Low dielectric and thermally stable hybrid ternary composites of hyperbranched and linear polyimides with SiO₂

Abstract

A hydroxyl-terminated hyperbranched polyimide was synthesized via the A₂ + B₃ reaction between dianhydride and triamine monomers. The hydroxyl groups at the peripheral positions were then introduced by modification of the anhydride end groups via a reaction with 4-aminophenol. Based on the hydroxyl-terminated hyperbranched polyimide, HBPI_{BPADA-TAP(OH)}, we successfully fabricated hybrid ternary composites, which were comprised of a linear polyimide (PI_6FDA-APB), HBPI_{BPADA-TAP(OH)}, and an inorganic SiO₂ component. The material was designed to satisfy the requirement for cutting-edge insulators with a low dielectric constant and a high thermal stability. Because of the appropriate choice of the hybrid ternary composite systems with HBPI_{BPADA-TAP(OH)} and inorganic silica, it is sensible to improve the dielectric properties and thermal resistant properties of unary systems or improve the disadvantages of the dielectric and optical properties of binary systems. For an optimized composition, the dielectric constant (D_k) of the PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-30%–SiO₂-20% composite reaches the lowest value of 2.24 at 100 kHz. Research also showed that the optical transparency is significantly improved with the increase of the HBPI_{BPADA-TAP(OH)} content in the composite. Compared with the binary linear PI _6FDA-APB–SiO₂ composite, the transmittance increases from 1% to 75% at the wavelength of 450 nm. The incorporation of SiO₂ can preserve the good thermal properties of the hybrid composites containing HBPI_{BPADA-TAP(OH)}. By adding 10% of HBPI_{BPADA-TAP(OH)} to the PI_6FDA-APB–SiO₂-20% system, the coefficient of thermal expansion of the hybrid ternary composite is 20.9 ppm °C⁻¹ in the temperature range from 100 to 150 °C, which is significantly lower than that of the linear polyimide (37.1 ppm °C⁻¹ for PI_6FDA-APB). Because of these optimized properties, hybrid ternary composites have the potential for use in applications in the micro-electronic insulator fields, such as interlayer dielectrics of advanced electronic devices.

---

Introduction

In the electrical and electronics industries, insulating interlayer materials with a low dielectric constant are urgently needed in order to improve the speed of electron flow transmission between the chips of large scale integrated circuits and to meet the requirement of high integration.¹ Polyimide (PI) with low dielectric constant (low D_k) is a desirable insulator material for applications in high-speed and high-frequency circuits. The development of PI with low D_k has been the focus of several recent investigations.^1–5 In recent years, research has been carried out on the organic–inorganic or other composite materials.^2,6,8 The fabricated PI composites with well-designed organic–inorganic structures can reduce the D_k by utilizing the effects of air volume, air gap or pore voids,^7,8 low polarization,^9–11 and increased free volume.^6,7,9,10,12 However, some drawbacks still remain in their thermal and mechanical properties as well as in the optical transparency. Therefore, there are some limitations for using them for real applications in the insulator material field. Generally speaking, ideal insulator materials should possess not only low dielectric properties, but also good mechanical properties and high thermal stability; in many cases, optical transparency is required as well. In order to satisfy these requirements, introducing the third component, such as a hyperbranched polymer, can be a promising way to reduce the dielectric constant and optimize the other properties at the same time.

Since they were reported by Kim and Webster in 1990,¹³ hyperbranched polymers (HBPs) have received considerable attention for applications in many fields.¹⁴ The properties of HBPs were predicted by Flory as early as 1952.¹⁵ The practical synthesis of HBPs can be traced back to the report by Kricheldorf et al. in 1982.¹⁶ HBPs are well known for their unique properties, such as possessing a large number of end groups, low solution viscosity, high solubility, and others. HBPs can be used where materials with improved properties are needed to supplement their linear analogues.

In recent years, hyperbranched polyimides (HBPIs) have been synthesized, characterized, and used for applications requiring high gas permeability with enlarged free volume.^17–19 Those studies showed that there are many nano-scale open and accessible cavities (typically several angstroms in size) in a rigid branched structure.^6,20,21 These characteristics have been exploited for the development of PI materials with improved properties.^22,23

It is of particular interest to optimize the dielectric, optical, and thermal properties by using HBPI as an additional component in a unary or binary system. For a unary or binary system with an inorganic component,^{6,7,9,10,12,24} there is a limit to the faults that can be supplemented. For example, although addition of inorganic silica could reduce the dielectric constant, the composites often appear opaque because of phase-separation. Fluorine-containing PI with reduced charge transfer (CT) interactions are optically transparent in the visible region, but their dielectric constants and coefficients of thermal expansion are not satisfactory for their application as insulators. Therefore, there is a requirement to systematically investigate the function of HBPIs as additional components in the composites to obtain PI composites with optimized properties.

In this paper, the preparation and properties of a series of novel hybrid ternary composites of PIs with SiO₂ are reported. The composites were prepared by using a hydroxyl-terminated hyperbranched polyimide (HBPI_{BPADA-TAP(OH)}) synthesized in this study. HBPI_{BPADA-TAP(OH)} and the precursor of a linear PI (PI_6FDA-APB) were used to prepare the composites with inorganic silica using the sol–gel method. From the results, it was concluded that HBPI_{BPADA-TAP(OH)} can be identified not only to have the effect to remedy the interaction faults between organic PI_6FDA-APB and the inorganic silica network, but also to be effective at improving the comprehensive properties of the materials in a complementary manner. It can reduce the dielectric constant, enhance the optical transmittance, and does not have negative effects on the thermal stability. The hydroxyl terminated HBPI plays a key role in connecting PI components and the inorganic silica network. The preparation, structures and properties of the hybrid ternary composites are reported in the next sections.

