highly nonlinear chromophore
Min-Cheol Oh, Hua Zhang, Cheng Zhang, Hernan Erlig, Yian Chang, Boris Tsap, Dan Chang,
Attila Szep, William H. Steier, Harold R. Fetterman, and Larry R. Dalton
Based on a nonlinear optical polymer with a highly nonlinear chromophore (CLD) dispersed in a amorphous polycarbonate (APC), we have developed electro-optic (EO) polymer modulators operating at 1550 nm wavelength with low loss and good thermal stability. By incorporating polymer insulation layer, push-pull poling was successfully performed without film damages. We also demonstrated the propagation loss of the EO polymer waveguide could be reduced as low as 1.2 dB/cm at 1550 nm when the large core waveguide structure was incorporated. The long-term reliabilities of EO polymer modulator made of CLD/APC polymer were investigated. When the modulator was hermetically sealed in an inert gas, the Vp change of Mach-Zehnder modulator was negligible over 30 days of operation with 20-mW exposure to the waveguide input. In the thermal stability measurement, 25% of Vp increase was observed from the sample heated to 60 ºC over 40 days, though the sample left at room temperature showed no decay of nonlinearity.
Min-Cheol Oh, Cheng Zhang, Hernan Erlig, Yian Chang, and Boris Tsap are with Pacific Wave Industries, Los Angeles, CA 90024. Hua Zhang, Attila Szep, William H. Steier, and Larry Dalton are with the Department of Electrical Engineering and Chemistry Department, University of Southern California, Los Angeles, CA 90089-0483. Dan Chang and Harold R. Fetterman are with the Department of Electrical Engineering, University of California Los Angeles, Los Angeles, CA 90095
Polymer electro-optic (EO) waveguide devices have been investigated for over a decade and there have been various experimental results to demonstrate the merits of EO polymer devices. The demonstrations have included high-speed modulators based on an inherent velocity match of RF and optical waves , integration of electronic circuits with polymer waveguides , multilevel-stacked integrated optical circuits , and polarization controlling waveguide devices . However, to be accepted in system applications, these EO polymer devices must demonstrate modest insertion loss, good long-term thermal stability, and sufficient photo-stability at the communication wavelengths.
Typical EO polymers have higher absorption loss than LiNbO3, especially at 1550 nm wavelength, due to the C-H vibration overtones. There has been recent work which shows the losses can be decreased by the substitution of fluorine or chlorine for the hydrogen . The thermal stability of the EO polymer materials has been improved by employing a high Tg backbone polymer, a side chain attached chromophore, and a crossliked polymer matrix [9-11]. However, in these polymers with an enhanced thermal stability, the efficiency of poling was reduced and the EO coefficient was sacrificed to improve the thermal stability. Recently an EO polymer based on a dendritic structure has demonstrated both high EO coefficient and excellent poled order stability at 85 C . The photo-stability of EO polymer waveguides under 1300 nm wavelength illumination has been investigated and it was found that some EO polymer materials degrade due to the chemical reaction with excited oxygen . The dependence of the photostability on the structure of the EO polymer and its backbone was investigated , and a double crosslinked chromophore was incorporated to demonstrate excellent photostability .
To develop an EO polymer material suitable for commercial devices, the material synthesis research has been directed toward the optimization of several requirements simultaneously. These include reduced insertion loss, long-term thermal stability at 85 ºC, and long term photo-stability at 10 mW input power at the communication wavelengths while at the same time maintaining a high electro-optic coefficient. There have been significant advances on all of these issues based on a guest-host polymer consisting of highly nonlinear chromophore with phenyltetraene bridge (CLD)  and amorphous polycarbonate (APC) . This chromophore was mixed in a PMMA host to achieve a push-pull modulator with a Vp of 0.8 V @ 1300 nm . In addition to this highly nonlinear chromophore, APC was identified as a promising host material because of its high thermal stability and low loss at 1550 nm wavelength . >From the simple guest-host mixture of CLD and APC, we obtained short term thermal stability of 120 ºC and waveguide optical propagation loss of 1.7 dB/cm @ 1550nm . In addition, the photo-stability of devices made with CLD/APC can be significantly improved via a hermatic packaging which removes the reactive oxygen. In this paper we will review the advances made based on the CLD/APC EO polymer which demonstrate that the polymer devices are now becoming a commercially viable device.
II. EO polymers and Cladding materials
The recent progress to improve the nonlinearity of the chromophore has been accelerated by the invention of a novel tricyanobutadiene acceptor with a furan-derivative ring. Fig. 1 shows the structure of three chromophores with the novel acceptor, and Table 1 summarizes the physical properties and the measured nonlinearity of the chromophores. FTC was the first chromophore which exhibited the high nonlinearity induced by the efficient electron acceptor, 2 - (3 – cyano - 4,5,5 – trimethyl - 5H – furan – 2 - ylidene) - malononitrile. The three cyano-groups in the acceptor are aligned along the charge-transfer direction of the chromophore, thus they can pull the electron more efficiently so as to provide relatively large dipole moment (12-15 Debye). The two butyl groups on the thiophene increase the solubility and reduce the interchromophore electrostatic interaction which can lead to aggregates. The FTC chromophore when doped into a PMMA host gives an EO coefficient r33 of 57 pm/V at 1060 nm. By using the FTC chromophore dispersed into polyurethane, an ordinary MZ modulator with a 2-cm electrode was fabricated and Vp of 4.5 V was obtained at 1300 nm.
