Material Analyses

The physical characterization of deposited La2O3 films consisted of the transmission electron microscopy (TEM) image for physical thickness determination, and the X-ray photoelectron microscopy (XPS) measurements for the composition scrutiny. And all of these material analysis techniques were briefly stated behind.

To begin with, the La2O3 physical thickness (cross section) of the fabricated samples was determined by the TEM image. In TEM, observation is made in ultrahigh vacuum, where an electron beam is focused onto the sample by electromagnetic lenses. Because the electron beam’s wavelength is less than that of visible spectra, the resolution of TEM is higher than that of the conventional optical type microscope. In this work, the deposited high-κ dielectric is prepared by using a focus ion beam (FIB) system with the model Nova 200 of FEI Company, and then it is transferred to JEOL JME-3000F TEM system for observing its thickness.

Next, the XPS system (Microlab 350, Thermal VG Scientific Company, England) was used to detect the composition of our deposited high-k dielectrics after annealing. The XPS, also known as the electron spectroscopy for chemical analysis (ESCA), is a quantitative spectroscopic technique to measure elemental composition, empirical formula, chemical state, and electronic state of elements existed within a material. Samples would irradiate with X-ray, and their emitted photoelectrons with kinetic energy (KE) are detected. The measured kinetic energy (KE) is given by

= − − _S

KE hv BE φ _(3-14)

, where hv is the photon energy, BE is the binding energy of the atomic orbital where electron generates, and φ_S is the spectrometer work function. The binding energy is the minimum energy to break the chemical bond inherent in each bond of the measured molecule.

Thus, the binding states could be identified by the positions of the binding energies where the peaks appear. In the case that the peak position is different from the expected positions, the chemical bond states are discussed the amount of shifting to the higher or the lower energy side.

A’

A Cap.

Bottom Electrode Definition Capacitor Hole Definition

Top Electrode Definition (Clear for lift-off process) Contact Hole Definition & High-k Etching

Al Pad Definition

(a)

Buffer Oxide Si Substrate

ILD1 TaN/Ni Top Electrode

ILD2 ILD2

ILD2

Al Al

TaN/Ta Bottom Electrode La

₂

O

₃

Film

(b)

Fig. 3-1. (a) The schematic layout of the La2O3 high-k MIM capacitors. (b) The cross-sectional structure along the A-A’ dashed line in the layout shown in (a).

100-nm Ta/20-nm TaN Bottom Electrode (BE)

Si Substrate PR

Buffer Oxide

(a) Bottom electrode deposition and patterning

Buffer Oxide Si Substrate

PR PR PR

200-nm PE-OX (ILD1) BE

(b) ILD1 deposition and capacitor hole region opening.

Buffer Oxide Si Substrate

ILD1 10-nm La₂O₃

BE

(c) High-k dielectric (HK) deposition and furnace annealing.

Buffer Oxide Si Substrate BE ILD1

PR

60-nm Ni/30-nm TaN Top Electrode (TE) PR

(d) Top electrode deposition and patterning by using the lift-off technique.

Buffer Oxide Si Substrate BE ILD1

TE PR

PR PR

300-nm PE-OX (ILD2)

(e) ILD2 deposition and contact hole opening.

Buffer Oxide Si Substrate BE ILD1

PR PR PR

TE HK

(f) HK etching.

Buffer Oxide Si Substrate

ILD1 TE

ILD2 ILD2

ILD2

500nm-Al

Al

BE

(g) Al deposition and patterning.

Fig. 3-2. The main fabrication steps of the MIM capacitor with the La2O3 high-k dielectric and the process steps corresponding to their used photo masks.

dQ

_inj

= ∫ J

_stress

(t

₁

)•dt

t₁+dt

t₁

t

0

t₁ t₁+dt

Q

_inj

(t) = ∫ J

_stress

(t)•dt

0 t

J

_stress

(t)

Fig. 3-3. The total injection charges density (Qinj) of MIM capacitor extracted from the area under the curve of the stress current density (Jstress) versus stressing time.

