Chapter 2 Literature Review
2.6 Post-integration porogen removal scheme
The novel post-integration porogen removal approach, which detail integration scheme is illustrated in Figure 2.21, has been proposed based on material design and integration scheme to circumvent the reliability issues of porous dielectrics mentioned previous section. Malhouitre et al. [112] and Fayolle et al. [113] have demonstrated that such novel material and scheme could be achieved in a copper single damascene structure with a 25% RC reduction after porogen removal. Recently, post-integration porogen removal approach has been applied to reduce the plasma-induced damage to porous low-k dielectrics because porogen can reduce effective pore size and limit the plasma radical diffusion inside pore [14,15]. For such a scheme, the decomposition temperature (Td) of high-temperature porogen candidates needs to be higher than the maximum processing temperature (> 350oC) with the post-integration porogen removal scheme [114]. Moreover, the mechanical strength of hybrid low-k material should be strong enough to pass all of the backend processing steps, especially the chemical-mechanical polish (CMP) step. On the other hand, for such hybrid dielectrics or its porous form, there are additional concerns relate to moisture uptake induced by
processing steps such as CMP, post-etch cleaning, post-CMP cleaning or resist removal [115,116,117]. Therefore, the structure-property relationship of hybrid low-k materials using in the post-integration porogen removal scheme will be further studied in this dissertation.
The thermal stability of the porogens used in the as-deposited porous low-k film are typically not so high such that porogens can be removed readily to form porosity during dielectric deposition or subsequent thermal treatment. As for the high-temperature porogens in the late porogen removal scheme, it is desirable to attain a high and sharp decomposition temperature by incorporating stable structures such as aromatic rings or double/triple bonds as well as by using similar moieties in a block copolymer. For example, polystyrene-b-polybutadiene-b-polystyrene (SBS) and polystyrene (PS) could be chosen as a good high-temperature porogen candidate due to its similar aromatic structure which can provide a good thermal stability and desired decomposition temperature. The study focus on the porogen surface modification by using ionic surfactant, and the thermal stability of surfactant is required to consider.
Material Dielectric constant Deposition method
Low-k materials
Polyimides 3.0-3.6 Spin-on
DLC (diamond-like carbon) 2.8-3.0 CVD
Hydrogen silsesquioxane (HSQ) 2.8-3.0 Spin-on
Black Diamond TM (SiCOH) 2.7-3.3 CVD
Spin-on glasses 2.7-3.1 Spin-on
Polyimide-SSQ hybrids 2.7-3.0 Spin-on
SILKTM 2.7 Spin-on
Fluorinated polyimides 2.6-2.9 Spin-on
Poly(arylene ethers) 2.6-2.9 Spin-on
Methyl silsesquioxane (MSQ) 2.6-2.9 Spin-on
Poly(arylenes) 2.6-2.8 Spin-on
Poly(norbornenes) 2.5-2.7 Spin-on
Fluorinated Parylene 2.5 CVD
Porous dielectrics
Polyimide nanofoams 2.2 Spin-on
Mesoporous silica 1.9-2.2 Spin-on
Silica xerogels 1.5-2.2 Spin-on
Silica aerogels 1.1-2.2 Spin-on
(Source:IBM)
Table 2.1 Type of ILD materials
Property MSQ HSQ SiO2
Dielectric Constant 2.8 3.0 4.0
Modulus (GPa) 3-5 6 59
Density (g/cm3) 1.2-1.3 1.4-1.5 2.4
Tensile Strength (MPa) 50 80 -
Table 2.2 Principle properties of SSQ based dielectric materials
Material Trade Name k-value Company
HSQ FOx (flowable oxide) 2.9-3.0 Dow Corning
MSQ RZ25-15 2.6 Hitachi
MSQ HOSP 2.6 Honeywell
Porous HSQ XLK 2.2 Dow Corning
Porous HSQ LKD 5109 2.2-2.3 JSR
Porous MSQ Zirkon 2.3 Shipley
Table 2.3 Commercially available SSQ-based low-k materials.
