Chapter 4 VCSEL design and measurement setup
4.6 RF measurement system
We adopted the RF measurement system with powerful software from Agilent Inc.
Figure 4.11 shows a RF measurement system, which contains a network analyzer of Agilent 8720ES, a DC source/Monitor of Agilent 4156, an optical platform, and a RF probe. The software is Agilent 85190A IC-CAP 2002, which is used to extract the
parameters of the device, which can be a component, circuitry, or Integrated Circuit. The Agilent 8720ES network analyzer is a crucial instrument in the RF measurement system.
The output of the DC source/Moniter (Agilent 4156) is connected to the DC input at the rear panel of the network analyzer and provides a bias current through the internal circuit of the network analyzer for VCSEL operation. Transmitter of the network analyzer provides a -10dBm RF signal and a bias current to VCSEL directly. The benefit of this connection is that can perform RF measurement with DC sweep. All of the measured data will be sent back to computer via GPIB and plotted/recorded by IC-CAP 2002 software.
All of test conditions we wanted can be set in IC-CAP 2002. The extraction of the parameters can be executed by IC-CAP 2002 software after that all test conditions have been completed. The RF test signal from the network analyzer is injected into VCSEL, which is bonded on sub-mount substrate shown as Figure 4.12, through the coaxial cable and the coplanar RF probe. The coplanar RF probe has 700µm pitch, which is space from ground (G) to signal (S) tips, and maximum working frequency is up to 40 GHz. Signal probe is higher than ground tips. Fine probing is observed by signal probe skating on contact substrate.
Optical platform contains microscope, beam splitter, objective and fiber coupler etc.
We used a long work distance objective (20X, Mitutoyo), fixed in a triple-divide translation stage, to pick up the light output from VCSEL. Light is separated by beam splitter and received by CCD and another 10X objective (Olympus). One of splitting light is received by a simplified microscope, which is constructed by beam splitter and CCD to make probing easily. Another light path through 10X objective is coupled into multimode fiber by five-axis fiber aligner (Newport).
The collected light is transmitted into a 12 GHz Photodetector (Model 1580, NewFocus) that converts the light into electrical signal. The output of Photodetor is connected to the port 2 of the network analyzer. The characteristics of transmission and reflection can be expressed as vector (magnitude and phase), scalar (magnitude only), or phase-only quantities, that is, S-parameter. The theorem of S-parameter has been reviewed briefly in the chapter 3.
Figure 4.1 Schematic diagram of inductively coupled plasma reactive ion
Figure 4.2 Schematics diagram of a typical commercial ion-implantation
Figure 4.3 A cross sectional view of a ring-trenched VCSEL structure
(a) Epitaxial growth a VCSEL (b) Made a ring trench by mesa etching
(c) Passivation process (d) Metal contact coating for electrode
(f) Made electrical confinement by oxidation process.
Figure 4.4 The fabricated steps of a oxide-confined VCSEL
Figure 4.5 Illustration of an oxidation process system setup
Figure 4.6 The oxidation rate of 98% Al-content layer
Figure 4.7 OM-image of VCSEL with 6µm oxide-aperture
(a) Epitaxial growth a VCSEL (b) Made a ring trench by mesa etching
(c) Passivation process (d) Metal contact coating for electrode
(f) The oxidation process (g) The proton implantation process Figure 4.8 The fabricated steps of a oxide-implant VCSEL
Figure 4.9 Proton-implant depth with different energy simulating by Trim
Figure 4.10 Probe station measurement system
Figure 4.11 A RF measurement system
Figure 4.12 Sub-mount substrate design
Chapter 5
Measurement and modeling analysis
Most of VCSELs are fabricated to be an oxide-confined VCSEL, which can work with high frequency modulation. [1] [2]. But the highest modulation frequency that an oxide-confined VCSEL can reach is not enough to challenge the conventional Edge Emitting Laser due to parasitic effect in the oxide-confined structure. [3] A considerable parasitic capacitor, located between the bonding pad and the oxide-layer, is a crucial parameter for high speed modulation. There are many studies that reported about decreasing parasitic capacitance by the reduction of pad area [4], mesa implantation [5], and BCB process [6] [7]. Mesa implantation is a popular process in the semiconductor fabrication industry and is easier than BCB process. So we compared the oxide-confined VCSEL with oxide-implant VCSEL in our experiment. In section 5.1, we have a comparison of the oxide-confined VCSEL and oxide-implant VCSELs for the static characteristics and the high speed modulation performance. In section 5.2, we compared the high speed modulation performance for the VCSELs with different oxide aperture sizes. The diameter of the oxide aperture is related to the high speed modulation performance due to the transverse mode and parasitic effect. In section 5.3, we presented an equivalent circuit and the result of the modeling. That represented the limitation of the high speed modulation performance is the structure of VCSEL according to the value of components in equivalent circuit, we extracted from the measured data.