Experimental section

Materials

4,4′-(Hexafluoroisopropylidene)diphthalic anhydride (6FDA, 95%), 2,4,6-triaminopyrimidine (TAP, 98%), 1-hydroxy-cyclohexyl phenyl ketone, and 1,3-bis(3-aminophenoxy)benzene (APB, 98%) were purchased from Adamas Reagent Co. Ltd. 4,4′-Bis(4,4′-isopropylidene diphenoxy) bis-(phthalic anhydride) (BPADA 97%) was purchased from Sigma-Aldrich and Aladdin Chemical Company. 4-Amino-phenol was purchased from Tianjin Chemical Engineering Laboratory. N,N-Dimethylacetamide (DMAc, 98%), N,N-dimethylmethanamide (98%), tetrahydrofuran (THF, 98%), and toluene which were used as the reaction media were purchased from the Beijing Chemical Works and Alfa Aesar. The solvent N-methyl-2-pyrrolidone (NMP, 97%) was purchased from Beijing Modern East Fine Chemical. Hydrochloric acid (HCl) and tetraethoxysilane (TEOS, 98%) were purchased from Alfa Aesar and used without further purification. If it is not mentioned specifically, the reactants and solvents were used as received without further purification.

Synthesis

The synthetic routes of the materials are shown in Schemes 1–3, which include the preparations of PI_6FDA-APB and its composite with silica (Scheme 1), the hydroxyl-terminated hyperbranched HBPI_{BPADA-TAP(OH)} (Scheme 2), the hybrid ternary PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}–SiO₂ composites (Scheme 3). The preparation details are described next.

Scheme 1 Synthetic route for the PI_6FDA-APB and its composite with silica.

Scheme 2 Synthetic route for the hydroxyl terminated hyperbranched HBPI_{BPADA-TAP(OH)}.

Scheme 3 Synthetic route for the PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}–SiO₂ hybrid ternary composites.

Linear polyimide (PI_6FDA-APB). The polymerization was performed by a traditional two-step condensation process. The polyamic acid (PAA) was firstly synthesized at 0 °C. Then, PAA was heated through several stages to a high temperature and then dehydrated to obtain the PI film. The diamine monomer, APB (0.657 g, 2.25 mmol) was dissolved in NMP (10 wt%) and cooled in an ice water bath. After complete dissolution, the anhydride monomer, 6FDA (1 g, 2.25 mmol) was added to the solution with stirring. The reaction to obtain the PAA solution proceeded, with magnetic stirring at 40 °C, for 24 h. The PAA solution was cast on a piece of glass plate and heated at 80 °C for 2 h. The film was thermally imidized by stepwise heating at 150 °C (1 h), 200 °C (1 h), and 300 °C (1 h). FTIR (KBr, cm⁻¹): 1783 cm⁻¹ (C=O sym. str.); 1720 cm⁻¹ (C=O asym. str.); 1585, 1479, 1448 cm⁻¹ (C=C str. Ar.); 1367 cm⁻¹ (C–N str. imide); 1238, 1187 cm⁻¹ (Ar–O–Ar); 1137 cm⁻¹ (–CF₃); 960, 844 cm⁻¹(Ar–H); 779, 717 cm⁻¹ (Subst. Ar.); and 1660–1670 cm⁻¹ (Non PAA structure band).

Linear polyimide–silica hybrid composite (PI_6FDA-APB–SiO₂-20%). The preparation of the linear silica composite was carried out by a typical sol–gel method (Scheme 1). The preparation of the composite containing 20 mol% SiO₂ is given here as a typical example. Stoichiometric quantities of TEOS in DMAc (10 wt%, 1.38 g), and HCl in deionized water (0.1 N, 0.048 g) were mixed and stirred at room temperature for 0.5 h to form the SiO₂ sol. Then, the SiO₂ sol was added dropwise to the PAA solution (10 wt%, 2 g) with stirring. The mixture was then stirred at room temperature for 6 h and a PAA/SiO₂ precursor was obtained. In the second stage, the PAA/SiO₂ precursor solution was cast on a glass plate and heated at 80 °C for 2 h. The film was then thermally imidized by step-wise heating at 150 °C (1 h), 200 °C (1 h), and 300 °C (1 h) to obtain the PAA_6FDA-APB–SiO₂ composite. FTIR (KBr, cm⁻¹): 1783 cm⁻¹ (C=O sym. str.); 1720 cm⁻¹ (C=O asym. str.); 1585, 1479 cm⁻¹ (C=C str. Ar.); 1369 cm⁻¹ (C–N str. imide); 1240, 1189 cm ⁻¹ (Ar–O–Ar); 1137 cm ⁻¹ (–CF₃); 1068–1003 cm⁻¹ (Si–O–Si); 960 cm ⁻¹(Ar–H); 890 cm⁻¹ (Si–OH); 844 cm⁻¹(Ar–H); 779, 717 cm⁻¹ (Subst. Ar.); and 1660–1670 cm⁻¹ (Non PAA structure band).