7.4 7.6 7.8 8.0 8.2 8.4

-3 -2 -1 0 1 2 3

Ln ( T

BD

)

scale factor (63% failure) α_C= exp(8.06)

Ln {- L n [1 -F

C

(t)]}

Slope = Shape factor β_C (Weibull slope)

8.06

Fig. 3-4. The shape factor βC (Weibull slope) and the scale factor αC extracted from the TDDB plot.

0 -1 -2 -3 -4 -10⁰

10² 10⁴ 10⁶ 10⁸

Time To 63% Failure (s ec)

5

Stress Voltage (V)

10 years

-4V A sec -3.5V B sec -3V C sec Stress

Voltage α_C

10-year Lifetime Operational Voltage

Fig. 3-5. The lifetime projection of MIM capacitors obtained from the failure plot with various stress voltages.

CHAPTER 4

Results and Discussion

This chapter falls into four main categories. Section 4.1 describes the basic properties and various physical analyses of La2O3 films, including the X-ray photoelectron spectroscopy (XPS) and the transmission electron microscopy (TEM). Secondly, section 4.2 studies the leakage current and the conduction mechanisms of the La2O3 MIM capacitor. Thirdly, considering for the accuracy of analog functions performed by the MIM capacitors, the relationships among the applied voltage, the temperature, and frequency on the capacitance of the La2O3 MIM capacitor are observed, corresponding to the quadratic voltage coefficient of capacitance (α), temperature coefficient of capacitance (TCC), and the frequency coefficient of capacitance (FCC), respectively. Next, as for the stability on the La2O3 MIM capacitors in practical use, section 4.4 investigates the stress induced variation on the capacitance density, the quadratic voltage coefficient of capacitance (α), the temperature coefficient of capacitance (TCC), and the leakage current density. Finally, the inquiries into breakdown and reliability characteristics of the La2O3 MIM capacitor are stated in section 4.5, such as the time-zero dielectric breakdown (TZDB) and the time-dependent dielectric breakdown (TDDB), respectively.

4.1 B

ASIC

P

ROPERTIES OF THE

E-

BEAM

E

VAPORATED

L

A₂

O

3

D

IELECTRIC

F

ILMS OF

MIM C

APACITORS

The chemical composition of the e-beam evaporated La2O3 film after 400-°C furnace annealing in O2 ambient for 10 minutes is determined by X-ray photoelectron spectroscopy (XPS) analysis. The La 3d and O 1s core level spectral regions are detected and shown in Figs.

4-1(b) and 4-1(c), respectively. Considering the area integrated from the peak spectral to the binging energy, the atomic concentration ratio of lanthanum atom to oxygen atom could be determined about 2: 3. The La 3d signals of the La2O3 film consisted of the energy splitting of the 3d3/2 and the 3d5/2 spin-orbit doublets. The main La 3d XPS peak is centered at 856.7 eV, and its spin-orbit component is separated at 836.1 eV. The binding energy and the spin-orbit component associated with the present La2O3 features are in agreement with the XPS reference book [122]. Besides, the XPS peak of O 1s core level spectral, as presented in the Fig. 4-2(c), at binding energy of 533.1 eV could be regarded as the La-O bonding. A broad signal existed at lower binding energy of 530 eV due to the overlap of various components is associated with oxide and hydroxide on the La2O3 film surface.

The thickness of the deposited La2O3 film on the bottom electrode is decided by the cross-sectional transmission electron microscope (TEM) image of the fabricated MIM structure, as indicated in the bottom inset of Fig. 4-2. It could be found that the physical thickness of the La2O3 film is about 10 nm after being annealed in oxygen ambient. Fig. 4-2 shows the typical capacitance-voltage (C-V) characteristic of the La2O3 MIM capacitor at 100 kHz from −2 V to 2 V. The capacitance density of the 10-nm La2O3 MIM capacitor measured at the zero bias is 11.4 fF/µm². As a result, the effective dielectric constant (kLa2O3) value and the equivalent-oxide thickness (EOT) derived by the Eq.3-1 and Eq. 3-2 are 12.9 and 3 nm, respectively. Besides, an obvious interfacial layer (IL) between dielectric and bottom electrode is observed. This unavoidable IL might be formed as the high-k layer directly

contacts the metal electrode.