5.1 Comparison oxide-confined and oxide-implant VCSELs
Figure 5.1 shows the structures of the oxide-confined VCSEL and the oxide-implant VCSEL. These two structures have similar fabricated process, and the only difference is that the oxide-implant VCSEL has an additional insulated layer in the top DBR mirrors built by proton implantation. The fabricated steps of these two structures have been described in Chapter 4. In this section, we compared these two VCSELs that both have the oxide aperture of 8µm and the mesa etched ring-trench of 22µm. For the
oxide-implant structure, the implanted region must be kept away from the ring-trench to prevent the reliability issues triggered due to the damages in active region During the proton implant process.
Figure 5.2 shows the L-I-V curve of these two structures respectively. Table 5.1 lists the static characteristics of these two structures. The threshold current of the oxide-implant structure is less than the oxide-confined structure due to the additional current confinement that increases the current density through active region and reduces the threshold current effectively. But the additional current confinement causes a larger series resistance in the oxide-implant structure. The light output power has increase of 18% for comparing the oxide-implant with the oxide-confined structure when an identical bias current is injected. Therefore, the oxide-confined structure has higher power dissipation when these two VCSELs output identical light power. The power dissipation will heat the entire structure, leads to variation of the parasitic components in structure, which degrades light output power further.
Figure 5.3 shows the measured modulation responses of the oxide-confined and the oxide-implant VCSEL. The maximum modulation bandwidth is 2.3 GHz for the oxide-confined structure and 8.6 GHz for the oxide-implant structure. In the modulation response of the oxide-confined structure, the roll-off is serious at the lower frequency.
The roll-off is caused by the parasitic effect. The resonance frequency increases with the bias current as expected. Figure 5.4 shows the resonance frequency (fr), and 3dB frequency (f3dB) of these two VCSELs, which are the function of the (I-Ith)1/2. The modulation current efficiency factor (MCEF) can be obtained from the Figure 5.4, represented the modulation bandwidth as function of (I-Ith)1/2. When the VCSEL operating with high bias current, the oxide-confined VCSEL is worse than the oxide-implant VCSEL due to the parasitic effect, we demonstrated by the result of the modeling in the section 5.3. The modulation response of a VCSEL is governed by the parasitic capacitance. This parasitic capacitance limits the modulation bandwidth of the oxide-confined structure. The oxide-implant has higher modulation bandwidth, illustrates the additional proton implantation process can reduce the parasitic capacitance of device
effectively.
Figure 5.5 shows the eye diagrams measurement of oxide-confined and oxide-implant VCSEL. In the eye diagram of oxide-confined VCSEL, the jitter is 30 ps and fall-time tail touches the mask of 10Gbps. Therefore, the oxide-confined VCSEL can not be employed for 10Gbps communication. In the oxide-implant VCSEL, the jitter is less than 20 ps and has 44 ps for rise-time and 54 ps for fall-time, respectively. The oxide-implant VCSEL can pass 10Gbps clarified.