Hydroxyl terminated hyperbranched polyimide (HBPI_{BPADA-TAP(OH)}). The anhydride terminated HBPI was prepared by A₂ + B₃ polycondensation (Scheme 2).^15,25,26 The molar ratio of dianhydride/triamine was at 2 : 1. TAP (0.06 g, 0.48 mmol) and BPADA (0.50 g, 0.96 mmol) were dissolved in NMP (100 mL) to obtain a solution with 60 wt% solid content. After reaction with stirring at 40 °C for 12 h, the anhydride-terminated hyperbranched polyamic acid (HBPAA) was obtained. An excess of 4-aminophenol (0.058 g, 0.53 mmol), and NMP (21 mL) were added to the HBPAA solution and the mixture was stirred with a magnetic bar at 40 °C for 24 h. Then toluene (50 mL) was added dropwise into the mixture and the reaction was carried out at 180 °C for 36 h under nitrogen gas protection. The mixture was poured into methanol and the gray powder was precipitated out. The product was obtained by filtration and dried under vacuum at 80 °C for 3 h. Yield = 82%. M_w = 8700, M_n = 3100, polydispersity index (PDI) (M_w/M_n) = 2.8. ¹H-NMR (600 MHz, deuterated dimethylsulfoxide), δ (ppm): 9.73 (OH, 1H_j), 7.88 (CH, 2H_f), 7.4–7.2 (CH, 4H_b, 2H_e), 7.29 (CH, 2H_d), 7.10 (CH, 4H_g), 7.01 (CH, 4H_c), 6.82 (CH, 2H_h), 3.71 (CH, 1H_i), 1.67 (CH₃, 6H_a). FTIR (KBr, cm⁻¹): 3500–3000 cm⁻¹ (OH); 2967 cm⁻¹ (CH₃ asym. str.); 1785 cm⁻¹ (C=O sym. str.); 1731 cm⁻¹ (C=O asym. str.); 1618 cm⁻¹ (N–C=N); 1598, 1479 cm⁻¹ (C=C str. Ar.); 1440 cm⁻¹ (CH₃); 1384 cm⁻¹ (C–N str. imide.); 1272, 1236 cm⁻¹ (Ar–O–Ar); 1175 cm⁻¹ (CH₃); 1014, 855, 717 cm⁻¹ (Subst. Ar.); and 1660–1670 cm⁻¹ (Non PAA structure band).

PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}–SiO₂ hybrid ternary composites. The hybrid ternary composites were prepared by using a typical sol–gel method (Scheme 3). As the methods for preparing the hybrid ternary composites are the same, to avoid repetition, only one of the methods for preparing the composite (PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-20%–SiO₂-20%) is given here as a typical example. Stoichiometric quantities of TEOS in DMAc (10 wt%, 1.38 g), and HCl in deionized water (0.1 N, 0.048 g) were mixed and stirred at room temperature for 0.5 h to form the SiO₂ sol. Then, the SiO₂ solution was added dropwise into the solution of linear PAA (10 wt%, 2 g) and HBPI_{BPADA-TAP(OH)} (0.04 g) with stirring. The mixture was stirred at room temperature for 12 h to obtain PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-20%–SiO₂-20% precursor. In the second stage, the precursor solution was cast on to a glass plate and heated at 80 °C for 2 h. The film was then thermally imidized by step-wise heating at 150 °C (1 h), 200 °C (1 h), and 300 °C (1 h).

The hybrid ternary composites with different compositions were prepared by a similar method by adjusting the composition of the PI and silica. The film formation property of the hybrid ternary composite depended on the contents of the HBPI_{BPADA-TAP(OH)} and TEOS. By using the linear PAA and HBPI, the hybrid ternary composites were prepared successfully. The films are called: SA-1 ∼ SA-5 for PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-0% ∼ 40%–SiO₂-20%, where the percentage given in the generic abbreviations is the weight percentage relative to the linear PI calculated from the reactant amounts. The FTIR results of the hybrid ternary composites are:

SA-2, SA-3, SA-5: FTIR (KBr, cm⁻¹): 2967 cm⁻¹ (CH₃ asym. str.); 1783 cm⁻¹ (C=O sym. str.); 1720 cm⁻¹ (C=O asym. str.); 1585, 1477 cm⁻¹ (C=C str. Ar.); 1369 cm⁻¹ (C–N str. imide); 1238, 1189 cm ⁻¹ (Ar–O–Ar); 1137 cm ⁻¹ (–CF₃); 1068–1003 cm⁻¹ (Si–O–Si); 960 cm ⁻¹(Ar–H); 890 cm⁻¹ (Si–OH); 846 cm⁻¹(Ar–H); 779, 717 cm⁻¹ (Subst. Ar.); and 1660–1670 cm⁻¹ (Non PAA structure band).

SA-4: FTIR (KBr, cm⁻¹): 2967 cm⁻¹ (CH₃ asym. str.); 1783 cm⁻¹ (C=O sym. str.); 1720 cm⁻¹ (C=O asym. str.); 1585, 1477 cm⁻¹ (C=C str. Ar.); 1371 cm⁻¹ (C–N str. imide); 1236, 1189 cm⁻¹ (Ar–O–Ar); 1137 cm⁻¹ (–CF₃); 1068–1003 cm⁻¹ (Si–O–Si); 960 cm⁻¹(Ar–H); 890 cm⁻¹ (Si–OH); 846 cm⁻¹(Ar–H); 779, 717 cm⁻¹ (Subst. Ar.); and 1660–1670 cm⁻¹ (Non PAA structure band).

Characterization

Proton nuclear magnetic resonance (¹H-NMR, 600 MHz) spectra were recorded on a JNM-ECA600 NMR spectrometer (JEOL). Fourier transform infrared (FTIR) spectroscopic measurements were performed using a Magna-IR 560 FTIR spectrophotometer (Nicolet) on samples incorporated in KBr disks. The spectra were recorded in the range of 4000–450 cm⁻¹ at a resolution of 0.35 cm⁻¹. The molecular weights and molecular weight distributions were determined by using GPC gel permeation chromatography (GPC) at room temperature with THF as eluent at a flow rate of 1 mL min⁻¹. The instrument was equipped with a Optilab T-rEX refractive index detector (Wyatt Technology) and fitted with a PLgel 5 μm mixed-D column (Agilent). The data were calibrated with linear polystyrene standards. The thermal properties were characterized by differential scanning calorimeter (DSC), thermal gravimetric analysis (TGA), and dynamic mechanical analysis (DMA). TGA was performed with a TGA 2050 thermogravimetric analyzer (TA Instruments) at a heating rate of 20 °C from room temperature to 900 °C under a continuous flow of nitrogen. Thermal phase transitions of the polymers were scanned using a DSC 2910 (TA Instrument Co.) with a heating rate of 20 °C min⁻¹ under a nitrogen atmosphere. The dielectric constant was determined using a NOVOCOOL Alpha-ANB (Novocontrol Technologies GmbH & Co. KG) dielectric analyzer with silver paint electrodes, at room temperature with scan frequencies from 10⁶ Hz to 1 Hz, and the reference was a commercial Kapton® HN type PI (D_k = 3.81 at 100 kHz). The thickness of the specimens was controlled to be 23–35 μm. The coefficient of thermal expansion (CTE) parallel to the film surfaces were measured using a DMA Q800 dynamic mechanical analyzer 9TA Instruments in extension mode over a temperature range from 25 to 320 °C with a force of 0.01 N. The samples used for the measurements were 14 mm in length, 5 mm in width, and 21–55 μm in thickness.