The basic structure of the CLD chromophore is obtained by replacing the thiophene unit in FTC with a diene. The aromatic resonance energy of the bridge is reduced to increase the nonlinearity of CLD. The two bulky t-butyldimethylsilyl groups used in CLD-1 enhance the molecular cross section so that aggregates are less likely to form, and they also reduce the interchromophore interaction during the poling process so as to increase the poling efficiency . As a comparison with FTC, when CLD-1 was doped into PMMA an r33 of 88 pm/V @ 1060 nm was obtained. When CLD-1 was used in the APC host, r33 of 90 pm/V @ 1060 nm was obtained. In a single-arm 2-cm MZ modulator made of the CLD-1/APC, a Vp was reduced to 2.4 V at 1300 nm. The loss at 1550 nm of the CLD/APC polymer was measured using the liquid immersion technique  and found to be 1.2 dB/cm
The effects of chromophore geometry on the optimum loading density and the poling efficiency have been investigated with another modified chromophore CLD-72. The EO coefficient obtained from CLD-72 is higher than that of CLD-1 by ~ 40% though the two chromophores have the same p-conjugate system and approximately the same absorption peak. It is believed that the two TBDMS bulky groups of CLD-1 while reducing the interchromophore interaction also restrict the poling efficiency by making it more difficult for the chromophore to rotate within the host polymer. Due to the limited solubility and poor quality of coated film, CLD-72 was not evaluated in a MZ modulator.
Cladding materials must satisfy a number of requirements, including higher electrical conductivity than the core, lower index of refraction than the core, good adhesion to the adjacent materials, low optical loss, and enough temperature stability to withstand the poling temperatures. An UV curable epoxy available from Masterbond Co., UV15, was found to be a suitable lower cladding for CLD/APC. The UV15 was spin-coated on a Au coated silicon wafer and cured by a UV lamp. The films were post baked to improve the crosslinking and to eliminate small molecules remaining in the UV cured film. The resistivity of the lower cladding was important in determining the efficiency of the poling. With UV15 lower cladding layer, we achieved ~ 75% of the poling efficiency compared to the result of CLD/APC single layer poling. The loss measured using the liquid immersion technique was 4.2 dB/cm at 1550 nm.
Because CLD/APC can be easily dissolved by organic solvents, the upper cladding should not contain any solvent. Moreover we found some polymers even without the solvent still can dissolve CLD/APC layer due to the small molecules. We have identified UFC170 polymer  as an appropriate upper cladding material. It has been designed to preclude small molecules that can penetrate and dissolve the surface of the CLD/APC core layer. UFC170 is curable with a relatively small dosage of UV light to form a thin film with good hardness. In addition, cured UFC170 is not attacked by the solvents used in photoresists, is compatible with Cr-Au upper electrode, and has an electrical resistivity which allows for efficient electrode poling of the CLD/APC. However, the UFC170 is not appropriate for the lower cladding layer because of the weak adhesion to the Au lower electrode. The measured loss of UFC170 at 1550nm is 3.1 dB/cm. The UV15:CLD/APC:UFC170 structure is robust and the end faces can be cut with a dicing saw to provide good optical coupling to the waveguide.
III. Design of rib waveguide and RF strip line
The waveguides used in polymer modulators have buried rib structure fabricated by dry etching in oxygen and they must be single mode to provide a high extinction ratio of the modulator. EO polymers have a relatively high index of refraction compared to the passive cladding polymers. Hence, in order to keep the waveguide single mode, it has been widely accepted that the thickness of the core layer has to be thin enough to suppress higher order modes in its slab waveguide. At 1550 nm this requires the typical thickness of the EO polymer layer to be less than 1.5 mm. However, the small vertical mode size in the thin waveguide leads to a high coupling loss between the optical fiber and the waveguide. Moreover, when the core layer is thin, the rib height is small for single mode operation and the fabrication tolerance is tight because the effective index change is so sensitive to the core thickness and the etch depth. As a result, it is difficult to maintain the required fabrication tolerance to achieve a high extinction ratio in the modulator.
In case of three dimensional rib waveguides, it has been shown that the single mode condition does not require the thin core layer . Even if the core layer is so thick that it could support several modes in its slab waveguide, it is still possible to design the rib waveguide to confine only one mode by radiating the higher order modes horizontally into the slab modes . Then the oversized rib waveguide will support only the single mode without additional radiation loss .
This approach can be used to design large core single mode waveguides in EO polymer devices. Fig. 2 shows the effective index of the planar waveguide as a function of the thickness of core layer. The refractive indices used in this calculation are 1.504 for the UV15 lower cladding, 1.612 for the CLD/APC core layer in the TM polarization, and 1.488 for the UFC170 upper cladding. The thickness of the core layer effects the EO modulation efficiency by changing the film thickness across the electrodes and also effects the fiber coupling loss to the waveguide because of the mode size mismatch. In Fig. 2, we have selected 3 mm as a reasonable compromise. The rib height DT is selected to obtain sufficient confinement in the lateral direction due to the effective index contrast DNeff, which is the difference of the effective index of waveguide side N0(side) and that of waveguide center N0(center) of the lowest order mode. For 6-8 mm wide waveguides, a DNeff of ~ 0.003 is required to produce the single mode waveguide with good confinement. From Fig. 2, one sees that waveguide side needs to be etched by 0.4 mm to provide DNeff of ~ 0.003 for the 3-mm thick waveguide center.