5.2 Comparison oxide-implant VCSELs with different oxide aperture sizes
In this section, we compared the high speed modulation performance of the oxide-implant VCSELs, which have different oxide aperture diameter. The diameter of the designed oxide aperture is 6, 7, and 8 µm respectively. Figure 5.6 shows the LIV curve and small signal modulation response of those VCSELs respectively. Table 5.2 lists the relevant static parameters. We found that the series resistance is inverse proportion to the diameter of the oxide aperture, and the modulation bandwidth of the smaller oxide aperture device is roll-off and it is flat for the larger oxide aperture device in lower frequency range.
Figure 5.7 shows the resonance frequency and the 3dB frequency of those VCSELs.
Table 5.3 lists the D-factor and the modulation current efficiency factor (MCEF) of those VCSELs. The D-factor of the smaller oxide aperture VCSEL is better and the MCEF is worse than the larger oxide aperture VCSEL at high bias current.
5.3 Equivalent circuit design and modeling
We established an equivalent circuit based on VCSEL structure to investigate the limitation of the modulation response caused by parasitic effect. Figure 5.8 (a) shows an equivalent circuit in the VCSEL structure and Figure 5.8 (b) shows an equivalent circuit for AC analysis, which neglects the ideal diode appeared in the equivalent circuit of the VCSEL structure. The measured data is obtained for the simulation and the fitting of the
parameters from the RF measurement system described in chapter 4. We used the build in function of the Agilent IC-CAP 2002 software [8] to simulate and fit the measured data for extracting the value of each component in equivalent circuit. Figure 5.9 shows a couple results after best fitting the simulated data with the measured data, which is reflection coefficient (S11) of the equivalent circuit. The best fitting can be obtained after performing the optimization function of ICCAP 2002. The value of the components can be extracted in the equivalent circuit after getting the best fitting. Table 5.4 lists the extracted values of the components for the oxide-confined VCSEL with different current bias. Table 5.5 lists the extracted values of the components for oxide-implant VCSEL with different current bias. We found that the parasitic capacitance (Cp) of the oxide-confined VCSEL can be reduced effectively after performing the proton implant process. The amount of the capacitance is reduced from 1.904 pF of the oxide-confined VCSEL to 0.36pF of the oxide-implant VCSEL. This result demonstrates that the modulation response can be improved by confining current flow, but the resistance of the DBR mirrors (Rm) increases due to the diameter reduction of the oxide aperture. This is a trade-off that needs to be optimized between the parasitic resistance (Cp) and the modulation bandwidth. In our experiment, we demonstrated the oxide-implant VCSEL is better than the oxide-confined VCSEL for modulation bandwidth.
Parameter
Oxide-confined 2.3 91 0.337
Oxide-implant 1.4 108 0. 484
Table 5.1 The static characteristics of the oxide-confined VCSEL and the oxide-implant VCSEL
Table 5.2 The static parameters for different oxide aperture size
Parameter
Table 5.3 The static parameters for different oxide aperture size
Bias
Table 5.4 The extracted values of the components for the oxide-confined VCSEL with different current bias
Table 5.5 The extracted values of the components for oxide-implant VCSEL with different current bias
Figure 5.1 The VCSEL structure of the oxide-confined and the oxide-implant
(a) Oxide-confined VCSEL (b) Oxide-implant VCSEL Figure 5.2 The L-I-V curves of the oxide-confined and the oxide-implant VCSEL
(a) Oxide-confined (b) Oxide-implant
Figure 5.3 Modulation responses of the oxide-confined and the oxide-implant VCSEL
Figure 5.4 The (I-Ith)1/2 versus the resonance frequency (fr) and 3dB frequency (f3dB) for the oxide-confined and oxide-implant VCSEL
(a) The eye diagram of the oxide-confined VCSEL
(b) The eye diagram of the oxide-implant VCSEL
Figure 5.5 The eye diagrams measurement of oxide-confined and oxide-implant VCSEL
(a) VCSEL of the 6 µm oxide aperture
(a) VCSEL of the 7 µm oxide aperture
(a) VCSEL of the 8 µm oxide aperture
Figure 5.6 the LIV curve and small signal modulation response of the 6, 7, and 8µm oxide aperture VCSEL
Figure 5.7 The resonance frequency and the 3dB frequency of the 6, 7, and 8µm oxide aperture VCSEL.