UV-visible absorption spectra of the films were measured on a Lamba Bio-40 spectrophotometer (Perkin-Elmer). Cross-sectional images of the PI hybrid films were studied by scanning electron microscopy (SEM). The SEM images were obtained using a S5500 electron microscope (Hitachi) operating at an acceleration voltage of 5.0 kV.

Results and Discussion

Synthesis and Characterization

A series of hybrid composites were prepared in this study. To facilitate the discussions, the abbreviations SA-1 ∼ SA-5 are used to denote the hybrid composites PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-0% ∼ 40%–SiO₂-20%. For comparison, a linear polyimide (PI_6FDA-APB) was also prepared and is referred to as S-1 in the following discussion. The chemical structure and synthetic routes of the materials are shown in Scheme 1–3. The preparation details are described in the experimental part of the paper.

The hydroxyl-terminated HBPI_{BPADA-TAP(OH)} is a key component for preparing the hybrid ternary composites, which was synthesized by the A₂ + B₃ polycondensation method. The number average molecular weight (M_n) and PDI (M_w/M_n) of HBPI_{BPADA-TAP(OH)} were 3100 and 2.8, respectively. The molecular structure of the hyperbranched PI was determined by ¹H-NMR and FTIR. Fig. 1 shows the ¹H-NMR spectrum of HBPI_{BPADA-TAP(OH)} and the assignment of the resonance signals. The signals corresponding to the HBPI main-chain can be clearly identified from the spectrum. The resonance signal of the hydroxyl group appears at 9.73 ppm. It is the hydroxyl terminal groups in HBPI from 4-aminophenol.²⁷ The signals at 6.82–7.88 ppm correspond to the benzene protons from BPADA (dianhydride monomer) and the 4-aminophenol (terminating agent). The resonance at 3.71 ppm is –CH from the TAP monomer. The characteristic signal at 1.67 ppm corresponds to the methyl protons from the –CH₃ of BPADA. This spectral evidence reveals that the hydroxyl terminated HBPI has been successfully prepared.

Fig. 1 ¹H-NMR spectra of the hydroxyl terminated hyperbranched polyimide (HBPI_{BPADA-TAP(OH)}).

The FTIR spectra of HBPI_{BPADA-TAP(OH)}, linear PI (S-1), and the composites (Series SA-1 ∼ SA-5) are shown in Fig. 2. From the FTIR spectra of HBPI_{BPADA-TAP(OH)} and S-1, the absorption bands of the imide groups are observed at 1785–1781 and 1731–1720 cm⁻¹ for the symmetric and anti-symmetric stretching vibrations of the carbonyl group. No obvious absorption bands of PAAs can be seen in the range form 1660 to 1670 cm⁻¹. The absorption band of the hydroxyl group (–OH) of HBPI_{BPADA-TAP(OH)} is clearly observed between 3500 cm⁻¹ and 3000 cm⁻¹ of the FTIR spectrum. It also proves that the hydroxyl groups from 4-aminophenol are successfully introduced into the termini of the hyperbranched polyimide.

Fig. 2 FTIR spectra of hydroxyl-terminated hyperbranched polyimide, linear polyimide (S-1), and PI/SiO₂ hybrid ternary composites, SA-1 ∼ 5 (PI_6FDA-APB–HBPI _{BPADA-TAP(OH)}-0% ∼ 40%–SiO₂-20%).

It can be seen that the absorption intensity of the bands around 1783 and 1720 cm⁻¹ (C=O stretching vibrations) increases with the increase of the amount of HBPI_{BPADA-TAP(OH)} by using the intensity of SiO₂ as the standard. The FTIR spectra of SA-1 ∼ 5 show significant absorption band shifts related to the transformation during the synthesis of the ternary composites. The absorption bands of the C=O shift are from 1785 and 1731 cm⁻¹ for HBPI_{BPADA-TAP(OH)} to 1783–1781 and 1720 cm⁻¹ for the hybrid ternary composites.

The absorption bands of C–N are shifted from 1384 cm⁻¹ for HBPI_{BPADA-TAP(OH)} to 1371–1369 cm⁻¹ for the hybrid ternary composites, and the bands for C–O–C between aromatic rings are also shifted from 1272 and 1236 cm⁻¹ for HBPI_{BPADA-TAP(OH)} to 1240–1238 cm⁻¹ and 1189–1187 cm⁻¹ for the hybrid ternary composites. These band-shifts are related to the reactions between the hydroxyl groups of HBPI_{BPADA-TAP(OH)} and Si–OH of TEOS, which enhance the molecular interactions. Therefore, with the increase of the percentage of HBPI_{BPADA-TAP(OH)}, the characteristic bands of C=O, C–N, and C–O are gradually shifted to lower wavenumbers accompanied by the increase of their absorption intensities. Because of the completed imidization, no obvious absorption bands of PAAs are observed in the range between 1660 and 1670 cm⁻¹.

In addition to the spectral characteristics related to the organic components, the FTIR spectra show absorption bands around 1000–1100 cm⁻¹ (Si–O–Si symmetric stretching vibrations) and 890 cm⁻¹ (Si–OH). The absorption bands show that the composites contain the inorganic component obtained from TEOS (SA-1 ∼ SA-5). It also verifies that not only are the PI precursors fully imidized, but also that the inorganic silica networks are formed in the hybrid ternary composites.

