Dielectric Anisotropy in CVD Polymer Thin Films
The directional characteristics of polymer intermetal dielectrics are affected by CVD parameters.
Jay J. Senkevich, Seshu B. Desu,Virginia Polytechnic Institute, Blacksburg, Va. -- Semiconductor International, 6/1/1998
| At a Glance | |||
| |||
All parylene polymers are semicrystalline either as-deposited or when subsequently annealed above their glass transition temperatures (Tg), unlike SiO2 , which is amorphous at lower temperatures(<400°C).3 The semicrystalline phase of the polymer may be characterized with techniques such as X-ray diffraction (XRD), but even within the polymer's unit cell, the morphology and resultant properties are complex. Also, polymers possess an amorphous phase that is often intimately related to the crystalline phase. This is due to the formation of three-dimensional spherulites in the polymer thin films that contain the amorphous phase within this spherulite structure.
To characterize semicrystalline CVD polymer thin films, an optical method using variable angle spectroscopic ellipsometry (VASE) is employed. Both the in-plane and the out-of-plane indices of refraction may be obtained, which can be related to the anisotropic k by use of Maxwell's relation, n2 = k. This relation, of course, assumes only electronic polarization contributes to k, which is only a good assumption with hydrocarbon or fluorocarbon polymers. The dielectric anisotropy of these thin films should be well characterized and well understood before any attempt is made to integrate these materials into ULSI devices.
This study investigates the optical birefringence of poly (tetrafluoro-p-xylylene) (VT-4), poly(p-xylylene) (PPXN) and poly(chloro-p-xylylene) (PPXC) to determine their dielectric anisotropies as a function of post-deposition anneals (Fig. 1). The birefringence is defined as the following:
D = nOut-of-plane nIn-plane
where nOut-of-plane is the index of refraction perpendicular to the plane of the substrate, and nIn-plane is the index of refraction in the plane of the film. Additionally, PPXC's birefringence was studied as a function of deposition temperature before and after anneal. The birefringence of PPXC and PPXN was also studied as a function of film thickness.
Birefringence of CVD polymer thin films
| Fig. 1. The repeat units of the polymers studied here. |
| Fig. 2. The optical birefringence is shown as a function of successive post-deposition anneals in N2 for 30 min. |
|
|
| Fig. 3. The optical birefringence is shown as a function of XRD peak intensity from the as-deposited condition to 290°C for PPXC and 400°C for PPXN and VT-4. |
Figure 2 shows the optical birefringence of PPXC, PPXN and VT-4 as a function of post-deposition anneals in nitrogen. Each anneal was 30 min. PPXC becomes more positively birefringent up to 290°C, and thereafter its birefringence drops. The melting point of PPXC is 290°C, at which point the film is structurally disrupted and becomes amorphous according to XRD measurements. The increase in birefringence may be associated with an increase in crystallinity (Fig. 3) where birefringence is plotted vs. XRD peak intensity. Similar results were obtained when birefringence was plotted vs. percent crystallinity; however, determining percent crystallinity is difficult, and hence rather large errors may result.
VT-4 and PPXN become more negatively birefringent after post-deposition anneals. The largest difference between VT-4 and PPXN is the crystallographic phase change that PPXN undergoes at 220°C from a-phase (monoclinic) to b -phase (hexagonal).7 In Figure 2, a large decrease in birefringence results when the b-phase starts to form. VT-4 is much more highly birefringent than PPXN as-deposited because of more crystallinity being present and the lack of a crystallographic phase transformation. The crystal structure of VT-4 is most likely hexagonal like b -PPXN.8
The three polymers studied have the same thermal stability, namely 400°C after successive anneals in nitrogen. At 400°C the birefringence increases for both VT-4 and PPXN (Fig. 2) because of the disruption of the crystalline phase and the breaking of the aliphatic carbon-carbon single bonds in the film. Therefore, birefringence, which is sensitive to the orientation of the benzene ring, is indicative of an increase in crystallinity, crystallographic phase transformations and thermal degradation.
