Vous êtes sur la page 1sur 5

RESEARCH NEWS

Perfluorocyclobutyl Copolymers for Microphotonics**


By Dennis W. Smith, Jr.,* Shengrong Chen, Suresh M. Kumar, John Ballato, Chris Topping, Hiren V. Shah, and Stephen H. Foulger
4 m

5 m

The copolymerization of aryl bis- and tris-trifluorovinyl ether monomers yields 30 m aromatic perfluorocyclobutyl (PFCB) polymers, via thermally initiated stepgrowth cycloaddition chemistry. PFCB polymers and their copolymers enjoy a unique combination of attributes well suited for applications in photonic technologies, such as broad tailorability of refractive indices and thermo-optic coefficients, low transmission losses at 1300 and 1550 nm, high thermal, mechanical, and optical stability, and excellent melt and solution processability. Planar PFCB structures can be processed by direct micro-transfer molding, which is a first step towards rapid soft-lithographic fabrication of polymer planar lightwave circuits. Copolymerization chemistry and processing parameters and characterization, including thermal (Tg = 120350 C) and optical properties (refractive indices from 1.443 to 1.508 at 1550 nm; thermooptic coefficients dn/dT = 7105 K1 to 1.5 104 K1), birefringence (< 0.003), and temporal stability of refractive index, are described.

1. Introduction
Organic polymers are increasingly attractive alternatives to inorganic materials in many photonic devices such as switches,[1,2] optical interconnects,[3] splitters,[4] and surface-relief structures.[5] As demand increases for greater bandwidth and component integration,[1,3] polymers offer low-cost fabrication, adequate transparency in the visible and near-infrared spectra, and versatility in structures and grades to cover a

[*] Prof. D. W. Smith, Jr., S. Chen, Dr. S. M. Kumar, Dr. C. Topping, Dr. H. V. Shah[+] Department of Chemistry and Center for Optical Materials Science & Engineering Technologies (COMSET), Clemson University Clemson, SC 29634 (USA) E-mail: dwsmith@clemson.edu Prof. J. Ballato, Prof. S. H. Foulger School of Material Science & Engineering and Center for Optical Materials Science & Engineering Technologies (COMSET) Clemson University Clemson, SC 29634 (USA) [+] Present address: Optical Coatings Laboratory, JDS Uniphase, Inc., Santa Rosa, CA 95407, USA.

[**] We thank the Defense Advanced Research Projects Agency (DARPA), the National Textiles Center, the National Science Foundation (CAREER Award DMR-9985160), Army Research Office, and South Carolina EPSCoR for financial support. We also thank Dr. E. Elif Gurel at Quantira Technologies, Inc. and Prof. G. Nordin at UAH for generous donation of the pattern silicon substrates and Mettler-Toledo for the donation of a DSC820 system to Clemson University. D. Smith is a Cottrell Scholar of Research Corporation.

broad range of property values, providing extensive device design flexibility. In particular, fluoropolymers represent leading alternatives to current optical polymers due to their many complementary properties, such as low transmission loss at the telecommunication wavelengths, good optical stability after thermal aging, and low moisture absorption.[1,6,7] For example, fluoroacrylates developed by Allied Signal and others,[1] Dupont's[8] Teflon AF (tetrafluoroethylene and perfluorovinyl ether copolymer), Amoco's[1] Ultradel (fluorinated polyimide), and Asahi's[9] CYTOP (perfluorovinyl ether cyclopolymer) are fluoroplastics currently pursued for both fiber and planar optical devices. However, commercial perfluoropolymers in general suffer from limitations such as poor processability using established fabrication techniques and other limitations that make it difficult to realize adhesion and mechanical integrity at interfaces, refractive indices matched to silica optical fibers, and the overall dimensional or thermomechanical stability required for higher-temperature applications. For example, the limitation of processability can be most critical when commercialscale production is considered. For a reasonable film thickness to be obtained from a single deposition when spin-coating for planar waveguides, solubility should typically exceed 50 wt.-% in common solvents suitable for optical and electronic device fabrication. Most fluoropolymers do not exhibit adequate solubility in common solvents desired for such solution processing of planar structures. Simple requirements such 1585