Morphology of hybrid ternary composites

Fig. 3 shows the typical SEM images of the PI and representative composite films. The reinforcement of HBPI_{BPADA-TAP(OH)} to the binding with PI_6FDA-APB and dispersion of SiO₂ in the matrix can be confirmed by the morphology. Fig. 3(a) shows the PI_6FDA-APB (S-1) without the inorganic component. For the composite of PI_6FDA-APB and SiO₂ (PI_6FDA-APB–SiO₂-20%, SA-1), the aggregated silica particles can be clearly seen as spherical beads with a smooth surface, which have an average diameter of around 1500 nm (Fig. 3(b)). When the hyperbranched PI is introduced, the dispersion of SiO₂ in the PI matrix is dramatically improved. For the ternary composites, such as PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-30%–SiO₂-20% (SA-4), the SEM observation indicates that HBPI_{BPADA-TAP(OH)} has an obvious effect in reducing the phase separation of SiO₂ (Fig 3(c) and 3(d)). The size of silica particles is significantly reduced and only particles with diameters of around 100–200 nm can be seen. The adhesion between PI and the inorganic component is also improved, which can be seen by the interface between the particles and matrix. Moreover, the image reveals that nano-scaled cavities exist in the hybrid ternary composite.

Fig. 3 Typical SEM images of PI and the composites, (a) PI_6FDA-APB (S-1), (b) PI_6FDA-APB–SiO₂-20% (SA-1), (c) and (d) PI_6FDA-APB_HBPI_{BPADA-TAP(OH)}-30%–SiO₂-20% (SA-4); scale bar: 2 μm (a, b, and c) and 500 nm (d).

The function of HBPI_{BPADA-TAP(OH)} to reduce the phase separation can be attributed to the peripheral hydroxyl (–OH) groups of the HBPI, which can form covalent linkages to the silica network. This enhanced interaction between the components can effectively reduce the aggregation of silica and suppress the phase separation between the PIs and the silica network. As will be discussed in the following sections, because of this effect, the transmittance of the composites can be significantly improved by incorporating an HBPI_{BPADA-TAP(OH)} component into the PI_6FDA-APB and the SiO₂ composite. The decrease in dielectric constants (D_k) at the frequencies between 1 and 10³ Hz can also be attributed to the reduced phase separation.

Dielectric properties of the hybrid ternary composites

Fig. 4(a) shows the dielectric constants (D_k) of the films measured at room temperature with the scanning frequencies from 10⁶ (1 MHz) to 1 Hz, which is a typical range for applied current electric insulator devices. The dielectric constants of this series of hybrid ternary composites are compared with PI_6FDA-APB (S-1), PI_6FDA-APB–SiO₂-20% (SA-1) and a commercial Kapton® HN type PI. The dielectric constants (D_k) show an increase with the decreasing frequency, which is typical behavior of the frequency dependence of D_k for solid matters.²⁸ This frequency-dependent behavior can be described by the Cole–Cole equation,²⁹

ε* − ε_∞ = (ε₀ − ε_∞)/[1 + (iωτ₀)^1−α] (1)

where: ε* is the complex dielectric constant, ε₀ and ε_∞ are the dielectric constants at “static” and “infinite frequency”, ω = 2π times the frequency, and τ₀ is a generalized relaxation time. The exponent parameter α can assume a certain value between 0 and 1, in which the former case corresponds to the result by Debye for polar dielectrics.²⁹

Fig. 4 Dielectric constant (D_k) of the linear polyimide (S-1) and hyperbranched polyimide with PI/SiO₂ hybrid ternary composite films SA-1 ∼ 5 (PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-0% ∼ 40%–SiO₂-20%), (a) scanning frequency from 10⁶ to 1 Hz, (b) frequency is 100 kHz.

The dielectric constants of the films prepared in this study are significantly smaller than that of the Kapton® HN type PI. Compared with the two PIs, the Kapton® HN type PI and S-1, the dielectric constants of the organic–inorganic composites show a more significant dependence on the frequency, which shows a sudden drop in the characteristic frequencies. Therefore, in the high frequency range (>10³ Hz), the dielectric constants of the composites are all lower than PI_6FDA-APB (S-1). The decrease of the D_k in the 1 Hz to 10³ Hz range shown in Fig. 4(a) is related to a decrease in space charge polarization between organic and inorganic phases. It is interesting to note that by adding the HBPI (HBPI_{BPADA-TAP(OH)}) as the third component, the dielectric constants of the hybrid ternary composites (SA-2 ∼ SA-5) are always lower than the binary composite (SA-1). Especially for the optimized condition, PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-30%–SiO₂-20% (SA-4) exhibits the lowest D_k compared with others.

The effect of HBPI_{BPADA-TAP(OH)} to reduce D_k can be attributed to the effect of the HBPI component to enhance the homogeneous dispersion of silica in the system. HBPI_{BPADA-TAP(OH)} is highly compatible with PI_6FDA-APB through hydrogen bonding and other intermolecular interactions. More importantly, it can react with inorganic silica networks in the Si–OH/SiO₂ transformation of TEOS. Because HBPI_{BPADA-TAP(OH)} can improve the inorganic–organic phase dispersion, it plays a very important role in reducing the dielectric constant of hybrid ternary composites. The effect of reducing the dielectric constants can also be attributed to nano-scale cavities (typically several angstroms in size) in the branched structure for hyperbranched polymers.^6,20,21

The dielectric constants measured at 100 kHz for this series of hybrid ternary composites are listed in Table 1 together with the data for PI_6FDA-APB (S-1) and PI_6FDA-APB–SiO₂-20% (SA-1) for comparison. The dielectric constants of PIs determined at this frequency are related to the fast polarizability of the films. The results are also given in Fig. 4(b). The binary composite (SA-1) shows a lower dielectric constant than that of the linear PI (S-1), which is reduced from 2.90 to 2.70. After the addition of the HBPI (HBPI_{BPADA-TAP(OH)}) into the composites, the dielectric constants of the hybrid ternary composites (SA-2 ∼ 5) further decrease. For PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-30%–SiO₂-20% (SA-4), the composite achieves the lowest dielectric constant (D_k = 2.24). When the content of HBPI_{BPADA-TAP(OH)} exceeds this critical point, the dielectric constant of the hybrid ternary composite begins to slightly increase with increase in the amount of HBPI_{BPADA-TAP(OH)} as shown for SA-5. This increase could be attributed to the aggregation and poor dispersion of the excess HBPI_{BPADA-TAP(OH)} in the system.