Also from Fig. 2, it is clear that the 3-mm slab waveguide can support higher order modes. However, the higher order mode will not propagate through the rib waveguide because the effective index of the first higher order mode in the 3-mm slab waveguide, N1(center) is lower than that of the fundamental mode in the 2.6-mm slab waveguide N0(side). The higher order mode supported by the 3-mm slab waveguide will be coupled into the fundamental mode of the 2.6-mm slab waveguide. In consequence, the over-sized rib waveguide will satisfy the single mode condition due to the radiation of higher order modes. This design has been verified by the high extinction ratio of the fabricated MZ modulators.
The minimum thickness of the cladding layers is determined by setting the metallic absorption loss due to the surface plasmon coupling to the upper or lower electrodes to an acceptable value. Making the cladding layers thicker than this minimum only increases Vp. To find the minimum thickness, the effective index of the multilayer planar waveguide structure including the two Au electrode layers is calculated. The imaginary part of the effective index is then used to obtain the metallic absorption loss. As shown in Fig. 3, the losses are calculated as a function of the upper cladding layer thickness for several core layer thicknesses. The thickness of the lower cladding layer is fixed at 2.5 mm throughout the calculation. From Fig. 3, one sees that the metallic loss is less than 0.1 dB/cm when the cladding thickness is over than 2 mm for the 3-mm core waveguide. It is also shown that for the same cladding thickness the thicker core layer has the lower metallic loss due to the better mode confinement into the core layer.
The total thickness of the waveguide consisting of lower cladding, core and upper cladding should be as thin as possible to minimize Vp. For the polymers used in the modulators, the minimum thickness for negligible metallic loss is approximately 8 mm for the 3-mm core waveguide. The upper and lower claddings are 2.5-mm thick. Even if the thickness of the core is reduced less than 3 mm, the total thickness of the device cannot be decreased because the thinner core waveguide requires the thicker cladding layers.
The high-speed electrode structure for the push-pull modulators, shown in Fig. 4, incorporates the two-arm microwave transmission line . The line is designed using commercial microwave design software. In the push-pull modulator, the arms of the MZ modulator are poled in the opposite directions and the rf electrode covers both arms to apply the rf electric field in the same direction on each arm. To achieve 50-W characteristic impedance, the width of the electrodes over each arm should be set to give 100-W impedance to each arm. However, the width of strip line cannot be reduced to less than 8 mm in order to maintain reasonable overlap of the rf field to the guided mode in 6-mm wide waveguides. In addition, the tolerance of alignment would be too small to reliably obtain good alignment if the strip line width is close to waveguide width. Consequently, the characteristic impedance of the strip lines over the MZ arms with the 8-mm width and 7.5-mm polymer thickness becomes 70 W. The two strip lines on MZ arms are connected to the wider strip line that has the impedance of 35 W with a width of 28 mm. Those lines are connected by the right angle bend without chamfers at the corner. According to the simulation, the reflection from the right angle bend is negligible and the total reflection is below -25 dB. The predicted loss coefficient for the line using Au with 4.1x107 S/m conductivity is 0.48. For the 2.6-cm long transmission lines including the RF feeding lines, the S21 is supposed to be -7.9 dB at 40 GHz. We assumed that the dielectric loss of the polymers is low  and the loss is dominated by metallic losses.
IV. Device Fabrication and Poling
A schematic diagram of the push pull polymer modulator fabrication procedure is shown in Fig. 5. The bottom electrode is made of Cr and Au with a thickness of 0.1 and 3.0 mm, respectively. UV15 was spin-coated on the bottom electrode as the lower cladding. A disposable filter with a 0.2-mm pore was used to filter the solution. The UV15 was cross-linked and hardened as described earlier.
To prepare the guest-host polymer solution for the core layer coating, the CLD chromophore was mixed with the APC polymer in the optimum concentration of 25 wt% . It was then dissolved in dichloroethane to make 10 % solution in a volume ratio. The solvent has a low boiling point (82 ºC) which requires the spin coating of the polymer solution to be done in a short time to prevent wrinkles on the film caused by the inhomogeneous drying of the solvent. After spinning, the 3-mm core layer was dried in a vacuum for one hour then baked at 120 ºC for one hour to remove the residual solvent.
The waveguide ridges are patterned by photo-lithography and reactive ion etching (RIE) in oxygen. One of the difficult issues in the fabrication of multi-layer polymer devices is the solvent attack on the first layer during the second layer coating. This is difficult to prevent in the guest-host EO polymers which are not crosslinked and which can be dissolved by the solvents used in the standard photolithography process. In order to overcome this problem, we used a sacrificial protective layer of UV15 polymer diluted with methanol to be 20 wt%. Methanol does not dissolve the CLD/APC polymer. The protective UV15 layer was spin coated to give ~0.5 mm thickness and then UV cured.
Standard photolithography was used on the protective layer to define the waveguide patterns. The waveguide patterns were etched by O2 RIE to have a rib height of 0.5 mm. The rib width was 6 mm. To minimize the roughness of the etched surface, a largely chemical etch was achieved at 450 mtorr with an RF power density of ~30 mW/cm2. After removing the residual photoresist using a developer, the entire surface of the sample was etched again to transfer the waveguide pattern into the CLD/APC core layer and to remove the remainder of the sacrificial layer. The final rib height on the CLD/APC layer became 0.4 mm due to the slower etch rate of CLD/APC compared to that of UV15. For the upper cladding, UFC170 polymer was filtered and spin coated to obtain 2.5-mm cladding thickness on top of the rib. After the UV exposure to cure the polymer, the polymer was post-baked. No cracking was occurred in the films of the multi-layer polymer devices.