Method Mixing-porogen Reactive-porogen
10 wt% PCL 5.3 5.2
20 wt% PCL 14.2 10.0
30 wt% PCL 20.0 15.1
40 wt% PCL -- 17.1
Table 2.4 The pore size (nm) from PCL porogen under different low-k process
Matrix Porogen
Table 2.5 The property of reactive-porogen on SSQ-based low-k materials
Matrix Porogen
Table 2.6 The impact of porosity change on SSQ-based low-k materials.
Figure 2.1 Industry average “Moore’s law” and chip size trends.
Figure 2.2 Device scaling projection trend shown in terms of gate length (half pitch) for Flash, DRAM, and MPU/ ASIC microelectronics products.
Figure 2.3 Typical schematic interconnect cross-section with parasitic capacitance Top Metal Layer
Bottom Metal Layer CV
CL
W
tm
Interconnect Layer
Figure 2.4 Decrease in interconnect delay and improved performance are achieved by using Cu and low-k dielectric.
Figure 2.5 Historical transition of ITRS low-k roadmap
Figure 2.6 Basic structure of FSG matrix.
O
Si O
Si
Si
Si Si
Si O
O
O O
O F
F
Figure 2.7 Depiction of possible bond rearrangements upon SiOF film hydration to produce Si−OH bonding and the release of HF from the film.
Si
O O
O Si
O
O O
O H
H
F
Figure 2.8 Carbon-doped silica glass and schematic bonding structure
Si
Figure 2.9 The typical precursors of CDO materials.
Figure 2.10 Organic SiLKTM chemical structure units.
Figure 2.11 Basic structure units of SSQ dielectric materials consist of random, ladder and cage structures. R= H, CH3 for HSQ and MSQ respectively.
0 20 40 60 80 100
Figure 2.12 Bruggeman’s effective medium approximation showing dielectric constant versus porosity for oxide and a low-k material.
Figure 2.13 Schematics of the nano clustering silica (NSC) formulation and its film deposition and curing processes.
Figure 2.14 The conventional formation of porous low-k by using template-type porogen method.
Figure 2.15 Schematic structures of phase behaviors of PEO-b-PPO-b-PEO triblock copolymers in MSQ matrixes.
Fig. 2.16 Procedure for preparation of a nanoporous organosilicate dielectric thin film from a curable polymethylsilsesquioxane precursor matrix and a thermally labile globular dendrimer porogen (EA-PPI-64 or EA-PPI-128).
Fig. 2.17 Star-shape polymer porogens: PCL4, four-armed poly(ecaprolactone) and PCL6, six-armed poly(e-caprolactone).
Figure 2.18 β-Cyclodextrin (β-CD), a cage supramolecular porogen
(a)
(b) Figure 2.19 The synthetic scheme: (a) sol-gel reaction and (b) ATRP method to prepare
Figure 2.20 The energy and the distance of relationship for two particles.
Figure 2.21 The integration scheme of the post-integration porogen removal scheme.
Chapter 3 Experimental
This chapter describes the preparation of MSQ/porogen solution, hybrid and porous low-k materials including the materials candidates, deposition processes, and thermal treatment. The approaches used to characterize the meso-/nano- porous dielectric films are also introduced in this chapter.
3.1 Materials candidates
3.1.1 MSQ/porogen hybrid materials
The low-k matrix, poly(methylsilsesquioxane) (MSQ), was obtained from Gelest Inc. (Morrisville, PA; CAS No.68554-70-1) as clear white flakes. The molecular-weight distribution of MSQ is from 6000 to 9000 g/mole, and its dielectric constant is about 2.9.
High-temperature porogens, PS-b-PB-b-PS (SBS Mw ~ 90,000 g/mole; 28 wt% PS) and PS (Mw ~ 790 g/mole) were both obtained from Sigma-Aldrich Co. The chemical structures of MSQ, and porogen are illustrated in Figure 3.1 (a) through (c), respectively.