Table 1 The parameters and properties of hydroxyl terminated HBPI with PI/SiO₂ hybrid ternary composites

Sample	Thickness^a μm	Dk^b	λcutoff nm	Transmittance		DSC	TGA			CTE^e ppm °C^−1
				450 nm/%	400 nm/%	Tg /°C	Td^5%^c/°C	Td^10%^c/°C	Rw800^d /%
a The thickness of specimens for dielectric constant measurement.b Measuring at frequency of 100 kHz.c Temperatures at which 5% and 10% weight loss occurred, respectively, recorded by TGA at a heating rate of 20 °C min^−1 and a N2 gas flow rate of 25 cm^3 min^−1.d Residual weight percentages at 800 °C.e The temperature range from 100 to 150 °C with a force of 0.01 N.
S-1 PI	24	2.90	325	94	81	205.1	549	571	53	37.1
SA-1 PI_SiO2-20%	25	2.70	331	1	1	205.0	527	565	57	17.8
SA-2 PI_HBPI(OH)-10%–SiO2-20%	24	2.63	327	2	1	197.6	487	552	58	20.9
SA-3 PI_HBPI(OH)-20%–SiO2-20%	27	2.34	324	9	4	200.5	487	547	57	23.1
SA-4 PI_HBPI(OH)-30%–SiO2-20%	23	2.24	326	75	50	200.2	485	533	54	28.7
SA-5 PI_HBPI(OH)-40%–SiO2-20%	35	2.35	326	33	14	192.2	481	538	53	29.5

---

The results above indicate that the incorporation of HBPI_{BPADA-TAP(OH)} shows a significant effect in lowering the dielectric constant, which can be attributed to the improvement of the dispersion of inorganic silica in the hybrid ternary composites and the nano-scale cavities of the hyperbranched structures. Therefore, the phase uniformity is enhanced by introducing an HBPI component up to 30%, which reduces the space charge polarization between the organic and inorganic phases. Meanwhile, this low value can also be attributed to the other characteristics of monomers and polymers, i.e., the high fluorine content from 6FDA,^30,31 the flexible or kinked structures from APB,^32,33 and the bulky substituent structure from BPADA.^34–36 In addition to these effects, the introduction of inorganic silica can effectively reduce humidity absorption of the material and enhances the free volume.²⁸

Optical properties of the hybrid ternary composites

The optical transparency of the linear PI (S-1), binary composite (SA-1) and hybrid ternary composite (SA-2 ∼ 5) was measured with UV-Vis spectroscopy. Fig. 5 shows the UV-Vis transmission spectra of the films. The cutoff wavelengths (absorption edge, λ_cutoff) and the transmittance at 450 nm and 400 nm are obtained from these spectra and listed in Table 1. S-1 (PI_6FDA-APB) shows the highest transmittance of 94% at 450 nm, and the film is entirely colorless. This is because it contains not only meta-linked diphenylether linkages from the APB moiety but also contains low polarizable fluorine (C–F) substituents from the 6FDA monomer, both of which are beneficial to the high optical transparency. The rigid and kinked molecular structure prevents dense chain packing and π-electron transfer between benzene and the imide rings. These factors effectively reduce the intermolecular CT-complex formation between the polymer backbones through the steric hindrance and the inductive effect by reducing the electron-donating capabilities of diamine moieties.

Fig. 5 UV-Vis spectra of linear polyimide (S-1) and HBPI with PI/SiO₂ hybrid ternary composites for SA-1 ∼ 5 (PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-0% ∼ 40%–SiO₂-20%).

After incorporating SiO₂ (20 wt%) into the PI system, the appearance of the SA-1 film becomes opaque and its transmittance is significantly decreased, which is attributed to the light scattering and opacity caused by the aggregated inorganic silica phase at the wavelength scale or above in the film. However, it is interesting to note that after introducing the HBPI_{BPADA-TAP(OH)} component into the PI_6FDA-APB and SiO₂ composite, the transmittance of the hybrid ternary composites is significantly improved until the amount of the hyperbranched PI component reaches 30%.

When the content of HBPI_{BPADA-TAP(OH)} is 30% (SA-4), the transmittance is the highest in the series of the hybrid composites (SA-1 ∼ 5), and shows a transmittance of about 75% at 450 nm. This content of HBPI_{BPADA-TAP(OH)} is appropriate for improving not only the optical properties but also the dielectric properties of the hybrid films. However, the transmittance of SA-5 (PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-40%–SiO₂-20%) is reduced. It could be caused by the excess of hydroxyl groups on the terminal surface of HBPI_{BPADA-TAP(OH)} (40%) for SiO₂-20%, which could cause inhomogeneous phase separation because of the strong aggregation and poor dispersion of HBPI_{BPADA-TAP(OH)}. This seems to be the same cause for the increased dielectric constant discussed in the previous section.

This result further confirms that HBPI_{BPADA-TAP(OH)} reduces organic–inorganic phase separation which was revealed by SEM observations. As mentioned above, this effect is attributed to the hydroxyl groups of HBPI_{BPADA-TAP(OH)} which introduce the linkage to the silica network and reduce the phase separation between PIs and the silica network. As a consequence, the size of the silica particles is decreased by enhanced interaction between the components. Because of this effect, the light scattering and opacity observed for SA-1 (PI_6FDA-APB–SiO₂-20%) film, which are caused by silica phase aggregation at the wavelength scale or above, can be significantly improved. The phase uniformity enhanced by introducing the HBPI component reduces both the dielectric constant and the light scattering to a proper level.