The Au poling electrodes were patterned by photolithography. For the push-pull operation, each arm of the MZ modulator must be poled in opposite directions. In this configuration, there is substantial electric field between the two top electrodes and unless the electrodes are covered there is a likelihood of avalanche breakdown across the surface at the poling temperatures . To prevent the breakdown, we coated the surface with a layer of the same APC polymer used for the host material of the EO core layer. With the APC layer, the current flow through the top electrodes could be kept below 1 mA during the entire poling process. Other steps to prevent the surface breakdown such as poling in SF6 and coating with AZ5214 photoresist were not successful. The SF6 gas was not effective in preventing the breakdown at the high temperatures used in the poling. The AZ5214 photoresist was effective but it was difficult to remove after the high temperature baking during the poling.
When the electrode poling was done in air, we observed a high poling induced loss in the waveguides. There was a significant change of the mode confinement after the poling due to the refractive index drop of the EO polymer. We believe this is caused by a reaction between the dissolved oxygen in the polymer and the chromophores at high temperature with high electric field. A similar effect was observed during the corona poling . To find the role of oxygen, straight waveguide samples were prepared and poled in either air or nitrogen by applying 600 V at 140 ºC for 20 minutes. The insertion loss of the samples was measured before and after the poling. The poling induced loss measured from 2-cm long waveguides was as high as 4-7 dB when the sample was poled in air.
No poling induced loss was observed when the sample was poled in nitrogen. We also measured the insertion loss of the unpoled waveguides that were located adjacent to the poled waveguides on the same substrate. Even though both waveguides were heated to the same temperature, there was no increase of the loss in the waveguide without the poling electrodes. It therefore appears that the assistance of poling current is necessary to initiate the oxidation of CLD chromophore at the poling high temperature. According to our observations, about 30 min. at 120 °C was sufficient to purge the dissolved oxygen from the polymer and to prevent the oxygen-current induced degradation.
To determine the optimum electrode poling conditions for the multi-layer device structure we fabricated several MZ modulators with CLD-1/APC using different poling conditions and the r33 @ 1550nm was calculated from the measured Vp of the modulators as shown in Fig. 6 (a). An r33 of ~43 pm/V could be attained using poling voltages from 300 V to 400 V at poling temperatures of 140-150°C. The r33 dependence on the poling time was also measured from MZ modulators. Fig. 6 (b) shows that at a temperature of 140 ºC with 400 V the maximum r33 is achieved after 20 min. of poling.
Based on these measurements the poling procedure used in the push-pull modulators is determined as shown in Fig. 7. The total applied voltage was 800 V (each arm 400 V) to give an estimated poling electric field of 50 V/mm if the poling field is assumed to be uniform across the device. The poling temperature was 145 ºC and the poling time was 30 min. After the poling, the insulating APC polymer layer coated on top of the poling electrodes was removed using acetone then the poling electrodes were etched away by a chemical etching.
To fabricate the modulation RF electrodes, a seed layer of Cr and Au were evaporated using an e-beam evaporator. On the metal seed layer, 3-mm thick photoresist was depositied to pattern the upper electrode of the RF strip line. To prevent the relaxation of the poling, the resist was dried in vacuum at room temperature instead of standard high temperature baking. The upper electrode was plated to ~ 3 mm thick Au. The end-facet of the waveguide was diced with a nickel blade. The polymers used in the device were sufficiently hard for the dicing to provide optical quality end faces.
V. Device Performance
Cut-back measurements at1550 nm were made to measure the propagation loss of the waveguides. The input single mode fiber was aligned to the waveguide and the output was collected by 20x lens and coupled to a detector. A fiber-optic polarization controller was used to adjust the input polarization to TM. To increase the accuracy of the measurement, several waveguides were measured at each length. Fig. 8 shows the results of the measurement for two waveguides with different core thickness. The 3-mm core waveguides were identical to those used in the modulators. To obtain single mode waveguides with the 5-mm thick core layer, the sample was etched to make a rib with a height of 1.3 mm. From the small scattering of the data, we believe the dicing saw produced reasonable quality end faces repeatedly and the film quality over the 4-cm long waveguide is quite uniform. The 3-mm core guides have a propagation loss of 1.7 dB/cm and a waveguide coupling loss of 4 dB. The 5-mm core guides have a propagation loss of 1.2 dB/cm and a coupling loss of 2.5 dB.
The propagation loss of the 5-mm core guides is the same as the 1.2 dB/cm core material loss measured in a slab waveguide. The 3-mm core waveguides are 0.5 dB/cm higher. Some of this additional loss can be explained by the loss of the claddings. The cladding materials have a measured higher loss than the core (4.2 dB/cm for UV15 and 3.1 dB/cm for UFC170) and the larger mode confinement of the thicker core waveguide therefore results in smaller loss due to the claddings. Moreover, the scattering at the cladding-core interfaces may be lower in the 5-mm core waveguides because the amplitude of the electric field is lower at the interface. The lower coupling loss of the 5-mm core waveguides is due to the better mode match to the input fiber.
For the push-pull MZ modulators made of CLD-1, the low frequency half-wave voltage, Vp and the extinction ratio were measured by applying a 1-KHz signal with a triangular waveform. As shown in Fig. 9, the source signal and the output waveform from the photodetector were observed to find the Vp of the device. The Vp was 1.4 V for 1300 nm and 2.1 V for 1550 nm corresponding to a VpL product of 2.8 and 4.2 V-cm, respectively. In case of high-speed LiNbO3 modulators, the electric field strength across the optical waveguide is reduced by the thick coplanar electrode used to reduce the velocity mismatch and this results in a VpL product of 9 V-cm with z-cut push-pull electrode structure  and 6 V-cm under the external push-pull operation mode where an additional RF circuit is required .
