavior of 22Å NO-annealed NIS nitrided oxides during CCS. It should be noted all samples show an obvious hole trapping characteristics. For NIS smaller than 1×10¹⁴ cm^-2, pre-oxidation nitrogen implantation will enhance the hole trapping. As NIS increases to 1×10¹⁵ cm^-2, the hole trapping becomes negligible. The trap generation rate as a function of injected charges is seen in Fig. 3.6(b). Similar trend is examined. Eliminated trap generation rate appears only at heavy N implantation case, i.e. 1×10¹⁵ cm^-2. Figure 3.7(a) presents the time-dependent dielectric breakdown (TDDB) Weibull distribution of 22Å NO-annealed NIS nitrided oxides during CVS at -4.3V. Nitrided oxides with NIS less than 1×10¹⁴ cm^-2 reveal poorer tBD than the samples without NIS due to the larger hole generation rate. As consistent to Fig. 3.6, heavy NIS is expected to exhibit the highest t_BD. The Weibull slope is also dependent on the NIS dosage, which is plotted in Fig.

3.7(b). β shows a valley at the NIS dosage of 1×10¹³ cm^-2 from residual implantation damage and a peak at 1×10¹⁵ cm^-2 due to the tighten nitrogen distribution profiles in the dielectric bulk. As shown in Fig. 3.4(b), 1×10¹⁵ cm^-2 NIS will retard oxidation rate and generate uniform N distribution, higher β is expected for heavy NIS. Although thinner oxide thickness of sample with 1×10¹⁵ cm^-2 NIS would partially contribute to less SILC characteristics, only 3~4Å thickness reduction may not enhance SILC immunity significantly than samples with 1×10¹³ and 1×10¹⁴ cm^-2 NIS. As a result, the

drastic SILC immunity improvement is accredited to uniform nitrogen distribution in the dielectric bulk.

In our study, NIS dosage smaller than 1×10¹⁴ cm^-2 will slightly increase the oxidation rate due to insufficient post-oxidation annealing out the damage from ion bombardment, which is shown in Fig. 3.1. The residual damage may increase weak bonding during oxidation and result in less SILC immunity, enhanced trap generation rate and poor tBD. Inferior surface roughness will also response for degraded dielectric reliability, as seen in Fig. 3.8 and 3.9 for N₂- and NO-annealed NIS nitrided oxides, respectively. On the other hand, samples with 1×10¹⁵ cm^-2 NIS not only can use to grow multiple oxide thickness but also improve reliability significantly. The results are quite opposite to samples with lighter implantation dose, which can attribute to several reasons.

First, surface is abounded with N atoms for heavy NIS implantation dosage, which may easily replace the weak Si-O bonds damaged by ion bombardment in the transition region by the strong Si₃≡N bonds during oxidation. Secondly, implantation would change interface morphology [70]. From HRTEM photos exhibited in Fig. 3.8 and 3.9, surface roughness is increased while NIS increases from 0 to 1×10¹⁴ cm^-2, which may be due to faster oxidation rate and less N area density. On the other hand, surface RMS roughness is decreased to 0.87nm with further increasing NIS to 1×10¹⁵ cm^-2. It can be seen that the nitrided oxide/silicon interface is very smooth and exhibits a highly uniform transition region from the crystalline silicon to the amorphorous nitrided oxide. This is beneficial for the electrical properties of the MOS device as described above. Surface roughness is also evidenced by AFM images and summarized in Fig. 3.10. A smoother interface is helpful in reducing the localized field and, hence, results in stronger SILC immunity and larger t_BD. Thirdly, it is

known that SILC becomes negligible as the oxide thickness down to the direct-tunneling dominate regime. The heaviest NIS can retard oxidation rate significantly and grow the thinnest oxide, as seen in Fig. 3.2. In consequence, nitrided oxides with NIS dosage of 1×10¹⁵ cm^-2 possesses stronger SILC immunity and better TDDB characteristics than others, regardless of post-oxidation N₂- or NO-annealing.