Thermal properties of the hybrid ternary composites

The thermal phase transition behavior of the hybrid ternary composites and related materials was investigated by DSC. The results are shown in Fig. 6(a) and summarized in Table 1. The hybrid ternary composites and related materials all show the phase transition behavior of an amorphous material, which means that the PI components exist in the amorphous phase. The hybrid ternary composites SA-2 ∼ 5 show T_g values ranging from 200.5 to 192.2 °C. The T_g of S-1 is 205.1 °C and the T_g of the binary composite with 20 wt% SiO₂ (SA-1) is 205.0 °C. After adding HBPI_{BPADA-TAP(OH)}, the T_g of the hybrid films becomes slightly lower than those of S-1 and SA-1. However, the differences in T_g of SA-2 to SA-4 are only a few degrees. The T_g of SA-5 is 192.2 °C, which is lowest in the series. As HBPI_{BPADA-TAP(OH)} has a hyperbranched structure and comparatively low molecular weight, it has a lower T_g compared with that of PI_6FDA-APB. However, when its concentration is below the critical value, HBPI_{BPADA-TAP(OH)} will not form a separate phase. Therefore, the addition of HBPI_{BPADA-TAP(OH)} does not show an obvious effect of decreasing the T_g of the hybrid ternary composites. Only when its concentration is too high, will the negative effects on the T_g be seen.

Fig. 6 Thermal properties of linear polyimide (S-1) and HBPI with PI/SiO₂ hybrid ternary composites for SA-1 ∼ 5 (PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-0% ∼ 40%–SiO₂-20%), (a) DSC, (b) TGA, and (c) CTE spectra.

The thermal decomposition temperatures of the hybrid ternary composites and related materials were measured by TGA analysis. The results are shown in Fig. 6(b) and summarized in Table 1. It can be observed for the SA series, that the residual weights of the organic–inorganic composites are nearly 100% below 400 °C. The hybrid ternary composites (SA-2 ∼ 5) show 5% weight loss at temperatures ranging from 487 to 481 °C, and 10% weight loss at temperatures ranging from 552 to 533 °C. The temperatures are obviously lower than those for the binary composite (527 and 565 °C for SA-1) and PI_6FDA-APB (549 and 571 °C for S-1). It can also be seen that the thermal decomposition temperature decreases with the increase of the HBPI_{BPADA-TAP(OH)} content in the systems. The decrease of the thermal decomposition temperatures is attributed to the property of the HBPI_{BPADA-TAP(OH)} component.

The CTE of the hybrid ternary composites and related materials was characterized by DMA. The CTE curves are shown in Fig. 6(c) and the CTE values are listed in Table 1. The materials all show significant thermal expansion in the temperature range higher than 200 °C, which corresponds to the glass transition as revealed by DSC analysis.

By comparing the CTE values below the T_g, the hybrid composites exhibit significantly smaller CTEs than that of the linear PI (S-1). The reduced CTE value is caused by the addition of inorganic silica. The CTE value of SA-1 (17.8 ppm °C⁻¹) is the smallest in the series. Compared to 37.1 ppm °C⁻¹ for PI_6FDA-APB, it is reduced by about 52%. For the hybrid ternary composites, SA-2 exhibits the most reduced CTE value (20.9 ppm °C⁻¹), which contains 10% of HBPI_{BPADA-TAP(OH)}. However, the CTE values increase with the further increase of HBPI_{BPADA-TAP(OH)}. The effectively reduced CTE value for the composites can be ascribed to the well-dispersed silica particles, which obstruct the expansion of polymer chains at elevated temperatures. Furthermore, there are many nano-scale cavities (typically several angstroms in size) in the branched structure of hyperbranched polymers.^6,20,21

As discussed previously, the structures can significantly reduce the dielectric constants of the hybrid ternary composites. However, the effect of the inorganic silica networks to reduce the CTE value is somehow counter-balanced by the HBPI component. This tendency can be seen by comparing the CTE values among the SA series. Even in this case, the CTE values of the hybrid ternary composites are still much smaller than that of the PI_6FDA-APB. The results shown previously indicate that there is a significant improvement of dielectric properties and transparency by introducing HBPI_{BPADA-TAP(OH)}. Its faults on the thermal properties can be complemented by the inorganic silica networks in the ternary system. Therefore, the hybrid ternary composites are very promising candidates for producing materials with optimized properties.

Conclusion

We synthesized a hydroxyl-terminated HBPI (HBPI_{BPADA-TAP(OH)}) via A₂ + B₃ condensation polymerization of TAP with BPADA and a post-polymerization reaction with 4-aminophenol. Based on HBPI_{BPADA-TAP(OH)}, a series of hybrid ternary composites were fabricated by using a linear PI (PI_6FDA-APB) and inorganic silica (SiO₂) through the sol–gel method. The hybrid ternary composites exhibit improved dielectric properties as an insulating material. At the appropriate content of HBPI_{BPADA-TAP(OH)}, the dielectric constant can reach the lowest value of 2.24 for PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-30%–SiO₂-20% (SA-4) in this series. The optical transmittance of the composites are also significantly improved with increasing HBPI_{BPADA-TAP(OH)}. The best result for the hybrid ternary composites was obtained for SA-4, where the corresponding transmittance at 450 nm was improved from 1% for the binary composite (SA-1) to 75% for SA-4. The CTE value of PI_6FDA-APB can be significantly reduced by introducing the inorganic SiO₂ phase. Although, incorporation of HBPI_{BPADA-TAP(OH)} will counter-balance the effect of SiO₂, the PI_6FDA-APB–HBPI_{BPADA-TAP(OH)}-10%–SiO₂-20% composite (SA-2) still exhibits the small CTE value of 20.9 ppm °C⁻¹. Compared to 37.1 ppm °C⁻¹ for PI_6FDA-APB, a reduction of 43% is obtained. The CTE value for the ternary composite with the lowest D_k (SA-4) is 28.7 ppm °C⁻¹, which corresponds to a decrease of CTE of 24%.