The insertion loss of the unpackaged 3-cm long MZ modulator, which includes two Y-branches and 2-cm long EO interaction length, was in the range of 9-10 dB for the most of the fabricated devices. This includes the coupling loss due to the input fiber and the output lens. According to the cut-back measurement, the chip loss of the 3-cm device was ~ 5 dB. Then the coupling loss between the normal single mode fiber and the waveguide input facet was ~ 3.5 dB, and the extra loss introduced by the lens to collect the light from the waveguide end would be ~ 0.5 dB. The extinction ratio of the modulator, when the output is collected by a lens, was greater than 20 dB, which indicates the oversized rib waveguide suppresses the higher order mode propagation and the two waveguide arms of MZ modulator poled in opposite direction have about the same loss.
The performance of the RF strip line was measured on an unpackaged chip in the range of 45 MHz to 40 GHz using a network analyzer and probes. The line was terminated in 35 W which is close to the characteristic impedance of the microstrip line. >From the S21 measurement the line loss is ~0.7 . This is higher than that predicted for very good gold (0.48) and is an indication that the quality of the plated gold could be improved. A measurement of the dc resistance of the upper electrode also shows that the resistivity of the gold is lower than that of bulk gold. The reflection loss from the line is less than 13 dB over the frequency band.
The optical frequency response of the 2-cm push-pull modulator was measured with a lightwave component analyzer (HP83420A) from 130 MHz to 20 GHz. Electrical contact to the modulator under test was established via an air spaced 50 W coplanar probe and the transmission line was terminated with a 35 W resistance on the chip. Because of the resistance of the upper electrode the input impedance of the modulator is dispersive for frequencies below ~2 GHz. The input impedance is higher for the lower frequencies and this results in an apparent increase in the modulation efficiency at low frequencies. This artifact can be eliminated by a low frequency matching circuit on the input to make use of the full bandwidth made available by the polymer modulator. Fig. 10 shows the results of the measurement from 2 to 20 GHz. The dip around 17 GHz may be due to a calibration error in our receiver since we see this structure in the frequency response of other lightwave components. To extend the 3-dB bandwidth, a 1-cm interaction length modulator was fabricated. Other than the shorter interaction length the design was identical to the 2-cm device. The Vp of the shorter modulator is twice that of the 2-cm modulator. The 1-cm device was completely packaged with RF connectors and optical fibers to demonstrate the performance. The optical frequency response, shown in Fig. 10, was measured with an optical network analyzer (HP8510) with the test set replaced by an HP8511B and a 50 GHz detector (U2T). The packaged 1-cm device has a 3-dB bandwidth of ~34 GHz.
The bandwidth of these modulators is limited by the rf loss of the microstrip line and, as noted above, the loss on our lines is about 40% larger than the predicted loss for high quality gold electrodes. The quality of our plated gold can be improved and if one could approach that of perfect gold the 3-dB bandwidth of the polymer modulators could be considerably extended. The velocity mismatch effect on the bandwidth should not be a factor until the modulation frequency is less than 100GHz.
VI. Stability of the CLD/APC Polymer Modulator
One of the major concerns for the polymer devices when used in systems is their long term stability. The devices must demonstrate good long term thermal stability, high photo-stability at 1550 nm, and a stable and trackable bias point.
Earlier work has shown that some EO polymers show chromphore damage when exposed to infrared wavelengths even though these wavelengths are well away from the principle absorption peak of the chromophore. Index of refraction changes and chromophore damage were observed at 1300 nm and in each case oxygen appeared to play a role . To study these effects, we aligned and attached input and output fibers to a group of modulators and straight waveguides fabricated with CLD-1/APC polymer. During the alignment we used 1 mW of TM polarized power at 1550 nm to avoid any initial degradation of the polymer waveguide. After the fiber attachment, 10 mW was launched into a straight waveguide. Within a few hours, there was a significant power drop in the waveguide output. After this exposure, a bleached area could be observed near the input. The pattern of the bleached area indicated that, because of the index drop in the core, the waveguide mode lost lateral confinement near the input.
After a continuous TM polarization exposure at 10 mW at 1550 nm for 1 day, the change of output mode profile was observed as shown in Fig. 11. Fig. 11 (a) is the TM mode picture of the unbleached waveguide adjacent to the exposed waveguide. The TE profile was essentially the same. Fig. 11 (b) and (c) show the TM and TE output mode profiles respectively after the high power exposure with the TM polarized input light. Clearly the TM index of refraction has decreased sufficiently to lose the lateral mode confinement. The TE polarization is reduced to a smaller extent.
The refractive index decrease in the waveguide core is due to a photochemical reaction which damages the chromophores and therefore reduces their contribution to the refractive index. At 1300 nm wavelength, the singlet oxygen generated by the intensive optical power is believed to play a role to this reaction . The excited oxygen attacks the carbon double bond or the nitrogen bond in the chromophore molecule. It has also been speculated that two photon absorption may play a role in the chromophore degradation. The CLD-1 chromophore has an absorption peak near 700 nm, which could be excited by absorbing 1550 nm light through the two-photon absorption process. The excited chromophore could react with the ground state oxygen to cause damage. The two-photon process could explain the larger index change for the TM polarization since the high intensity light was TM polarized and tended to preferentially excite chromphores oriented in the TM direction. As a result, the refractive index for the TM polarization will be decreased much faster than the index for the TE polarization. A depoling of the material due to two photon induced trans-cis excition and relaxation is not the dominate process in our experiments because the index appears to decrease for both polarizations.