3.1.2 Porogen surface modification
For porogen surface modification, we choose two kind of ionic surfactant to modify the porogen. Dodecylbenzensulfonic acid sodium salt (C12H25C6H4SO3Na, NaDBS, Mw=348.48 g/mole), an anionic surfactant, was procured from Showa Chemical Industry Co., Ltd. On the other hand, a cationic surfactant Domiphen Bromide CH3(CH2)11N(CH3)2(CH2CH2OC6H5)Br, DB, Mw=414.48 g/mole), was obtained from Sigma-Aldrich Co. The chemical structures of ionic surfactant are illustrated in Figure 3.2.
3.2 Sample preparation
The study divided into three steps, including (1) the porogen activity in the thermal profile (different curing rate, 2oC/min vs. 200oC/min), (2) the activity of the modified and ummodified porogen under 2oC/min slow curing, and (3) the properties of porous film at high porosity without and with modification.
3.2.1 The in-situ porogen activity test
To form hybrid low-k films, MSQ and SBS at 5 wt% loading were dissolved in tetrahydrofuran (THF) to form a hybrid low-k solution without special treatment such as surfactants to further minimize the porogen size. The solution was initially filtered
through a 0.20 µm PTFE filter (Millipore Inc.), and then spun onto (100) silicon wafer at 2000 rpm for 30 s at room temperature to obtain a film thickness of 500 nm.
The size of porogen in the hybrid low-k films were then characterized by in-situ grazing incidence small angle X-ray scattering (GISAXS) under two curing profiles: (1) slow curing at 2oC/min and (2) rapid curing at approximately 200oC/min. In-situ 2D GISAXS data were collected from 30oC to 210oC in intervals of 20oC, in which data collection time are 5 sec and 2 sec for slow and rapid curing profile, using beamline 23A of the National Synchrotron Radiation Research Center (NSRRC), Hsinchu, Taiwan. All of the GISAXS data were obtained using an area detector that covered a q range from 0.01 to 0.1 Å-1, and the incident angle of the X-ray beam (0.5 mm diameter) was fixed at 0.2° with an X-ray energy of 10 KeV. Then, the pore size was analyzed by sphere model fitting and Guinier’s law [118,119].
The interaction between MSQ and SBS was examined by an in-situ viscosity measurement using ARES (Rheometric Scientific). The viscosity data were collected from room temperature to 200oC (1) at a slow curing rate (2oC/min), and (2) isothermally at 200oC. The structural change and the degree of cross-linking of MSQ in the MSQ/porogen hybrid films were investigated as a function of temperature ranging from 25 to 200oC by Fourier-transform infrared spectroscopy (FT-IR) using a MAGNA-IR 460 (Nicolet Inc.) in a transmission mode with 64 scans at a spectral
resolution of 2 cm-1.
The elastic modulus (E) of hybrid film was measured using a nanoindenter (MTS, Nano Indenter XP system) with a Berkovich tip. The samples were cured up to 200oC following two different curing profiles to form hybrid low-k films. In the slow curing method, the sample was cured in a quartz tube furnace under N2 at a curing rate of 2oC /min to 200oC for 30 min. In the rapid curing method, the samples were cured for 30 min directly on the hot plate preheated at 200oC. The nominal film thickness of hybrid films for nanoindentation is 1200 nm, while the indentation depth is kept at 100 nm.
3.2.2 Porogen modification
PS particles were first dissolved in tetrahydrofuran (THF) to form a solution without modification (pH ~7.0). In addition, pH value and surfactants were used to modify the surface property of the PS particles. For pH effect, PS solutions with pH values of 3 and 11 were prepared by adding acid and base, respectively. Moreover, the well-dispersed PS/THF solutions were modified by two types of surfactants: (1) anionic surfactant, NaDBS, below its critical micelle concentration (CMC) of 522.75 mg/L, and (2) cationic surfactant, DB below its CMC of 730.74 mg/L [120]. The zeta potentials of the PS/THF solutions with and without modification were measured using a Zeta Potential Analyzer (Zetasizer HSA3000, Malvern Instruments). The size of the PS
particles in the THF solution was measured using an Ultrafine Particle Analyzer (Honeywell UPA 150).