In general, from the results of these hybrid ternary composites, it can be concluded that the hydroxyl-terminated HBPI (HBPI_{BPADA-TAP(OH)}) is a new promising modifier to reduce dielectric constants and improve the overall performance for PI_6FDA-APB–SiO₂ (SA-1) binary systems. The fabricated hybrid ternary composites with HBPI_{BPADA-TAP(OH)} can be complementary in preventing the dielectric and thermal resistance drawbacks of unary (S-1) or dielectric and optical drawbacks of the binary system (SA-1) to improve overall properties. Thus, the hybrid ternary composite developed can meet the requirements for being an interlayer dielectric in advanced electronic devices.

Acknowledgements

The financial support from National Basic Research Program of China (973 Program) under Project 2011CB606102 and that for the academic visit of X. W. in Tokyo from Tokyo Institute of Technology are gratefully acknowledged.

Reference

1. S. H. Choi and S. J. Song, Polym. Sci. Technol., 2005, 16, 20.
2. H. Chen, L. Xie, H. Lu and Y. Yang, J. Mater. Chem., 2007, 17, 1258.
3. L. Tan, S. M. Liu, F. Zeng, Z. Y. Lin and J. Q. Zhao, Polym. Adv. Technol., 2010, 21, 435.
4. Y. H. Zhang, Z. M. Dang, J. H. Xin, W. A. Daoud, J. H. Ji, Y. Y. Liu, B. Fei, Y. Q. Li, J. T. Wu, S. Y. Yang and L. F. Li, Macromol. Rapid Commun., 2005, 26, 1473.
5. Y. H. Zhang, S. G. Lu, Y. Q. Li, Z. M. Dang, H. H. Xin, S. Y. Fu, G. T. Li, R. R. Guo and L. F. Li, Adv. Mater., 2005, 17, 1056.
6. J. C. Huang, P. C. Lim, L. Shen, P. K. Pallathadka, K. Y. Zheng and C. B. He, Acta Mater., 2005, 53, 2395.
7. V. E. Yudin, J. U. Otaigbe, S. Gladchenko, B. G. Olson, S. Nazarenko, E. N. Korytkova and V. V. Gusarow, Polymer, 2007, 48, 1306.
8. G. F. Zhao, T. Ishizaka, H. Kasai, M. Hasegawa, T. Furukawa, H. Nakanishi and H. Oikawa, Chem. Mater., 2009, 21, 419.
9. E. Hamciuc, C. Hamciuc and M. Olariu, Polym. Eng. Sci., 2010, 50, 520.
10. Y. T. Chern and J. Y. Tsai, Macromolecules, 2008, 41, 9556.
11. H. Deligöz, S. Özgümüş, T. Yalçınyuva, S. Yildirim, D. Deger and K. Ulutas, Polymer, 2005, 46, 3720.
12. H. W. Wang, R. X. Dong, H. C. Chu, K. C. Chang and W. C. Lee, Mater. Chem. Phys., 2005, 94, 42.
13. Y. H. Kim and O. W. Webster, J. Am. Chem. Soc., 1990, 112, 4592.
14. D. J. Massa, K. A. Shriner, S. R. Turner and B. I. Voit, Macromolecules, 1995, 28, 3214.
15. P. J. Flory, J. Am. Chem. Soc., 1952, 74, 2718.
16. H. R. Kricheldorf, Q. Z. Zang and G. Schwarz, Polymer, 1982, 23, 1821.
17. R. M. Harrison and W. J. Feast, Abstracts of Papers of the American Chemical Society, 1997, 214, 147.
18. K. L. Wang, M. Jikei and M. A. Kakimoto, J. Polym. Sci., Part A: Polym. Chem., 2004, 42, 3200.
19. T. Suzuki, Y. Yamada and Y. Tsujita, Polymer, 2004, 45, 7167.
20. Y. J. Lee, J. M. Huang, S. W. Kuo and F. C. Chang, Polymer, 2005, 46, 10056.
21. T. Seçkin, S. Köytepe and H. I. Adıgüzel, Mater. Chem. Phys., 2008, 112, 1040.
22. S. Baskaran, J. Liu, K. Domansky, N. Kohler, X. Li, C. Coyle, G. E. Fryxell, S. Thevuthasan and R. E. Willford, Adv. Mater., 2000, 12, 291.
23. Y. J. Lee, J. M. Huang, S. W. Kuo, J. S. Lu and F. C. Chang, Polymer, 2005, 46, 173.
24. S. J. Park, H. J. Sohn, S. K. Hong and G. S. Shin, J. Colloid Interface Sci., 2009, 332, 246.
25. J. Hao, M. Jikei and M. Kakimoto, Macromolecules, 2002, 35, 5372.
26. F. Jianhua, K. Hidetoshi and O. Kenichi, Macromolecules, 2000, 33, 4639.
27. A. Akar, A. Tunca and N. Talini, Eur. Polym. J., 1995, 31, 9.
28. J. O. Simpson and A. K. St Clair, Thin Solid Films, 1997, 308–309, 480.
29. K. S. Cole and R. H. Cole, J. Chem. Phys., 1941, 9, 341.
30. G. Hougham, G. Tesoro and J. Shaw, Macromolecules, 1995, 27, 3642.
31. A. E. Feiring, B. C. Auman and E. R. Wonchoba, Macromolecules, 1993, 26, 2779.
32. G. C. Eastmond and J. Paprotny, Macromolecules, 1995, 28, 2140.
33. Y. Imai, High Perform. Polym., 1995, 7, 337.
34. Y. T. Chern and H. C. Shiue, Macromolecules, 1997, 30, 4646.
35. L. J. Mathias, A. V. G. Muir and V. R. Reichert, Macromolecules, 1991, 24, 5232.
36. X. T. Han, Y. Tian, L. H. Wang and C. F. Xiao, J. Appl. Polym. Sci., 2008, 107, 618.

---