Adv. Mater. 2002, 14, No. 21, November 4

2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0935-9648/02/2111-1585 $ 17.50+.50/0

D. W. Smith, Jr. et al./Perfluorocyclobutyl Copolymers for Microphotonics

RESEARCH NEWS

as this severely limit the use of many F F F F F F F F fluoropolymers that are based on 150-200C F F F F F O Ar O F H3C F 5-25h chain-addition polymerization or F F y O O F F O Ar O bulk or 50% wt semi-fluorinated polyimide condenX F F in solvent F sation polymers mentioned above. F + Promising exceptions for processO F F F able low-loss polymer-integrated O F optics include polycyanurate materiFor Ar = H3C als[10] and fluorinated acrylates and 2 4 [11] copolymers thereof. F3C CF3 We have focused on the developF O O F ment of a unique fluoropolymer F F F F technology based on perfluorocyclo1 3 5 butyl (PFCB) aromatic ether repeatScheme 1. Copolymerization of selected trifluorovinyl ether monomers. ing units, for which thermomechanical robustness (i.e., Tg > 200 C) and The cycloaddition polymer chemistry also occurs without the solution processability are not compromised.[7,12] The general class of polymers were originally targeted for microelectronics preference to 1,2-disubstituted cis or trans stereochemistry. This contributes (along with other factors) to the fact that applications at The Dow Chemical Company.[13] Although PFCB polymers are typically amorphous and therefore exhibsome early work at Dow explored electro-optic (EO) PFCB it exceptional processability characteristics. polymers,[14] current efforts elsewhere have revitalized PFCB The relatively new hexafluoroisopropylidene-containing EO programs.[15] In addition, other derivativesfor example monomer (3) has also recently been reported, and provides those containing imide groupshave also been developed.[16] twice as much fluorocarbon per repeat unit than traditional Here we report the PFCB copolymerization strategy, and the PFCB polymers.[17] Polymer 4 contains the established first detailed structure/property relationship account offered for telecommunication applications. a-methylstilbene mesogen and forms lyotropic lamellar mesophases over a wide range of temperatures and molecular weights. Mesophase behavior probed by optical microscopy, 2. Synthesis atomic force microscopy (AFM), NMR, and X-ray scattering was recently reported.[12] New monomers and their PFCB polymers have also been prepared, for example, the CF3 anaInitially, we focused on preparing copolymers with precisely tailored optical, thermo-optical, and thermomechanical proplog of monomer 1, thereby replacing aliphatic CH segments, erties with long-term stability. For optical applications to date, which contribute to greater optical loss.[18] High-refractive-inwe have concentrated on the monomers (15) shown in dex monomers and polymers containing phenyl phosphine Scheme 1 with primary emphasis on monomers 13.[7] Monooxide linkages have also been prepared.[19] mers were prepared from commercially available phenolic Some selected copolymer experimental data are shown in starting materials in two steps as reported previously.[7] Table 1. Monomers can be polymerized in the melt, or dissolved in a variety of solvents suitable for commercial device fabrication. 3. Random Copolymerization Copolymer solutions were prepared by dissolving one or more monomers (5080 wt.-%) in freshly distilled and filtered mesitylene. The solution was stirred mechanically and degassed Random amorphous copolymers with tailorable refractive for 30 min with a dry nitrogen sparge. Polymerization was carindices, glass transition temperatures (350, 165, 120, 110, and ried out at 150 C for several hours, depending on the func180 C, respectively for monomers 15), and long-term 350 C tionality and viscosity desired. Exothermic formation of the thermal stability were accessed by simple choice of comonocyclobutyl linkages does not require catalysts or initiators nor mer structure (Scheme 1). A variety of techniques have demdoes the polymerization produce condensates or by-products. onstrated that random step-growth copolymers are produced
Table 1. Selected copolymer properties. Copolymer (w/w) wt.-% polymer in mesitylene 80 50 75 Polym. time [h] Mn [a] at 150160 C 2.5 8 3.2 589 1859 2508 Mw [a] Mw/ Mn [a] 2.06 2.5 1.93 Olefin conv. [%][b] 37.3 44.2 40 Viscosity RI Glass transition [Pa s][c] @1550 nm [d] Tg [C] [e] 0.133 6.8 0.1 1.5036 1.4785 1.4892 225 221 285