3.4 Summary

The dielectric properties and reliability characteristics of NIS nitrided oxides are investigated in this chapter. An obvious oxidation rate retardation effect is observed for the nitrided oxides with nitrogen-implanted Si substrate. Dielectric property is strongly depended on NIS dosage and post-oxidation annealing temperature. NIS dosage less than 1×10¹⁴ cm^-2 is helpless to oxidation rate suppression accompanying with degraded dielectric reliability. On the contrary, the samples with 1×10¹⁵ cm^-2 NIS not only can use to grow multiple oxide thickness to meet SOC requirement, but also improve stress immunity apparently. Nitrogen implantation also generates a uniform distribution nitrogen profile in the dielectric bulk, which can be used as an effective diffusion barrier to resist boron penetration. NIS nitrided oxides could effectively suppress trap generation and improve time-to-breakdown and charge-to-breakdown, also suitable to reduce process steps of the SOC technology.

Table 3.1 Experimental conditions of nitrided gate oxides formed by pre-oxidation NIS and post-oxidation NO-annealing.

NIS dosage (cm^-2) Anneal Gas TOX (Å)

0 1 × 10¹³ 1 × 10¹⁴ 1 × 10¹⁵

30 √ √ √ √

22 √ √ √ √

NO anneal

19 √ N/A N/A N/A

30 √ √ √ √

22 √ √ √ √

N2 anneal

19 √ N/A N/A N/A

Table 3.2 Flat-band voltages and interface state densities of NO- and N2-annealed NIS nitrided oxides.

V_FB

NIS NO-GOX (V) N2-GOX (V) ∆VFB (mV)

0 -1.085 -1.055 -30

1 × 10¹³ cm^-2 -0.992 -0.958 -34 1 × 10¹⁴ cm^-2 -0.925 -0.896 -29 1 × 10¹⁵ cm^-2 -0.825 -0.816 -9 D_it

NIS NO-GOX (V) N2-GOX (V) ∆Dit (cm^-2 eV^-1) 0 7.06×10¹¹ 4.88×10¹¹ 2.18×10¹¹ 1 × 10¹³ cm^-2 8.82×10¹¹ 4.93×10¹¹ 3.89×10¹¹ 1 × 10¹⁴ cm^-2 9.11×10¹¹ 5.11×10¹¹ 4.00×10¹¹ 1 × 10¹⁵ cm^-2 2.45×10¹² 1.90×10¹² 5.50×10¹¹

Table 3.3 Thickness and dielectric constant variation for NO- and N2-annealed NIS nitrided oxides.

NIS Dosage QMCV TEM κ

0 26.3Å 27.3Å 4.01

1 × 10¹³ cm^-2 27.3Å 27.6Å 3.94

1 × 10¹⁴ cm^-2 26.6Å 27.2Å 3.99

N2 anneal

1 × 10¹⁵ cm^-2 20.2Å 23.3Å 4.50

0 24.9Å 25.8Å 4.20

1 × 10¹³ cm^-2 25.7Å 27.2Å 4.13

1 × 10¹⁴ cm^-2 25.2Å 26.8Å 4.15

NO anneal

1 × 10¹⁵ cm^-2 19.7Å 23.2Å 4.60

-2.0 -1.5 -1.0 -0.5 0.0 0.5

14 N₂-annealed NIS nitrided oxides

Capacitance ( pF )

14 NO-annealed NIS nitrided oxides

Capacitance ( pF )

Fig. 3.1 Plots of high frequency C-V curves of 22Å (a) N2-annealed (b) NO-annealed NIS nitrided oxides. Significant oxidation rate retardation is inspected as NIS dosage larger than 1×10¹⁴ cm^-2. As NIS dosage larger than 1×10¹⁵ cm^-2, clear V_FB shift and D_it increment is observed.