The calculated kLa2O3 value is 12.9 in this work, which is smaller than that in the other reports [22], [34], [36], [37]. The main reason for this low kLa2O3 value might be the low-temperature post deposition annealing [40], [123]. As the paper reported in [40], the dielectric constant of the La2O3 film deposited by the atomic layer deposition (ALD) at 300

°C is 9. When as-grown ALD La2O3 film is subjected to anneal from 400 °C to 500 °C, its dielectric constant could be raised from 12.5 to 17.3 since the improvement in its crystallinity and densification. In this study, the La2O3 MIM capacitor for RF/analog applications is located in the interconnection levels and above the active integrated circuit layers. Although the increase in annealing temperature could raise the dielectric constant of the La2O3 film, the maximum process temperature is limited to 400 °C to fulfill the thermal budget requirement of the backend process. Despite that the dielectric constant of La2O3 film in this research is 12.9 annealed in a low temperature of 400 °C, the capacitance density with the magnitude of 11.4-fF/µm² is sufficient for RF applications until 2018, as mentioned in chapter 1.

4.2 L

EAKAGE

C

URRENT AND

C

ONDUCTION

M

ECHANISMS OF

L

A₂

O

₃

MIM C

APACITORS

In spite of possessing higher capacitance density in the same physical thickness, the lager leakage current of the high-k La2O3 MIM capacitor is higher than that of the SiO2 one, which could be ascribed to the smaller conduction band gap offset (ΔECB) of La2O3 dielectric. Since the leakage current issue limits the aggressive scaling of dielectric thickness which has an impact on the capacitance as depicted in Eq. 3-1, the technological evolution of high-k MIM capacitors requires a lower leakage current to enhance the capacitance density and reduce the power consumption. The capacitor fabricated by dielectric material with a large energy band gap and a high dielectric constant value could obtain a low leakage current and a large

capacitance density, respectively. However, the dielectric constant and bandgap for high-κ dielectric materials are usually trade-off. Hence, we choose the La2O3 dielectric with the largest ΔECB of 2 eV, compared to the other reports, as our insulator in MIM capacitor. In this section the leakage currents at measurement temperatures varied from 25 °C to 125 °C were obtained and current transport mechanisms of the La2O3 MIM capacitor were studied.

Fig. 4-3 sketches the energy band diagram of the Ni/La2O3/TaN MIM capacitor. The ideal work-functions of the top electrode (Ni) and the bottom electrode (TaN) are about 5 eV and 4.6 eV, respectively, and the electron affinity of La2O3 dielectric is 1.7 eV. The top electrode is applied voltage and the bottom electrode is grounded during measurement. When the positive voltage is applied on the top electrode, the electron carriers inject from the bottom electrode to the insulator named as the bottom injection, whereas the top injection condition is when negative voltage is applied. Fig. 4-4 illustrates the leakage current density of the 10-nm La2O3 MIM capacitor measured at 25 °C, where the leakage current at −1V and +1 V were 9.4 and 45 nA/cm², respectively, even if the biased voltage is −6 V that the corresponding electrical field applied on the MIM capacitor is up to 6 MV/cm, the leakage current density is still below 10⁻⁵ A/cm², and the features are superior to any other reported data of the MIM capacitors with La2O3 high-k dielectrics.

Fig 4-5 shows the leakage current of 10-nm La2O3 MIM capacitor measured at a negative bias on nickel electrode from 25 °C to 125 °C. It could be found that every curve has two transition regions, where the leakage current density with the increase of the temperature.

The leakage current density is kept below 100 nA/cm² in the first region, whereas it is hugely increased in the second region. These phenomena represent various conduction mechanisms of carriers (electrons) depending on the electric field and the temperature. Therefore, we use the Poole-Frenkle emission and the Schottky emission to verify the conduction mechanisms of La2O3 MIM capacitor in the high field region and in the low field region, respectively. Fig.

4-6 plots the natural logarithm of the leakage current density divided by electric field ln(J/E)

versus the square root of electric field E^1/2 in the high field region according to the Eq. 2-5.