To determine the role of oxygen, we operated pigtailed modulators and waveguides in a sealed container with a continuous Ar gas flow through the container. Before launching the high optical power into the device, we waited for 1 day at ambient temperature until the absorbed O2 in the polymer diffused out completely in an Ar environment in which the amount of oxygen was on the order of a few ppm. Fig. 12 shows the monitored output power from the straight waveguide exposed to 10-mW input power at 1550 nm. The sample in the Ar environment shows no drop in output power for 120 hours while the sample in air shows a rapid decrease.
Perhaps a more sensitive measure of optical damage is to measure the Vp of the devices. We monitored the Vp of a single arm (not push-pull) modulator kept in Ar at an input power of 20 mW at 1550 nm. Due to the coupling loss between fiber and waveguide, the power coupled into the waveguide was ~ 8 mW. During 32 days of continuous operation, as shown in Fig. 13, the device exhibited excellent stability with no measurable change in Vp.
For the comparison of the thermal stability of materials, the monitoring of the generated second harmonic signal as the temperature of a poled sample is increased by 10 ºC/min. is a useful measurement. The poled nonlinearity of the CLD/APC material can be maintained up to 110-120 ºC in this measurement . Although this short-term measurement can give some guidelines for material selection, it is not a measure of the long term thermal stability. To get a better assessment of the long term stability, a wafer containing four modulators was held at 60 ºC in air and the Vp of the modulator was monitored over 40 days. A similar wafer was kept at room temperature as a reference. As shown in Fig. 14, we observed the increase of Vp from 4.8 V to 6.0 V (~ 25%) from the sample heated at 60 ºC, though the sample kept at room temperature did not show increase of Vp. No similar measurements have been made in the Ar atmosphere so we do not know if oxygen played any role in the increase of Vp.
This paper reviews the current state-of-the-art of the CLD/APC based high speed modulators. Based on this material polymer EO modulators are becoming a commercially viable component. The low frequency Vp L product @ 1550 nm is 4.2 V-cm. The Vp @ 1550 nm is 2.1 V and the frequency response is near 20 GHz for a 2-cm long modulator. By the modification of chromophore structure to reduce the intermolecular interaction, less than 2 V of Vp was achieved very recently . Shorter devices with higher Vp have nearly 35 GHz of bandwidth. There is still considerable improvement that is possible in the fabrication to reduce the rf micro-strip line losses which will considerably extend the bandwidths. The optical loss of the modulators is acceptable and material research to further reduce the loss is underway. The long term thermal stability of the current devices shows 25% poling relaxation at 60°C and needs to be improved. We are currently measuring a cross-linked polymer which appears to extend the thermal stability by ~ 30°C. It appears that the packaging that removes O2 can solve most of the photo-stability problems. To confirm this, measurements must be made at higher temperatures and for longer times. There appears to be no problems in fabricating single mode waveguides and therefore in achieving high extinction ratios in the modulators.
In summary, there has been considerable recent progress in polymer EO modulators and the long years of materials research by many investigators is now bearing fruit. Although not all of the questions have been answered regarding the critical long-term stability of the polymer devices, the progress made makes the outcome look promising
The authors would like to appreciate Uray Co., Korea for supplying the UFC polymers and the Advanced Technology Center at Lockheed Martin for identifying the APC material. This work was supported by AFOSR, BMDO, ONR, and Pacific Wave Industries.
 C. C. Teng, “Traveling-wave polymeric optical intensity modulator with more than 40 GHz of 3-dB electrical bandwidth,” Appl. Phys. Lett., vol. 60, no. 13, pp. 1538-1540, 1992.
 D. Chen, H. R. Fetterman, A. Chen, W. H., Steier, L. R. Dalton, W. Wang, Y. Shi, “Demonstration of 110 GHz Electro-optic Polymer Modulators”, Appl. Phys. Lett., vol. 70, pp. 3335-3337, 1997
 S. T. Kowel, S. Wang, A. Thomsen, W. Chan, T. M. Leslie, and N. P. Wang, “High field poling of electrooptic etalon modulators on CMOS integrated circuits,” IEEE Photon. Technol. Lett., vol. 7, no. 7, pp. 754-756, 1995.
 S. Kalluri, M. Ziari, A. Chen, V. Chuyanov, W. H. Steier, D. Chen, B. Jalali, H. Fetterman, and L. R. Dalton, "Monolithic Integration of Waveguide Polymer Electrooptic Modulators on VLSI Circuitry", Photon. Tech. Lett., Vol. 8, pp. 644-646, 1996.
 M. Hikita, Y. Shuto, M. Amano, R. Yoshimura, S. Tomaru, and H. Kozawaguchi, “Optical intensity modulator in a vertically stacked coupler incorporating electro-optic polymer,” Appl. Phys. Lett., vol. 63, no. 9, pp. 1161-1163, 1993.
 S. M. Garner,S-S Lee, V. Chuyanov, A. Chen, A. Yacoubian, and W. H. Steier, “Three dimensional integrated optics using polymers”, IEEE J. Quantum Electron. vol. 35, pp. 1146-1155, 1999
 M.-C. Oh, W.-Y. Hwang, and K. Kim, "TE/TM polarization converter using twisted optic-axis waveguides in poled polymers," Appl. Phy. Lett., vol. 70, no. 17, pp. 2227-2229, 1997.