Then, MSQ and PS particles (with and without surface modification) at 10 wt%
loading were dissolved in THF solvent to form a hybrid low-k solution. The solution was initially filtered through a 0.20 µm PTFE filter (Millipore Inc.), and then spun onto a (100) silicon wafer at 2000 rpm for 30 s at room temperature to obtain a film thickness of 500 nm. The size and distribution of the porogen in the hybrid low-k films during the curing step at 2oC/min were then characterized by in situ GISAXS. In situ 2D GISAXS data were collected from 30 to 200oC at intervals of 10oC. All of the GISAXS data were obtained using an area detector covering a q range from 0.01 to 0.1 Å−1, and the incident angle of the X-ray beam (0.5 mm diameter) was fixed at 0.2° with an x-ray energy of 10 keV. Then, the porogen size was analyzed using sphere-model fitting and Guinier’s law [118,119].
The interaction between MSQ and PS was then examined by an in situ viscosity measurement using ARES (Rheometric Scientific). The viscosity data were collected from room temperature to 200oC for the MSQ/PS hybrid films with and without modification by surfactants. The interaction between MSQ and surfactant-modified PS was further investigated using a FTIR spectrometer, MAGNA-IR 460 (Nicolet Inc.) in a transmission mode with 64 scans at a spectral resolution of 2 cm−1.
Finally, the hybrid films (with and without modification) were cured in a quartz tube furnace under N2 at a heating rate of 2oC/min to 400oC for 1 hr to form the porous low-k films after completely burning out the porogens. Their pore sizes were also characterized using the GISAXS technique. The porosity of the porous low-k films were obtained from its density, which was measured by X-ray reflectivity (XRR) (Bruker D8 Discover), with a Cu Kα source (λ=0.154 nm), using ω-2θ scan mode. The scanning region ranged from 0 to 2°. The XRR data was analyzed by LEPTOS simulation software, to fit the density of porous low-k film.
3.2.3 Different porosity low-k film with and without modification
First, PS particles were dissolved in THF to form a well-dispersed solution using different treatments: (1) without modification, (2) pH modification at a value ranging from 3 to 11, and (3) with a cationic surfactant, DB below its critical micelle concentration 730.74mg/L. The zeta potentials of PS/THF solutions with and without modification were measured using a Zeta Potential Analyzer (Zetasizer HSA3000, Malvern Instruments). The size of the PS particles in THF solution was measured using an Ultrafine Particle Analyzer (Honeywell UPA 150).
Then, MSQ and PS (with and without surface modifications) of 10, 20, 30, 40,
and 50 wt % loading were dissolved in THF solvent to form a hybrid low-k solution.
The solution was initially filtered through a 0.20 µm PTFE filter (Millipore Inc.), and then spun onto a (100) silicon wafer at 2000 rpm for 30 s at room temperature to obtain a film thickness of 500 nm. The sample was cured in a quartz tube furnace under N2 at a curing rate of 2oC/min to 400oC for 60 min to remove the porogen.
3.2.3.1 Porosity
All of the film density in the porous low-k films was measured by X-ray reflectivity (XRR). The porosity of the porous low-k films were obtained from its density, which was measured by X-ray reflectivity (XRR) (Bruker D8 Discover), with a Cu Kα source (λ=0.154 nm), using ω-2θmethod with scan angle from 0° to 2°). The XRR data were analyzed by LEPTOS simulation software using a genetic algorithm model [121]. Film thickness were also confirmed by using an n&k Analyzer 1280 (n&k Technology, Inc.) at wavelengths ranging from 190 to 900 nm.
3.2.3.2 Pore size and porogen size
The pore size in the porous low-k films were then characterized by 2D GISAXS.