One further point should be mentioned regarding Figure 3. The XRD peak intensity of PPXN and PPXC are close, which is indirectly related to the crystallite size and disorder in the crystalline unit cell. However, VT-4 has a significantly smaller XRD peak intensity, which is mostly due to paracrystallinity or disorder in the unit cell. This disorder may not be intrinsic to VT-4, but future work may be needed to investigate this phenomenon. Details concerning the morphological transformations of these polymers will be reported elsewhere.9
Dielectric anisotropy
The most important property to be extracted from the birefringence data is the dielectric anisotropy of the polymer films. The polymers studied here show a relatively high in-plane k (Fig. 4). For VT-4 and PPXN, the k reported from the birefringence data is accurate; however, PPXC's k at lower frequencies(<1 MHz) is higher than what is reported here because of the presence of chlorine. The out-of-plane index for an as-deposited film 295 nm thick is 3.13 at 1 MHz, 2.99 at 100 kHz and 2.88 at 10 kHz. The values reported here would be more appropriate at the higher operating frequencies of a typical ULSI device.
|
|
| Fig. 4. Out-of-plane (a) and in-plane (b) dielectric constants taken from the optical birefringence data as a function of successive post-deposition anneals. |
PPXC has a high in-plane k, but its out-of-plane k is still higher. This is reflected in its positive birefringence. VT-4 and PPXN in contrast have a low out-of-plane k. The more desirable material is one with a low in-plane capacitance, like PPXC, which will have a greater impact on interconnect delay.1
Two adjacent polymer chains or benzene rings will be most stable if the benzene rings lie in the same plane. This would be the case when the plane of the benzene rings are coplanar with the plane of the substrate, which would result in a negative birefringence as with VT-4 or PPXN. VT-4 and PPXN are symmetrical with respect to their repeat unit, and fluorine does not occupy much more space than hydrogen in the case of VT-4.
PPXC, in contrast, is asymmetrical with respect to its repeat unit and possesses a chlorine atom on the benzene ring, which occupies much more space. The C-Cl bond is longer (1.77 Å) than the C-F (1.36 Å) bond and the C-H bond (1.09 Å). The existence of the chlorine atom creates a greater distance between adjacent coplanar benzene molecules, destabilizing the coplanar structure and causing the benzene main-chain rings to be at an angle to one another. This noncoplanarity is a compromise between 100% positive birefringent orientation where little molecular polarizability exists to stabilize adjacent bonding and 100% negative birefringent orientation that is destabilized because of the presence of chlorine.
With this finding, the precursor species can be designed to produce a film that has a low in-plane k relative to its out-of-plane k. The use of a fluoropolymer such as VT-4 and AF-4 has potential drawbacks, such as adhesion and liberation of fluorine during metal reflow, but its k should be fully exploited. The data here show a high in-plane k for the three parylene polymers studied here. Most of the reduction in k that comes from polymer fluorination occurs in the out-of-plane k. Until enough understanding and insight is gained about the structure-property relations in polymer thin films, little gain may be made to improve their desired properties.
Thickness and deposition temperature effects
| Fig. 5. The optical birefringence of PPXC is shown as a function of the deposition temperature. The films were also annealed at 205°C for two hours in N 2. |
The deposition temperature is an important parameter since it, in conjunction with the sublimation rate of the precursor, controls the deposition rate of the polymer film. The deposition rate is a function of the adsorption and polymerization rates. At higher deposition temperatures (Fig. 5), the polymerization rate is high and the adsorption rate is rate controlling. Figure 5 shows a linear increase in the birefringence of the PPXC thin films as the deposition temperature is increased.
As the deposition temperature is increased, the polymeric chain becomes more conformationally ordered. This is reflected in the XRD data, which show increased crystallinity in the as-deposited condition for films deposited at successively higher temperatures above ~36°C, PPXC's Tg. 9 After the PPXC thin films were annealed at 205°C in N2 for two hours, all the films showed close to the same structure from the birefringence data of Figure 5. A slight slope exists in the birefringence as a function of deposition temperature even after anneal. However, the films deposited at lower temperatures did not possess the same structure as the films deposited at higher temperatures after anneal, according to the XRD data.9
It is well known that the structure of polymeric films changes from a thin film to an ultrathin film at around 100 nm.10 Any property that is dependent on long-range cooperative movement of atoms such as polymer crystallization and ferroelectricity will show ultrathin film effects. The birefringence of PPXN does not change as a function of thickness from ~800 nm down to ~112 nm, with a value of 0.055 to 0.06. However, below that critical thickness the birefringence rapidly increases to 0.08 at 60 nm. What effect thickness has on birefringence after post-deposition anneal has yet to be studied.