1-co-2(50:50) 1-co-3(70:30) 1-co-2 (80:20)

1212 4656 4846

[a] GPC in chloroform vs. PS. [b] 19F NMR. [c] TA AR1000-N Rheometer at 25 C. [d] Prism Coupler at 25 C. [e] DSC at 10 C min1 in N2.

1586

2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0935-9648/02/2111-1586 $ 17.50+.50/0

Adv. Mater. 2002, 14, No. 21, November 4

D. W. Smith, Jr. et al./Perfluorocyclobutyl Copolymers for Microphotonics

RESEARCH NEWS

from PFCB polymer chemistry, including GPC, NMR, IR, Raman, and DSC.[7] Melting endothermic transitions and subsequent exothermic polymerization beginning near 150 C (at 10 C min1) indicate similar monomer reactivity.[20] Copolymer and terpolymer reactions show the same behavior by DSC. Similar rates of polymerization and copolymerization were measured by 19F NMR for monomer 1 and 2.[7] An example of PFCB random copolymer formation is shown in Figure 1, expressed as the glass transition temperature change as a function of composition for the network copolymerization of monomer 1 and 5. Clearly, random PFCB

350

Tg Observed Tg Predicted by Fox eq.

300

250

a branched copolymer of variable conversion, and thus molecular weight and viscosity can be controlled prior to thermal cure to a cross-linked network. For example, PFCB copolymer solutions in excess of 80 wt.-% polymer can be easily obtained, which enables the preparation of reasonably thick films (20+ lm) in a single deposition. We have found that processing parametersand thus optical polymer propertiescan be precisely controlled by choice of co-monomer and polymerization time and temperature. Upon spin-coating, the cyclopolymerization is progressed by heating under inert atmosphere to the desired degree of conversion. Functional group (f) conversion at gelation (Pgel) has been shown to follow Flory's theory of rubber elasticity (Pgel = 1/f1), therefore the extent of conversion can be precisely controlled given the composition at zero time.[20] Crosslinking chemistry can then be tailored to give fast cure with minimal conversion for etch resistance, or high Tg (i.e., dense cross-linking) for higher performance applications. Table 2 illustrates selected conditions and properties of PFCB copolymers targeted for optical applications.
Table 2. Summary of selected property ranges for PFCB copolymers. Condition or Property Typical Controlled Range 50 90 wt.-% 1 20 lm 140 160 C / 1 10 h 1,200 30,000 Mw 1.2 10 Mw/Mn 0.02 100 Pa s 120 350 C / 0.1 3 h 175 350 C Micromolding or RIE < 0.25 dB cm1 1.443 1.508 <0.003

Tg (C)

200

150 0 10 20 30 40 50 60 70 80 90 100

% monomer 1 in poly 1-co -5

Fig. 1. Glass transition (determined by DSC) of polymer 1co5 vs. composition.

copolymerization obeys the Fox equation (1/Tg12 = wt1.-%/ Tg1 + wt2.-%/Tg2) as is expected for equally reactive monomers.[21] Slight differences in reactivity are observed for highly deactivated arenes (containing electron-withdrawing groups), yet this data is preliminary and the subject of further study. Mechanical properties for some polymers have also been reported.[7] Free-standing optically clear polymer films or molded plaques can be prepared by melt-mixing variable composition monomer mixtures and heating at 150200 C under an inert atmosphere. High-molecular-weight homopolymers or pregelled copolymers could also be micromolded to form submicrometer waveguides and gratings (see below).[22] One of the most useful copolymer compositions has been poly1co2 (50 wt.-% composition), which has a measured Tg = 220 C, whereas the theoretical Tg value calculated from the Fox equation is 224 C (Table 1).[13] The ability to spin-coat very thick films is a unique feature of high-performance PFCB polymers, and thereby ensures a homogeneous layer in the final structure versus multiple spin and cure cycles otherwise needed to achieve thicknesses necessary for planar waveguide applications. Copolymer solutions suitable for spin-coating thick films were prepared in mesitylene solution, yet other solvents such as other high-boiling-point hydrocarbons, 1-methyl-2-pyrrolidinone (NMP), dimethyl sulfoxide (DMSO), etc., can also be used. For optical device fabrication, most copolymers studied to date include trifunctional monomer 1, so that the advanced intermediate is