18

Open : N₂-annealing Close : NO-annealing

NIS Implantation Dosage ( cm^-2 ) QMCV

NIS Implantation Dosage ( cm^-2 ) N₂-annealing

NO-annealing

(b)

Fig. 3.2 (a) Oxide thickness (b) dielectric constant as a function of nitrogen dosage implanted into the silicon substrate before oxidation. Dielectric constant increases from 3.9 to 4.6 for NO-annealed NIS nitrided oxides with heavy implantation dosage.

0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0

N₂-annealed NIS nitrided oxides

Leakage Current Density ( A / cm2 )

Gate Voltage ( V )

Leakage Current Density ( A / cm2 )

Gate Voltage ( V )

Fig. 3.3 Direct tunneling leakage current density as a function of gate voltages of 22Å (a) N2-annealed (b) NO-annealed NIS nitrided oxides. Direct tunneling leakage current density is strongly dependent on the dosage of nitrogen implanted into Si substrate..

(a)

Nitrogen Concentration ( cm-3 )

Depth ( A ) w/o NIS

10¹⁴ NIS 10¹⁵ NIS

(b)

Fig. 3.4 (a) XPS depth profiles of 22Å NO-annealed nitrided oxides with 4×10¹⁵ cm^-2 NIS. Evidence of Si3N4-like interface is observed. (b) SIMS nitrogen depth profiles of 30Å NO-annealed NIS nitrided oxides. NIS incorporates uniform nitrogen distribution in the bulk. Atomic Concentration (%) O 1 s

N 1 s NO-annealed NIS nitrided oxides NIS = 4×10¹⁵ cm^-2

0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0

N₂-annealed NIS nitrided oxides CVS, -4V, 100sec

Leakage Current Density ( A / cm2 )

Gate Voltage ( V )

Leakage Current Density ( A / cm2 )

Gate Voltage ( V )

Fig. 3.5 SILC characteristics of 22Å (a) N2-annealed (b) NO-annealed NIS nitrided oxides at -4V CVS for 100sec. Maximum current increase after CVS occurs near flat band region. Increased SILC is negligible for 1×10¹⁵ cm^-2 NIS nitrided oxides.

0 20 40 60 80 100

Stressing Time ( sec ) (a)

Maximum Trap Generation Rate ( % )

Injected Charges ( C / cm² ) (b)

Fig. 3.6 (a) Transient hole-trapping behaviors (b) maximum trap generation rates of 22Å NO-annealed NIS nitrided oxides at -20 mA/cm² constant current stress. NIS nitrided oxides with 1×10¹⁵ cm^-2 dosage has negligible trap generation rate for both CVS and CCS.

1 10 100 1000 10000 -3

-2 -1 0 1 2 3

TDDB Weibull plots ( CVS -4.3V ) NO-annealed NIS nitrided oxides

ln ( -ln ( 1-F ) )

Time ( sec ) w/o NIS

10¹³ NIS 10¹⁴ NIS 10¹⁵ NIS

(a)

2.0 2.5 3.0 3.5 4.0

w/o NIS 10¹³ 10¹⁴ 10¹⁵ TDDB Weibull plots ( CVS -4.3V )

NO-annealed NIS nitrided oxides

Weibull Slope

NIS Implantation Dosage ( cm^-2 ) (b)

Fig. 3.7 (a) tBD Weibull distribution (b) Weibull slopes of 22Å NO-annealed NIS nitrided oxides under -4.3V constant voltage stress. The reliability of NIS nitrided oxides is substantially relied on NIS dosage, tBD and Weibull slope are increased only

Fig. 3.7 (a) tBD Weibull distribution (b) Weibull slopes of 22Å NO-annealed NIS nitrided oxides under -4.3V constant voltage stress. The reliability of NIS nitrided oxides is substantially relied on NIS dosage, tBD and Weibull slope are increased only

在文檔中 氮氧化層及高介電常數介電層在金氧半元件及快閃記憶體上之特性研究 (頁 43-0)