The pore morphology of porous MSQ films was examined by FIB-SEM (Nova Nanolab 2000 system, FEI Company). The pore morphology of porous low-k films was
examined by a focus ion beam/ scanning electron microscope (FIB/SEM) dual-beam system (FEI Nova 200) after sputtering etch ~40 nm off the surface of the low-k films.
Moreover, the pore size and porogen size were further obtained by GISAXS. GISAXS was a versatile tool for characterizing nanoscale density correlations and/or the shape of nanoscopic objects at surfaces, at buried interfaces, or in thin films. The incidence beam extracted from a super-conducting wavelength-shifter X-ray source, was monochromated to a wavelength (λ) of 1.55 Å by a Ge (111) double crystal monochromator, with Δλ/λ~10-3 resolution. The two dimensional image were recorded by a low-noise 16-bit charge-coupled device (CCD) camera. All GISAXS data were corrected for sample transmission, background, and the detector sensitivity. Then the pore size was analyzed by Guinier’s law. A typical geometry of GISAXS measurement was depicted in Figure 3.3. The area detector records the scattering intensity of scattered rays over a range of exit angles () and scattering angles () in the surface plane [122]
3.2.3.3 Electrical characteristics
The k was characterized by capacitance-voltage (C-V HP 4280) measurement, using metal insulator semiconductor (MIS) structure configuration [Al electrode/MSQ film/Si (50 ohm-cm)] at room temperature. To accurately measure the dielectric constant by C–V dot measurement, three circular aluminum dots of nominal diameters
200, 400, and 800 mm were used to minimize the geometric effect. Aluminum electrodes with a thickness of 1m were coated onto the dielectric films by ULVAC EBX-6D thermal evaporator through a shadow mask. Prior to CV-dot measurement, the samples were kept in a small vacuum chamber which was pumped to 10-2 torr. The dielectric constant (k) of the films was determined by the following Equation (3.1)
0
d A
k C (3.1)
where C is the capacitance of the MIS structure, d is the in film thickness, A is the area of the aluminum dot, ε0 is the permittivity of free space (8.85410-12 F/m)
3.2.3.4 Chemical characteristics
The structural transformation and MSQ’s ratio of network/cage in the matrix were studied using Fourier-transform infrared spectroscopy (FT-IR). The measurements were performed using a MAGNA-IR 460 (Nicolet Inc., Waltham, MA) in specular mode with 30º incident angle, and 64 scans at 2 cm-1
3.2.3.5 Mechanical strength
The elastic modulus of the low-k films were measured using a Nanoindenter (MTS,
Nano Indenter XP system) with a diamond tip with Berkovich geometry. A holding segment was inserted in the end of each loaded segment to allow time for the system to equilibrate before it was unloaded. Then, unloading was performed at a constant rate.
The early unloaded portion was used to calculate the stiffness, and the holding segment was used to correct the thermal drift. The indentation depth was 100 nm, but the elastic modulus was determined from the first 10% of the film thickness to avoid the substrate effect. About ten identical indents were made in each test to ensure reproducibility and accuracy.
(a)
(b)
(c)
Figure 3.1 The molecular structure of (a) MSQ as the low-k matrix; (b) PS-b-PB-b-PS and (c) PS as the high-temperature porogens.
(a)
(b) Figure 3.2 The molecular structure of surfactant: (a) NaDBs, and (b) DB.
Figure 3.3 Typical geometry of GISAXS measurement.
Chapter 4
Effect of Curing on the Porogen Size in the Low-k MSQ/SBS Hybrid Films
In this chapter, the interaction between porogen and low-k matrix during curing and their impact on the porogen size of low-k/porogen hybrid films will be discussed. A commercial, spin-on organosilicate, MSQ was selected as the matrix and a high-temperature porogen such as SBS, was employed as the sacrificial component. The effects of curing rate (slow: 2 oC/min vs. rapid: 200 oC/min) and cure temperature on the porogen size in the hybrid low-k films cured up to 200 oC were studied by in-situ GISAXS, viscosity measurement, and FT-IR analysis. The impact of the MSQ structure during thermal cure process on the porogen aggregation behavior and its correlation with porogen size will be described and elucidated.