 I. McCulloch, and H. Yoon, “Fluorinated NLO polymers with improved optical transparency in the near infrared,” J. Polymer Science: Part A: Polymer Chemistry, vol. 33, pp. 1177-1183, 1995.
 T. C. Kowalczyk, T. Z. Kosc, K. D. Singer, A. J. Beuhler, D. A. Wargowski, P. A. Cahill, C. H. Seager, M. B. Meinhardt, and S. Ermer, “Crosslinked polyimide electro-optic materials,” J. Appl. Phys., vol. 78, no. 10, pp. 5876-5883, 1995.
 H.-T. Man, and H. N. Yoon, “Long term stability of poled side-chanin nonlinear optical polymer,” Appl. Phys. Lett., vol. 72, no. 5, pp. 540-542, 1998.
 W. Sotoyama, S. Tatsuura, and T. Yoshimura, “Electro-optic side-chain polyimide system with large optical nonlinearity and high thermal stability,” Appl. Phys. Lett., vol. 64, no. 17, pp. 2197-2199, 1994.
 H. Ma, B. Chen, T. Sassa, L. R. Dalton, and K.-Y. Jen, “Highly efficient and thermally stable nonlinear optical dendrimer for electrooptics,” J. Am. Chem. Soc., vol. 123, no. 5, pp. 986-987, 2001.
 M. A. Mortazavi, H. N. Yoon, and C. C. Teng, “Optical power handling properties of polymeric nonlinear optical waveguides,” J. Appl. Phys., vol. 74, no. 8, pp. 4871-4876, 1993.
 M. A. Mortazavi, K. Song, H. Yoon, and I. McCulloch, “Optical power handling of nonlinear polymers,” Polymer Preprint, vol. 35, pp. 198-199, 1994.
 Y. Shi, W. Wang, W. Lin, D. J. Olson, and J. H. Bechtel, “Double-end crosslinked electro-optic polymer modulators with high optical power handling capacity,” Appl. Phys. Lett., vol. 70, no. 11, pp. 1342-1244, 1997.
 C. Zhang, A. S. Ren, F. Wang, L. R. Dalton, S.-S. Lee, S. M. Garner, W. H. Steier, “Thermally Stable Polyene-Based NLO Chromophore and Its Polymers with Very High Electro-Optic Coefficient,” Polymer Preprints, vol. 40, pp. 49-50, 1999.
 Poly(Bisphenol A carbonate-co-4,4’-(3,3,5-trimethylcyclohexylidene)diphenol, available from Aldrich Chemical Co., 1001 West Saint Paul Avenue, Milwaukee, WI 53233, U.S.
 Y. Shi, C. Zhang, H. Zhang, J. H. Bechtel, L. R. Dalton, B. H. Robinson, W. H. Steier, “Low (sub-1-volt) halfwave voltage polymeric electro-optic modulators achieved by controlling chromophore shape,” Science, vol. 288, pp. 119-122, 2000.
 S. Ermer, D. G. Girton, L. S. Dries, R. E. Taylor, W. Eades, T. E. van Eck, A. S. Moss, and W. W. Anderson, “Low-voltage electro-optic modulation using amorphous polycarbonate host material,” Proc. SPIE, vol. 3949, pp. 148-155, 2000.
 M.-C. Oh, H. Zhang, A. Szep, V. Chuyanov, W. H. Steier, C. Zhang, L. R. Dalton, H. Erlig, B. Tsap, H. R. Fetterman, “Electro-optic polymer modulators for 1.55 mm wavelength using phenyltetraene bridged chromophore in polycarbonate,” Appl. Phys. Lett., vol. 76, no. 24, pp. 3525-3527, 2000.
 I. Liakatas, C. Cai, M. Bösch, M. Jäger, Ch. Bosshard, P. Günter, C. Zhang, and L. R. Dalton, “Importance of intermolecular interactions in the nonlinear optical properties of poled polymers,” Appl. Phys. Lett., vol. 76, no. 11, pp. 1368-1370, 2000.
 C.-C. Teng, “Precision measurements of the optical attenuation profile along the propagation path in thin-film waveguides,” Appl. Optics, vol. 32, no. 7, pp. 1051-1054, 1993.
 UFC polymers are synthesized and supplied from URAY Co., 245-1, Kwanhang-lee, Jungnam-Myun, Hwasung-goon, Kyungki-do, Korea.
 R. A. Soref, J. Schmidtchen, and K. Petermann, “Large single-mode rib waveguides in GeSi-Si and Si-on-SiO2,” IEEE J. Quantum Electronics, vol. 27, no. 8, pp. 1971-1974, 1991.
 S. P. Pogossian, L. Vescan, and A. Vonsovici, “The single-mode condition for semiconductor rib waveguides with large cross section,” J. Lightwave Technol., vol. 16, no. 10, pp. 1851-1853, 1998.
 U. Fisher, T. Zinke, J.-R. Kropp, F. Arndt, and K. Petterman, “0.1 dB/cm waveguide losses in single-mode SOI rib waveguides,” IEEE Photon. Technol. Lett., vol. 8, no. 5, pp. 647-648, 1996.
 K. H. Hahn, D. W. Dolfi, R. S. Moshrefzadeh, P A. Pedersen, and C. V. Francis, “Novel two-arm microwave transmission line for high-speed electro-optic polymer modulators,” Electron. Lett., vol. 30, pp. 1220-1222, 1994.