Regarding to various measurement temperatures, a well linear correlation between ln(J/E) and E^1/2 could be obtained in the high field region. On account of the slopes of these fitted straight lines, the extracted εr value is largely increased with the increasing temperature. The reason for large increase of the extracted εr value can be contributed to unsuitable fitting region. The straight line could explain this conduction mechanism belongs to Poole-Frenkle emission, but the extracted εr value and the effective trapping level energy are not good agreement. Next, we take the other method to extract εr value. The extracted slope in Fig, 4-7 is chosen to re-plot the slope versus (qE/π)^1/2 curve, as shown in the Figs. 4-7(a) and 4-7(b). The best fitting region on applied voltage could be determined by negative slope in these plots.

Therefore, the εr value and the effective trapping level energy could be extracted as 3.77~3.53 and 1.15~1.50 eV, respectively. The extracted εr value is close to the dynamic dielectric constant of La2O3 at optical frequencies, i.e., the square of its refractive index (~1.87~1.93).

To study the leakage mechanism in low field region, the ln(J) is arbitrarily plotted as a function of E^1/2 to verified whether Schottky emission or not. Fig. 4-8 shows the typical plotting of ln(J) versus E^1/2, and the fitted straight lines with various temperatures. The extracted εr value is increased with the increasing temperature and these curves appear the linearity in the low electric field. Further, a linear relationship between ln(J/T²) and q/kT also should be displayed in terms of Schottky emission. Therefore, we also plot this linear relationship from the applied voltage from −0.6V to −1.6V in Fig. 4-9. It indicates the data could be fitted well by a straight line at various electric fields, which many research conclude this leakage mechanism is the Schottky emission, but it seems not always true. As the same method in the PF emission extraction, we also plot the extracted slope versus the E^1/2 in the inset of Fig. 4-9. According to this inset, however, the slope in the inset is negative, not compatible with the positive square root term in the extracted slope from Fig. 4-9. Therefore, the Schottky emission mechanism for this 10-nm La2O3 MIM capacitor in low electric field is

not valid. In fact, the Schottky emission mechanism is electrode limitation and contributed by the carriers that overcome the barrier between the electrode and the insulator. It has been verified in many researches for metal-insulator-silicon (MIS) capacitor with ultra-thin high-k dielectric without considering defects inside the bulk film. If there are more defects in the thick high-k dielectric, especially on processing at low temperature of 400 °C in this work, the traps assisted tunneling at high temperature would become more obvious and complex in low field region. As mentioned in chapter 1, the high-k dielectrics have high trap density in themselves and they have lots of interface states in their bandgap. Moreover, the deposited films have huge amounts of defects compared to the thermally grown films, especially for the e-beam evaporated films. Therefore, we may reasonably suppose that a trap-related conduction also affect the leakage current of high-k La2O3 MIM capacitors in the low field region.

In summary, it could be concluded that the conduction mechanism of 10-nm La2O3 MIM capacitors is dominated by the Poole-Frenkel emission at high applied voltage region. On the other hand, at low field region, the Schottky emission is not a unique conduction mechanism, another trap-related conduction also influences the conduction behavior of MIM capacitors at low bias, due to the trap-rich characteristic of La2O3 films.

4.3 E

FFECTS OF

V

OLTAGE

, T

EMPERATURE

,

AND

F

REQUENCY ON

T

HE

C

APACITANCE OF

L

A₂

O

3

MIM C

APACITORS

The performance of RF circuit would be ultimately limited by the accuracy of its passive components. It is also known that the capacitance of the MIM capacitor with high-k dielectric would vary with the operational voltage, the temperature, and the frequency to lead to distortion in analog signals for RF application [6]. However, the physical mechanisms of these influences upon capacitance are still unclear and they are necessary to investigate. In

this section, accordingly, we evaluate the relationships among the applied voltage, the temperature, and the frequency on the capacitance of the La2O3 MIM capacitor, corresponding to the quadratic voltage coefficient of capacitance (α), temperature coefficient of capacitance (TCC), and the frequency coefficient of capacitance (FCC), respectively.