Copolymer solution concentration Single spin coat thickness Copolymerization temperature / time Weight average molecular weight (GPC) Molecular weight distribution (GPC) Solution viscosity (RMS) Cure temperature / time Glass Transition Temperature (DSC) Patterning Technique Loss at 1550 nm Refractive index at 1550 nm Birefringence at 1550 nm (gTM-gTE)

4. Optical Characterization
Optical absorption for homopolymers 1, 2, and 3 was studied in the UVinfrared, ranging from ~ 200 nm to 3200 nm.[21] Of particular note is the lack of absorption peaks in the wavelength range of interest for telecommunication applications. With the exception of weak absorptions at ~ 1150 nm, 1400 nm, and a stronger one at 1700 nm, PFCB polymers, in general, show great potential for low-loss photonic applications. The commercially significant spectral windows of 1300 nm and 1550 nm are essentially void of molecular vibrations for PFCB polymers, whereas transparency within this energy range is typically inaccessible to hydrocarbon-based polymers due to strong CH harmonic absorptions. Many other factors, of course, also affect the observed optical loss of polymers, including birefringence, stress, interfacial defects, and impurities.[23] Optical attenuation measurements for copolymers designed specifically for waveguides have been measured using prism coupling techniques, and values typically approach the report-

Adv. Mater. 2002, 14, No. 21, November 4

2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0935-9648/02/2111-1587 $ 17.50+.50/0

1587

D. W. Smith, Jr. et al./Perfluorocyclobutyl Copolymers for Microphotonics

RESEARCH NEWS

ed value of 0.25 dB cm1 at 15151565 nm for the homopolymer from monomer 1.[23,24] Actual waveguide loss at 1550 nm has also been measured for selected compolymers by industrial partners and was found to be consistently and significantly lower than that reported for the homopolymer from 1, the details of which will be published elsewhere. The lower loss of PFCB copolymer waveguides is expected due to the decrease of aliphatic groups present in 1 and the intimate compatibility at the core/clad interface for waveguide structures made solely from PFCB copolymers. The dependence of refractive index on wavelength up to 800 nm has been reported for the homopolymers 13 and the (1:1 by weight) copolymers 2co1 and 2co3.[21] Figure 2a gives the refractive index data at 1550 nm and at 630 nm as a
a) 1.555
co -2 (633 nm)

der study. Our initial results are also shown in Figure 2b where the refractive index is plotted as a function of cure time at 200 C for monomer 1. As is shown, the refractive index changes dramatically during cure yet quickly stabilizes in less than 2 h at 200 C. Furthermore, the refractive index is completely stable after multiple heat/cure cycles and continued heating at 200 C. The temperature dependence of refractive index has also been studied for the copolymer 2co1 with different monomer ratios (Fig. 3). The dn/dT response can be tailored as a function of copolymer composition where, to date, values
1.525 1.520 dn/dT= -1.4x10 1.515 1.510 dn/dT= -1.5x10
-4 -4

1.535
RI
1.505 1.500 1.495 dn/dT= -1.4x10 1.490
-4

1.515

co -2 (1550 nm)

dn/dT= -1.2x10

-4

RI

1.495 1.475 1.455 1.435 0


co -3 (1550 nm)

co -3 (633 nm)

1.485 10 20 30 40
o

50

60

70

80

Temperature ( C)

10

20

30

40

50

60

70

80

90

100

Fig. 3. Refractive index vs. temperature for polymer 2co1 (1:1 by weight, squares) and 2co1 (1:4 by weight, triangles) at 633 nm (dotted line) and 1550 nm (solid line).

