Organic Thin Films - American Chemical Society

Chapter 27

Polymeric Organic-Inorganic Hybrid Nanocomposites 1,3

1

1

1

1

J. L. Hedrick , R. D. Miller , D. Yoon , H. J. Cha , H. R. Brown , S. A. Srinivasan , R. Di Pietro , V.Flores ,J.Hummer ,R. Cook , E . Liniger , E. Simonyi , and D. Klaus 1

1

1

Downloaded by UNIV OF SYDNEY on May 3, 2015 | http://pubs.acs.org Publication Date: August 6, 1998 | doi: 10.1021/bk-1998-0695.ch027

2

2

2

2

2

1

2

IBM Almaden Research Center, 650 Harry Road, San Jose, CA 95120 IBM T. J. Watson Research Center, Route 134 Kitchawan Road, Yorktown Heights, NY 10589 Transparent, nanophase-separated inorganic-organic hybrid polymers with high modulus and excellent crack resistance have been prepared from reactively functionalized poly(amic ester) precursors of polyimide and substituted, oligomeric silsesquioxanes. These polyimide/ poly(silsesquioxane) hybrid materials are stable to 400 °C and exhibit an isotropic dielectric constant of ca. 2.9. Induced cracking and crack propagation studies performed with the application of external stress suggest a maximum critical film thickness in excess of 2.0 μm. These hybrid materials appear to be significantly toughened by the chemical incorporation of the polyimides relative to organically modified silicates and spin-on-glasses without adversely affecting other important properties of spin-on-glass.

As on-chip device densities increase and active device dimensions shrink, signal delays increase due to capacitive coupling and crosstalk between the metal interconnects (1). The situation is exacerbated by the need to keep conductor lines as short as possible in order to miriimize transmission delays, thus necessitating complex multilevel wiring schemes for the chip. Since both capacitive delays and power consumption each depend critically on the dielectric constant of the insulators (2,3), much attention has focused recently on the replacement of the standard silicon oxide with new intermetal dielectrics (IMD) having dielectric constants lower than conventional oxide (k = 3.9-4.2) (2,4). This is not a simple matter given the complexities and demands of current semiconductor integration processes which are such that replacement of oxide as the IMD material with a material of significantly lower dielectric constant could be considered to be one of the great materials challenges of the 90's. A replacement dielectric must not only have a significantly lower andfrequencyindependent dielectric constant, but also must withstand temperatures of 400-450 °C associated with current integration processes, provide a barrier to metal ion diffusion, adhere strongly to a 3

Corresponding author.

©1998 American Chemical Society

In Organic Thin Films; Frank, C.; ACS Symposium Series; American Chemical Society: Washington, DC, 1998.

371

372 variety of metals, ceramics and substrates, resist cracking under extreme conditions, be inert to common metallurgy and liners, be resistant to 0 -RlE etching and tolerant to chemical mechanical polishing (CMP) yet amenable to high resolution lithographic imaging, and show low water absorption, etc. In addition, an IMD material should be electrically isotropic, show minimal leakage currents and possess a high dielectric breakdown threshold. Although many organic polymers have dielectric constants below those of thermal and C V D oxide, their thermal stabilities and/or glass transition temperatures are often inadequate for current semiconductor processing. This has focused attention recently on inorganic-organic network materials such as spin-on-glasses (SOG), organically modified silicates, silsesquioxanes, etc., prepared by sol-gel condensation processes (4,5). These materials have many attributes. When properly cured, they have dielectric constants ranging from 2.7-3.2, are chemically unreactive and are exceptionally thermally stable. However, due to the loss of solvent and other small molecules through the polycondensation, shrinkage (6) occurs during curing which often leads to cracking, a situation which is exacerbated as the functionality number of the material and/or film thickness increases. For this reason, there have been numerous attempts to toughen silicates and organically modified derivatives through the incorporation of a variety of thermoplastic or network forming polymers (7). The advent of the sol-gel or spin-on-glass process for the preparation of high purity glass or ceramics has created the possibility to incorporate a polymeric component into the precursor solution (8-10). The chemistry of the sol-gel process involves the hydrolysis and condensation oligomerization of a group xiv metal alkoxide in solution (sol) and subsequent formation of a three-dimensional network. In such cases, the organic polymer must be thermally stable to allow proper curing of the silicate matrix. For use of such materials in IMD applications, the additive polymer must not only be thermally stable but should also have a relatively low dielectric constant.


2

Polyimide-inorganic hybrids are an emerging class of materials which may be designed to offer a range of properties depending on the relative composition of each component, the size scale of phase separation, and the chemistry and interactions between the organic and inorganic components (11-20). The initial attempts at hybridization involved mixing of a poly(amic-acid) solution with tetraethylorthosilicate (TEOS) in an aprotic dipolar solvent (11). In these systems, gross macroscopic phase separation was observed. Partial control of the phase separation and resulting morphology was accomplished by introducing inorganic functionality in the polyimide so as to chemically incorporate the organic component into the TEOS-based network (12-14). This was accomplished for low inorganic compositions by binding the metal alkoxide precursor to the carboxylic sites of the poly(amic-acid) (15,16). Upon imidization, the metal alkoxide is released and condenses, producing dispersed inorganic particles in the polyimide matrix. The rigidity of the polyimide prevents coarsening of the inorganic component. The resulting films were clear and small particles were reported. Alternatively, Mascia and Kionl reported the use of glycidylpropyltrimethoxysilane as a means to compatibilize TEOS with poly(amic-acids) (17-20). The use of the