Circuit component definition Rm: mirror resistance Ra: active region resistance Ca: active region capacitance Rp: shunt resistance under pad Cp: under pad capacitance L : bonding wire
(a) An equivalent circuit in the VCSEL
structure.
(b) An equivalent circuit for AC analysis
Figure 5.8 The equivalent circuit of VCSEL
(a) The fitting results of the real and the imaginary part
(b) The fitting results of the magnitude
(c) The fitting results of the Smith Chart.
Figure 5.9 The fitting results of the oxide-confined VCSEL, which works at the bias currtnt of 4mA
Reference
[1] K. L. Lear, A. Mar, K. D. Choquette, S. P. Kilcoyne, R. P. Schneider jr., and K. M.
Geib, “High-frequency modulation of oxide-confined vertical cavity surface emitting lasers,” Electron. Letter, pp.457-458, February 1996.
[2] C. Carlsson, H. Martinsson, R. Schatz, J. Halonen, and A Larsson, “Analog modulation properties of oxide confined VCSELs at microwave frequencies,” IEEE Journal of Lightwave Technology, vol.20, pp.1740-1749, September 2002.
[3] C. H. Chang, L. Chrostowski, and J. Chang-Hasnain, “Parasitics and design considerations on oxide-implant VCSELs,” IEEE Photonics Technology Letters, vol.13, December 2001.
[4] A. K. Dutta Dutta, H. Kosaka, K. Kurihara, Y. Sugimasa, and K. Kasahara,
"High-speed VCSEL of modulation bandwidth over 7.0 GHz and its application to 100 m PCF datalink," IEEE Journal of Lightwave Technology, vol.16, pp.870-8755, May 1998.
[5] A. Larsson, C. Carlsson, J. Halonen, and R. Schatz, “Microwave modulation characteristics of 840nm oxide confined / proton implanted VCSELs,” Microwave Photonics technical report, October 2001.
[6] A. Larsson, C. Carlsson, A. Haglund, J. Halonen, and R. Schatz, “Microwave modulation characteristics of BCB-planarized oxide confined 850nm VCSELs,”
Microwave Photonics technical report, June 2002.
[7] A. N. AL-Omari, and K. L. Lear, “Polyimide-planarized vertical cavity surface emitting lasers with 17.0GHz bandwidth,” IEEE Photonics Technology Letter, vol.16, pp.969-971, April 2004.
[8] “Agilent 85190A IC-CAP 2002 User’s Guide”, Agilent Technologies.
Chapter 6 Conclusion
We investigated the high speed performance of the oxide-confined VCSEL in this thesis. We found the parasitic capacitance of the oxide-confined VCSEL can be reduced by employment of the proton implant process. We demonstrated the parasitic capacitance of the oxide-confined VCSEL can be reduced effectively by using an equivalent circuit. The small signal modulation bandwidth of the oxide-confined VCSEL can be improved from 2.3 GHz to 9 GHz after using proton implantation process. The eye diagram of the oxide-confined VCSEL, which operating at bias current of 6mA, at 10Gps bias and 6dB extinction ratio showed a very clean eye with a jitter of less than 20 ps.
To investigate the extrinsic bandwidth limitation of the oxide-confined VCSELs, an equivalent circuit instead of the oxide-confined VCSEL impedance was introduced.
The extrinsic bandwidth can be obtained by combining the bandwidth of the equivalent circuit with the measured data of the probe station. We analyzed the difference of the parasitic components between the oxide-confined and oxide-implant VCSEL. The limitation factor of the modulation bandwidth can be found out through the extraction of each component value in the equivalent circuit by adopting Integrated Circuit Characterization and Analysis Program (IC-CAP). We found the bondpad capacitance of the oxide-confined VCSEL can be reduced from 1.854 pF to 0.277 pF after proton implantation process. This extraction method, the use of the IC-CAP, was proved that is very useful to characterize the high speed performance of VCSELs and this extraction method also can be applied to most diode based optoelectronics devices.