 S. K. Mohapatra, C. V. Francis, K. Hahn, and D. W. Dolfi, “Microwave loss in nonlinear optical polymers,” J. Appl. Phys., vol. 73, pp. 2569-2571, 1993.
 C. Zhang, G. Todorova, C. Wang, T. Londergan, and L. R. Dalton, “Synthesis of new second-order nonlinear optical chromophores: implementing lessons learned from theory and experiment,” Proceedings of International Society for Optical Enginering (SPIE), vol. 4114, pp. 77-87, 2000.
 Akria Otomo, G. I. Stegeman, W. H. G. Horsthuis, and G. R. Möhlmann, “Strong field, in-plane poling for nonlinear optical devices in highly nonlinear side chain polymers,” Appl. Phys. Lett., vol. 65, pp. 2369-2371, 1994.
 P. T. Dao and D. J. Williams, “Constant current corona charging as a technique for poling organic nonlinear optical thin films and the effect of ambient gas,” J. Appl. Phys., vol. 73, pp. 2043-2050, 1993.
 M. M. Howerton, R. P. Moeller, A. S. Greenblatt, and R. Krahenbuhl, “Fully packaged, broa-band LiNbO3 modulator with low drive voltage,” IEEE Photon. Technol. Lett., vol. 12, no. 7, pp. 792-794, 2000.
 K. Noguchi, H. Miyazawa, and O. Mitomi, “LiNbO3 high-speed modulator,” CLEO Pacific Rim ’99, FS2(invited), pp. 1267-1268, 1999.
 H. Zhang, M.-C. Oh, A. Szep, W. H. Steier, C. Zhang, L. R. Dalton, H. Erlig, Y. Chang, D. H. Chang, and H. R. Fetterman, “Push-pull electro-optic polymer modulators with low half-wave voltage and low loss at both 1310 and 1550 nm,” Appl. Phys. Lett., vol. 76, pp. 3136-3138, 2001.
Table 1. Physical properties of the FTC and CLD chromophores: the absorption peak wavelength was measured with two different solvents, dioxane and CHCl3, the decomposition temperature of chromophore Td was measured in a nitrogen atmosphere with 10 ºC/min temperature ramp, the electro-optic coefficient r33 was measured from a single layer EO material at 1060 nm, and the Vp were measured at 1300 nm from the fabricated MZ modulators made of each chromophore with single arm 2-cm long modulation electrode.
Fig. 1. Chemical structures of the high mb chromophores, FTC and phenyltraene-bridged CLD chromophores.
Fig. 2. Effective index of the planar waveguide as a function of the thickness of core layer. The waveguide consists of UV15 lower cladding (1.504), CLD/APC core layer (1.612), and UFC170 upper cladding (1.488). In the rib waveguide structure, the waveguide side needs to be etched by DT to obtain the effective index contrast of DNeff for the fundamental mode.
Fig. 3. Metallic loss of the planar waveguide structure including top and bottom electrode layers made of Cr and Au as a function of cladding thickness for the three different core layer thickness. The loss will be less than 0.1 dB/cm if the cladding thickness is greater than 2 mm for the 3-mm core waveguide.
Fig. 4. The structure of RF strip line incorporated for the push-pull MZ modulator.
Fig. 5. Schematic diagram of the polymer modulator fabrication procedure. The polarity of electric field applied for the push-pull poling is denoted.
Fig. 6. Electrooptic coefficients measured from the fabricated MZ modulators with different poling conditions: (a) Poling efficiency depending on the poling temperature and the applied voltage, (b) Poling efficiency depending on the poling time with a fixed poling temperature of 140 ºC and an applied voltage of 400 V.
Fig. 7. Typical temperature and voltage profiles used for push-pull poling of the CLD/APC polymer modulator. The sample is purged with nitrogen at 120 ºC for 30 min. to remove the dissolved oxygen from the polymer, then the sample is poled at 145 ºC for 30 min. in nitrogen atmosphere.
Fig. 8. Propagation loss of CLD/APC waveguide measured by the cutback method for the TM polarization, which shows the propagation loss of 1.2 dB/cm and 1.7 dB/cm and the coupling loss of 4 dB and 2.5 dB for the waveguide with a core thickness of 3 mm and 5 mm, respectively.
Fig. 9. Oscilloscope trace of the input and output signal during the Vp measurement. The result shows the MZ modulator made of CLD-1/APC has 1.4 V of half wave voltage @ 1300 nm.
Fig. 10. The frequency response of the polymer MZ modulator with 1-cm and 2-cm long electrodes. The shorter device exhibits the 3-dB electrical bandwidth over 30 GHz from 2 GHz.
Fig. 11. Comparison of guided mode profile exhibiting the effect of waveguide bleaching: (a) is the initial mode profile before the bleaching, (b) and (c) are the TE and TM polarized mode pictures respectively after the light exposure.
Fig. 12. The change of waveguide transmission during the high optical power illumination for the waveguide sealed in Ar and the other waveguide exposed to air.
Fig. 13. Long-term measurement of half-wave voltage from the CLD/APC polymer MZ modulator in Ar atmosphere with a power of 20 mW exposed at the waveguide input. No considerable change of the EO coefficient was observed over 30 days.
Fig. 14. The result of long-term thermal stability measured from the CLD/APC polymer MZ modulator. Although the sample kept in room temperature does not show poling relaxation, we observed 25 % of Vp increase from the sample heated to 60 ºC over 40 days.