373


coupling agent served to significantly reduce the size scale of phase separation, which occurred via a spinodal decomposition process. Other key factors in structure control included the molecular weight of the poly(amic-acid), and the reaction time for the coupling agent and catalyst. Other polyimide-silica hybrid materials were prepared by hydrolysis and cocondensation of TEOS with polyimides containing the triethoxysilane functionality. Relatively tough, free standing films were obtained for silica compositions as high as 70 wt.%. The degree of phase separation was controlled by the composition of TEOS in the mixture and the number of inorganic functionalities. In general, the size of the phases increased with increasing silica content and decreased with triethoxysilane content (18). The synthetic procedure for the preparation of polyiniide-silica hybrids involves mixing varying quantities of TEOS and water with the poly(amic-acid) solution. The resulting mixtures are cast and cured to effect imidization and concurrent network formation. For this, water is essential, and the stability of the poly(amic-acid) to hydrolysis is of concern. To this end, we have surveyed an alternative route to polyimide-silica hybrids based on a poly(amic ethyl ester) precursor to the polyimide (20). In this route, pyromellitic dianhydride (PMDA) is opened by ethanol to yield a meta, para mixture of half acid esters which can be separated by fractional recrystallization and converted to the respective acid chlorides. Polymerization with a diamine yields the target poly(amic alkyl esters). These precursors are hydrolytically stable allowing isolation, characterization and copolymerization in a wide variety of solvents and solvent mixtures. We have primarily used the poly(amic ethyl ester) precursor since imidization occurs at a substantially higher temperature than the poly(amic-acid) analog. In polyimide-silica hybrids derived from poly(amic-acid) solutions, the morphology is strongly influenced by the hydrogen bonding between the organic and inorganic components. If this interaction is lost prior to vitrification (i.e., imidization), significant coarsening during phase separation is observed. Chujo and coworkers (21) and others (22) have reported that nano-level phase separation is obtained for organic-inorganic hybrids only when there is inorganic functionality on the organic component and there is a strong interaction (i.e., hydrogen bonding) between the components. The onset of imidization of the poly(amic ethyl ester) is 250 °C and a cure temperature of 350 °C is required for quantitative imidization. In this article, we will discuss the preparation of triethoxysilyl functional poly(amic ethyl ester) oligomers, and the chemical modification of inorganic precursors as a route to microphase separated inorganic-organic hybrids. Experimental Materials. The N-methyl-2-pyrrolidone (NMP), dimethylpropylene urea (DMPU), and pyridine were purchased from Aldrich and used without further purification. The aminophenyltrimethoxysilane was purchased from Gelest Inc. and used without further purification. The bis(p-arninophenyl) phenyl trifluoromethylmethane was prepared according to a literature procedure (23).


374


Polymer Synthesis Trimethoxysilyl Functionalized Poly(Amic Ethyl Ester) Oligomers ( = 10,000 g/mol) (Scheme 1) (24). To a three-neck flask equipped with an overhead stirrer, nitrogen inlet and addition funnel was charged 3.2456 g (9.48 mmol) of 3FDA, 0.2218 g (1.04 mmol) of aminophenyltrimethoxysilane, 2 g (25 mmol) of pyridine and 50 mL of distilled NMP. The reaction mixture was maintained under a positive nitrogen pressure and cooled to 0 °C. The P M D A diethyl ester diacyl chloride (3.4716 g, 10 mmol ) was dissolved in ~ 100 mL of methylene chloride, quantitatively transferred to the addition funnel and added dropwise to the cold, stirred reaction mixture. After the addition was complete, the polymerization was allowed to proceed overnight at room temperature. The poly(amic ethyl ester) oligomer was isolated by precipitation under high shear conditions (Waring blender) in methanol, filtered and dried in a vacuum oven at 60 °C. An oligomer ( = 20,000 g/mol) was synthesized likewise employing the appropriate stoichiometry. n

Characterization The inherent viscosities were obtained with 0.5 g/dL solutions in N M P at 30 °C employing a Cannon Ubbelohde viscometer and were calculated from an average of five different runs. H N M R spectra were obtained on a Bruker A C 250 MHz N M R spectrometer in either CDCL3 or deuterated DMSO and are reported in ppm (δ) downfield from TMS. In the case of the trimethoxysilyl functionalized oligomers, the number average molecular weights were estimated from the ratio of the methyl protons of the ester functionality to the methoxy protons of the trimethoxysilyl end-group. FTIR analyses were carried out on a Nicolet FTIR on films prepared from a solution of poly(amic ethyl ester) in THF on NaCl plates. Dynamic T G A was performed on a Perkin-Elmer TGA-7 in air at a heating rate of 10 °C/min. Dynamic DSC was carried out on a DuPont 1090 instrument at 10 °C/min. Measure of Fracture Properties. The hardness and modulus of the films were determined using an instrumented nano-indenter. A Berkovich diamond-pyramid triangular indenter was loaded onto the surface of 1 μτη thick films to a peak load of 2 mN and then unloaded. The monitored load and displacement of the indenter exhibited a hysteresis and residual displacement on loading typical of brittle materials. Fracture properties were characterized using a controlled-flaw stressing technique in which the films were exposed to a reactive environment under applied stress. Strips 10 mm wide and 30 mm long were cleaved from the coated silicon wafers and a 5N indentation made in the center of the strips. The Vickers diamond-pyramid square indenter used produced a contact impression with cracks approximately 30 μτη long emanating from the corners that extended through the film to the silicon substrate. The indented strips were then immersed in water either as indented or in a small four point bend fixture that applied 70 MPa tension to the outer fiber of the silicon substrate resulting in approximately 6 to 9 MPa additional stress on the film. Control strips that contained no indentations were also monitored or were exposed to air


375


Ο

Ph-

Organic Thin Films - American Chemical Society

Recommend Documents