% monomer 1

b) 1 .54
1 .52
RI

R I@1550nm

R I@633nm

1 .5 1 .48 1 .46 0 1 2 3 4 5 6 7 8

Time (h)
Fig. 2. a) Refractive index vs. copolymer compositions. b) Stability of polymer 1 vs. cure time at 200 C at 633 and 1550 nm.

function of copolymer composition. Refractive index tailorability for combinations of the three monomers ranges from 1.443 to 1.508 at 1550 nm. Birefringence was also measured (gTEgTM) to be < 0.003. The resolution of this technique is, however, admittedly limited, and other techniques such as the polarization splitting method should be considered.[1] The evolution and stability of optical properties as polymerization proceeds is also a major concern for processing and performance of polymer integrated optics. As is the case for most fluoropolymers, moisture penetration in PFCB optical polymers is not an issue, even at elevated temperatures and high relative humidity. The stability of network optical properties over time and during multiple heat/cool cycles is currently un-

from 7 105 K1 to 1.5 104 K1 have been measured.[25] This broad range in the thermo-optic (TO) coefficient indicates that PFCB copolymers can exhibit a lower or higher (negative) dn/dT value than typical fluorinated acrylates and other amorphous fluoropolymers. The lower dn/dT values provide stability for non-TO devices, whereas the larger TO coefficient offers the potential for realizing TO switches, modulators, and interconnects with lower power consumption than with other materials.[26] A complete thermo-optic characterization study for PFCB copolymers is underway. Due to the excellent processability of PFCB polymers, we have recently pursued micro-transfer molding of waveguide structures. Copolymers can be solution or melt microfabricated via standard methods, and can also be processed by soft-lithography techniques. Specifically, themoplastic monomers have been used to fabricate optically diffractive linegratings with 0.5 lm feature sizes by direct micromolding using only a photolithographically generated silicon master.[22] Figure 4 exhibits surface scanning electron microscopy (SEM) images and cross section (inset) for the polymer film with transferred grating, as well as a photograph of the film cast and removed from a 4 in. (1 in. 2.54 cm) silicon wafer. Strong reflectivities from the green to red portions of the visible spectrum, depending on incident beam angle, were observed with 3 dB bandwidths of approximately 30 nm. This negative mold-free technique permits feature reproduction at submicrometer size scales, eliminates several steps from

1588

2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0935-9648/02/2111-1588 $ 17.50+.50/0

Adv. Mater. 2002, 14, No. 21, November 4

D. W. Smith, Jr. et al./Perfluorocyclobutyl Copolymers for Microphotonics

RESEARCH NEWS

a)