PPXN does not show a strong correlation between birefringence and deposition temperature because its Tg is below room temperature (~13°C). This also makes rearrangements and molecular motion possible after deposition. However, because the Tg is near room temperature, only slight changes in the morphology of the film may take place. Drastically different structure, like that found when PPXN is deposited at liquid nitrogen temperatures, may be "locked in" until high-temperature annealing takes place.
PPXC has a Tg above room temperature at ~36°C. When the film is at room temperature, the atoms do not have enough energy for cooperative motion to take place, so the structure is "locked in." Birefringence of PPXC, then, is a function of the deposition temperature. The films that showed the same thickness effect as the PPXN films were those deposited above ~36°C. The PPXC films deposited below its Tg did not show a thickness effect below 100 nm, but films deposited above the Tg showed the same thickness effect as PPXN.
The consequences of this finding still needs to be fully explored, but some hypotheses can be made. Films deposited below the Tg and not heated above it after deposition show a structure that is independent of the substrate. Films deposited above the Tg have time to rearrange during film growth. The substrate has a large impact on the orientation of the benzene rings, and therefore on the morphology of films that are <100 nm thick. This has been seen before in polymer crystallization studies and the effect of the substrate on the Tg of the film, as in the case of polystyrene spin-coated onto silicon.10 -12
Conclusion
Of the three CVD polymer thin films studied here, only PPXC is positively birefringent after post-deposition anneals. This is the desired condition, since the in-plane k is lower than the out-of-plane k, and the capacitance within the same metalization level is minimized. PPXN and VT-4 show morphological differences compared with PPXC, primarily because of the symmetry of the repeat unit and the strong asymmetry in PPXC. The chlorine atom in PPXC may destabilize adjacent coplanar benzene rings, adding a perpendicular component to their orientation, which causes a positive birefringence in the film.
The increase in the birefringence for PPXC and the decrease in birefringence for VT-4 and PPXN were directly related to an increase in crystallinity in the films. Reducing the crystallinity content for VT-4 and PPXN could drastically improve its dielectric anisotropy and in-plane k. Further gains in reducing the in-plane k for VT-4 and PPXN cannot be realized until this large negative birefringence is eliminated.
The structure of PPXC showed a strong dependence on the deposition temperature of the film. At lower temperatures, the polymer chain is conformationally disordered and becomes more ordered as the deposition temperature is increased. After the same films are annealed, most of the dispersion between birefringence and the deposition temperature is eliminated, but some structural differences still persist. PPXN and PPXC show a large increase in birefringence below ~100 nm, but PPXC only shows it with films deposited above its Tg.
References
1. S.-P. Jeng, R.H. Havemann, M.-C. Chang, Mat. Res. Soc. Symp. Proc. 337, 25-31 (1994).
2. S.P. Murarka, Solid State Tech, March, 83-90 (1996).
4. E.T. Ryan, A.J. Mckerrow, J. Leu, P. Ho, MRS Bulletin October, 49-54 (1997).
5. E. Saiz, U.W. Suter, P.J. Flory, J. Faraday Trans. (London) 73(9), 1538-52 (1977).
6. M.J.S. Dewar, J.J.P. Stewart, Chem. Phys. Lett. 111(4-5), 416-20 (1984).
7. M. Miles, H. Gleiter, J. Macromol. Sci. -Phys. B15(4), 613-7 (1978).
8. J.J. Senkevich, S.B. Desu, Appl. Phys. Lett. 72(2), 258-60 (1998).
9. J.J. Senkevich, S.B. Desu, Unpublished.
11. Y. Cohen, S. Reich, J. Polym Sci: Polym.. Phys. Ed. 19, 599-608 (1981).
12. W. Wu, J.H. van Zanten, W.J. Orts, Macromolecules 28(3), 771-4 (1995).
|
|
Jay J. Senkevich is a doctoral student in
materials engineering science at Virginia Polytechnic Institute.
He is expected to finish his degree late this summer. Phone: (540) 231-2107 Fax: (540) 231-8919 E-mail: jsenkevi@vt.edu |
|
|
Seshu B. Desu is a professor of materials
science and engineering and electrical and computer engineering
at Virginia Polytechnic Institute and director for the advanced
ceramic materials and thin films laboratory. Phone: (540) 231-6820 Fax: (540) 231-8919 E-mail: sdesu@vt.edu |