4.1 In-situ porogen size test
The porogen size of the hybrid low-k films before burning out were first characterized by in-situ GISAXS. Figures 4.1(a) and (b) show the 2D GISAXS scattering patterns of the hybrid low-k films for various curing temperatures (50-210oC) at two different curing rates. In Figure 4.1(a), the scattering patterns of the hybrid low-k
film cured at slow curing rate showed significant transition between 90 and 170oC. In contrast, the scattering patterns of low-k films cured at a rapid curing rate exhibited little variation as illustrated in Figure 4.1(b). These scattering patterns indicate that the SBS porogens showed much diffusion and aggregation activities within MSQ matrix when the hybrid film was cured at a slow curing rate. In contrast, the diffusion and aggregation of SBS within MSQ matrix is very limited when the hybrid low-k film was cured at a rapid rate.
From GISAXS scattering patterns, the porogen sizes in the MSQ/SBS films cured under different rates at temperatures between 50 and 210oC can be further determined.
In this paper, a particulate system was used to analyze the scattering data by treating the pores in the matrix the same as the particles in the hybrid film. When the particle concentration is low, the scattering of the pores in the matrix does not mutually interact.
Accordingly, the scattering intensity of individual particle, I(q), could be defined by Equation (4.1), where the wavevector q = 4-1sin is defined by the wavelength and the scattering angle 2 of X-rays; np is the number density of particles; ρ p and ρ m are the scattering density of the particle and the matrix; Vp denotes the volume of particle; P(q) is a form factor, and S(q) is a structure factor. The structure factor S(q) is close to 1 in a
low-concentration system and thus can be ignored. Therefore, the scattering profile of I(q) is only related to the form factor P(q) of the particles. As a result, I (q) can be furthered to the Guinier’s expression [123,124,125], involving radius of gyration Rg as described by Equation (4.2). In this case, a linear relationship exists between ln(I) and q2, with a slope of (-Rg2/3).
Figures 3(a) and 3(b) show ln(I) plot against q2 in the 2D GISAXS scattering patterns of the hybrid low-k films cured at various temperatures under slow curing and rapid curing conditions, respectively. The slope, i.e. Rg size, showed that the porogen Rg size increased non-linearly with the curing temperature between 50 and 210oC in the slowly cured system as shown in Figure 4.2(a), but varied little with temperature between 50 and 210oC in the rapidly cured system as shown in Figure 4.2(b). Since the shape of SBS porogen in the hybrid film is close to sphere as confirmed by scanning electron microscope, the pore size, d could be deduced from d = 2(5/3)1/2Rg [126,127]. In addition, the porogen size distribution is estimated from the change of the slopes in the low-q region; for instance, monodispersity corresponds to a single slope. As a result, the calculated porogen size and distribution of the hybrid low-k films as a function of curing temperatures under the slow curing and rapid curing conditions are shown in
o
size/distribution increased from 12.7 ± 2.3 nm to 32.8 ± 5.4 nm. Moreover, the increase rate of pore size became noticeable at T > 110oC, and more significant between 130oC and 170oC (from 17.5 ± 2.9 nm to 31.2 ± 4.9 nm), but little variation between 170oC and 210oC (from 31.2 ± 4.9 nm to 32.8 ± 5.4 nm). In contrast, the porogen size/distribution remained about 12 ± 2 nm with a narrow size distribution upon rapid curing up to 210oC as shown in Figure 4.3(b).
4.2 In-situ visosity test
Subsequently, the interaction between SBS porogen and MSQ matrix was examined by an in-situ viscosity measurement from room temperature to 200oC (1) at a slow curing rate (2oC/min), and (2) at a rapid curing rate, i.e. isothermally at 200oC as
Subsequently, the interaction between SBS porogen and MSQ matrix was examined by an in-situ viscosity measurement from room temperature to 200oC (1) at a slow curing rate (2oC/min), and (2) at a rapid curing rate, i.e. isothermally at 200oC as