4 m

variable film thickness, high thermal stability, variable dn/dT, and stable optical properties. Excellent versatility in processing and property control is provided by the PFCB copolymerization strategy.
[1] L. Eldada, L. Shacklette, IEEE J. Sel. Top. Quantum Electron. 2000, 6, 54. [2] U. Siebel, R. Hauffe, K. Petermann, IEEE Photonics Technol. Lett. 2000, 12, 40. [3] B. Lee, M. Kwon, J. Yoon, S. Shin, IEEE Photonics Technol. Lett. 2000, 12, 62. [4] M. Oh, M. Lee, H. Lee, IEEE Photonics Technol. Lett. 1999, 11, 1144. [5] N. Stutzmann, T. Tervoort, C. Bastiaansen, K. Feldman, P. Smith, Adv. Mater. 2000, 12, 557. [6] E. Nihei, T. Ishigure, N. Tanio, Y. Yasuhiro, IEICE Trans. Electron. 1997, E80-C, 117. [7] D. W. Smith, Jr., D. Babb, H. Shah, A. Hoeglund, R. Traiphol, D. Perahia, H. Boone, C. Langhoff, M. Radler, J. Fluorine Chem. 2000, 104, 109. [8] P. R. Resnick, W. H. Buck, in Modern Fluoropolymers (Ed: J. Scheirs), Wiley, New York 1997, p 397. [9] N. Sugiyama, in Modern Fluoropolymers (Ed: J. Scheirs), Wiley, New York 1997, p. 541. [10] C. Dreyer, M. Bauer, J. Bauer, N. Keil, H. Yao, C. Zawadzki, Microsyst. Technol. 2002, 7, 229. [11] N. Keil, C. Weinert, W. Wirges, H. Yao, S. Yilmaz, C. Zawadzki, J. Schneider, J. Bauer, M. Bauer, K. Losch, K. Satzke, W. Wischmann, J. Wirth, Electron. Lett. 2000, 36, 430. [12] R. Traiphol, H. V. Shah, D. W. Smith, Jr., D. Perahia, Macromolecules 2001, 34, 3954. [13] D. A. Babb, B. Ezzell, K. Clement, W. Richey, Kennedy, J. Polym. Sci., Part A: Polym. Chem. 1993, 31, 3465. [14] M. Newsham, M. Mang, R. Gulott, Jr., D. W. Smith, Jr., US Patent 576 374, 1998. [15] H. Ma, J. Wu, P. Herguth, B. Chen, A. Jen, Chem. Mater. 2000, 12, 1187. [16] W. S. Choi, F. W. Harris, Polymer 2000, 41, 6213. [17] D. W. Smith, Jr., H. Shah, B. Johnson, Am. Chem. Soc. Div. Polym. Chem., Polym. Prepr. 2000, 41, 60. [18] S. Kumar, E. Nelson, S. Chen, J. Ballato, S. Foulger, D. W. Smith, Jr., Am. Chem. Soc. Div. Polym. Chem., Polym. Prepr. 2001, 42, 501. [19] D. Babb, H. Boone, D. W Smith, Jr., P. Rudolf, J. Appl. Polym. Sci. 1998, 69, 2005. [20] M. Cheatham, S. N. Lee, J. Laane, D. W. Smith, Jr., D. A. Babb, Polym. Int. 1998, 46, 320. [21] D. W. Smith, Jr., A. Hoeglund, H. Shah, M. Radler, C. Langhoff, in Optical Polymers (Ed: J. Harmon), ACS Symp. Ser., Vol. 795, American Chemical Society, Washington, DC 2001, Ch. 4, pp. 4962. [22] H. Shah, D. W. Smith. Jr., J. Ballato, S. Foulger, P. Deguzman, G. Nordin, IEEE Photonics Technol. Lett. 2000, 12, 1650. [23] R. Norwood, R. Gao, J. Sharma, C. Teng, Proc. SPIEInt. Soc. Opt. Eng. 2000, 44, 2001. [24] G. Fishbeck, R. Moosburger, C. Kostrzewa, A. Achen, K. Petermann, Electron. Lett. 1997, 33, 518. [25] E. E. Gurel, presented at Optics in the Southeast, Clemson, SC, October 2001. [26] C. H. Jang, L. Sun, J. H. Kim, X. Lu, G. Karve, R. T. Chen, J. Maki, IEEE Photonics Technol. Lett. 2001, 13, 490.

5 m

30 m
b)

Fig. 4. Field emission scanning electron micrographs (SEM) and photograph of polymer 2 diffraction grating prepared by direct micro-transfer molding.

conventional soft-lithographic methods, and marks itself as a practical means for rapidly generating planar photonic structures that operate spectrally in the visible and the near-IR telecommunication bands.

5. Conclusions
The thermal cyclopolymerization of aromatic trifluorovinyl ether monomers to aromatic perfluorocyclobutane polymers and copolymers affords high-temperature, low-transmissionloss materials with tailored optical properties for potential use in telecommunication devices. Branched pre-network copolymer solutions can be spin-coated to give low loss at 1550 nm,

______________________

Adv. Mater. 2002, 14, No. 21, November 4

2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

0935-9648/02/2111-1589 $ 17.50+.50/0

1589

Vous aimerez peut-être aussi