4.3-1 Characteristics of Voltage Coefficient of Capacitance

Fig. 4-10 shows the C-V curves of the 10-nm La2O3 MIM capacitor measured at frequencies varied from 10 kHz to 500 kHz and at the temperature of 25 °C. As can be seen, the capacitance density rises with the increase of applied voltage and it expresses a parabolic curve with a positive curvature. This reason on the increase of capacitance might be attributed to the high degree of electric field polarization and carrier injection [6], [23], [103]. When the voltage applied on the electrode sweeps from zero to a high voltage level, some of injection carriers would be captured by interface trap states existing in the dielectric near the injection electrode. Then, these trapped charges could induce internal dipoles to follow the alternating signals with a dipole relaxation time. Besides, the other excess mobile charges in the insulator also follow the small ac signals with a space charge relaxation time. Consequently, both dipole relaxation and space charge relaxation behaviors could modulate the capacitance, resulting in the capacitance fluctuation (ΔC=C(V)–C(V=0)) with the varied applied voltages.

This phenomenon, named as voltage nonlinearity or voltage dispersion of capacitance, could be depicted by the voltage coefficient of capacitance (VCC).

The VCC fitting curves and the extracted α values according to the Eq. 3-3 are indicated in Fig. 4-11(a), under the top injection condition from 10 kHz to 500 kHz at 25 °C. The quadratic voltage coefficient of capacitance (α) decreases from 775 to 595 ppm/V² as the frequency increases from 10 kHz to 500 kHz. The reason for the lower VCC with the frequency increasing is that the trapped charges induced dipoles and the excess mobile

charges hardly follow the ac signal with higher frequency, corresponding to the higher relaxation times of these dipoles and mobile charges [24], [103], [124]. On the other hand, in Fig 4-11(b), the α value under the top injection is significant smaller than that under the bottom injection with the same measurement frequency. This asymmetry on the quadratic voltage coefficient of capacitance (α) of the top injection and bottom injection results from the difference in workfunction between top (Ni, 5 eV) and bottom (TaN, 4.6 eV) electrodes.

Since the top electrode (Ni) with high workfunction has a larger barrier height to suppress the charges injection, the amount of trapped charge induced dipoles and excess mobile carriers in the La2O3 dielectric could be reduction to improve the voltage linearity. This experimental result further confirms that the charge injection effect at metal/dielectric interface dominates the voltage dispersion of capacitance rather than the bulk effect, such as a field dependent polarization, which is consistent with observations in other works.

In addition, the α value increases with the rise in temperature, as illustrated in Fig. 4-12.

As the temperature elevates, more energetic carriers could easier inject from the electrode into the dielectric, and further the field enhanced electron hopping process in the dielectric is more remarkable. It could be expected that the higher charge in trapping/detrapping rate of dielectric would increase the quantity of mobile charges to affect the voltage coefficient of capacitance (VCC) at high temperature. As a result, the variation in capacitance becomes more sensitive to alternating voltages at high temperature and then causes the degradation in voltage linearity, especially in the lower frequency measurement condition. Moreover, the magnified distinction in α among various frequencies at elevating temperature might be attributed to the reduction in mobility of injection carriers [103]. The injection charges caused the VCC variation are classified two types, trapped charges induced dipoles and excess mobile charges. However, the trapped charges induced dipoles are more energetic and easier to follow high frequency alternating signals at high temperature. Therefore, these dipoles are

not the main reason to cause the magnified distinction in α. Although the larger amount of excess mobile charges would be generated inside the insulator to cause large α value at high temperature, the fraction of total excess mobile charges with enough high mobility, which could follow high frequency signal at higher temperature, is less than that at low temperature.

Accordingly, the deviation in α measured at various frequencies at high temperature is larger than that at low temperature. This result clarifies the significance of interface charge injection on voltage dependency of capacitance for La2O3 MIM capacitors again.

Accordingly, the deviation in α measured at various frequencies at high temperature is larger than that at low temperature. This result clarifies the significance of interface charge injection on voltage dependency of capacitance for La2O3 MIM capacitors again.

在文檔中 應用於射頻/類比電路的三氧化二鑭金屬-絕緣層-金屬電容之特性研究 (頁 55-0)