You are on page 1of 8

New photon density rate equation for Fabry-Perot Semiconductor

Optical Amplifiers (FP SOAs)


Pengyue Wen*a, Michael Sancheza, Matthias Grossa, Osman Kibara, Sadik C. Esenera
a
Electrical and Computer Engineering Dept., University of California, San Diego

ABSTRACT

Two different approaches are commonly used for Fabry-Perot Semiconductor Optical Amplifiers (FP SOAs)
performance analysis: the Fabry-Perot resonator approach and rate equation approach. Compared with the Fabry-Perot
resonator approach, the rate equation approach is more powerful because noise and mode-related performance analysis
can be included. However, it has been shown that the results based on Fabry-perot approach contains multiplicative
factor which arise from an explicit consideration of the resonator and those factors are missing in the rate equation
approach. As a result, the existing rate equations provide a poor description of FP SOAs. Our analysis shows that this is
due to the fact that the interference between the injected optical field and the intracavity optical field has not been taken
into account properly. In this paper, a new photon density rate equation for Fabry-Perot semiconductor optical amplifiers
is derived based on the electric field rate equation. By taking this interference into account, our derivation shows that the
input coupling term in the photon density rate equation is a function of the top and bottom mirror reflectivity, as well as
the bias condition. Optical gain predictions from this new photon density rate equation match well with experimental
measurements.

Keywords: Optical amplifiers, Semiconductor optical amplifiers, Fabry-Perot resonators, Optical coupling, Interference,
Electric fields

1. INTRODUCTION

Fabry-Perot semiconductor optical amplifiers (FP SOAs) have been a subject of research for more than two
decades. As one type of FP SOAs, Vertical-cavity semiconductor optical amplifiers (VC SOAs) have drawn increasing
research attention in recent years [1] [2]. They have at least three advantages over traditional edge-emitting
semiconductor optical amplifiers (EE SOAs). First, coupling loss due to the narrow geometry of the gain region in EE
SOAs is reduced by adopting the relatively large, circular input / output distributed Bragg reflector (DBR) structure in
VC SOAs. Second, the mode-partitioning noise is suppressed by single longitudinal mode operation of VC SOAs. Third,
the polarization sensitivity of the signal gain due to anisotropic properties of waveguide structures is eliminated, again
because of the circular device geometry. Furthermore, compared with other types of optical amplifiers, 2-D arrays of VC
SOAs are attractive for parallel applications such as optical information processing, optical interconnects and remote
imaging systems [3]. In a remote imaging system, a VCSOA array could be used as optical pre-amplifier to enhance the
sensitivity of detectors in the system. In such an intensity detection (ID) system, optical detectors are the square-law
detectors, therefore, optical amplification is more favorable than its electric counterpart. Meanwhile, VCSOA is the only
candidate that can achieve optical amplification in 2-D form.
Two different approaches are commonly used for device performance analysis: the Fabry-Perot resonator
approach [4] and rate equation approach [5][6]. Compared with the Fabry-Perot resonator approach, the rate equation
approach is more powerful because noise and mode-related performance analysis can be included. But previous analyses
[5][6] have shown that this approach misses the correct coupling term of the incoming photons, because the physics of
the Fabry-Perot cavity is not taken into account properly. The multi-pass amplification nature of the devices has not been
accurately reflected in the rate equations. Furthermore, the gain prediction based on this existing rate equation approach
cannot explain the gain measured in our experiments with VC SOAs. This is because the photon density has typically
been used as the most basic parameter to describe the devices so far. However, for an optical field, the electric field is

*
Correspondence: pwen@ece.ucsd.edu, UC San Diego ECE Dept., 9500 Gilman Dr., La Jolla, CA93093-0407; phone 1
858 822-1295; fax 1 858 534-1225; http://soliton.ucsd.edu/~pwen

Physics and Simulation of Optoelectronic Devices X, Peter Blood, Marek Osinski, Yasuhiko Arakawa,
Editors, Proceedings of SPIE Vol. 4646 (2002) © 2002 SPIE · 0277-786X/02/$15.00 243

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/28/2013 Terms of Use: http://spiedl.org/terms


the fundamental description since it includes not only the amplitude, but also the phase information of the field, in
contrast to the photon density. For certain cases, in which interference is important, e.g. Fabry-Perot cavity based
devices, this leads to significantly different results.
In this paper, we derive a new photon density rate equation for Fabry-Perot semiconductor optical amplifiers
based on the electric field rate equation. Our analysis shows that the input coupling term is not only a function of the top
mirror reflectivity, but depends also on the reflectivity of the bottom mirror and the bias condition. This conclusion
cannot be obtained with the existing rate equation approach. Expressions for optical gain in both reflection mode (signal
output through top mirror) and transmission mode (signal output through bottom mirror) operation are derived, based on
this new rate equation. Furthermore, an approximation method is proposed to compare the theoretical predictions with
experimental data. Results show that our expressions are in very good agreement with the experimental measurements.

2. THEORY

2.1 Standard (existing) rate equation


In the existing rate equation [5][6], the time derivative of average photon density Np is expressed as following
dN P 2 ⋅ N Pin
= ( G − γ ) N P + (1 − R t ) + R sp (1)
dt τ RT
Where G = Γ g 0 v g is the optical gain caused by the stimulated emission, with Γ the optical mode confinement factor, g0
the linear gain coefficient, v g the group velocity. γ = v g (α m + α int ) is the total photon loss, with α m the mirror
loss, and α int the internal loss. Rt is the top mirror reflectivity. τ RT = 2 n g L / c is the cavity round-trip time with
n g the group refractive index, L the cavity length, and c is the vacuum velocity of light. The second term describes the
photon density increase resulting from the input (signal) photon density Npin entering through the top mirror. The last
term represents the spontaneous photon generation rate.
It is evident in equation (1) that the coupling term is not a function of input wavelength. In other words, the
coupling efficiencies for photons with different wavelengths are the same. Even the resonance wavelength has the same
efficiency of (1-Rt) as others. This is contrary to the wavelength selection observed in all Fabry-Perot type of devices.
Furthermore, the optical gain predicted by this equation has been found to be one or two orders lower than measured
values for most bias conditions. A more fundamental treatment is needed in order to overcome those problems.

2.2 New photon density rate equation


In this section, we describe the new rate equation approach for a FP SOA, which is exposed to external light
injection. Although a single mode device is assumed, the derivation is easily modified for multimode devices by using
the same equations for each optical mode.
In order to overcome the problems in the existing approach, the rate equation for the complex electric field is
used to describe the device. The intracavity complex electric field ε ( t ) and the injected complex electric field ε inj (t )
are given as
ε ( x , t ) = E ( x , t ) e − j ( ω t + φ ( t ))
− j ( ω t + φ inj ( t ))
(2)
ε inj ( x , t ) = E inj ( x , t ) e
Where ε inj (t ) = t t ⋅ ε in (t ) is the complex electric field of input light as it is just passed through the top mirror
of FP SOAs, t t is the top mirror transmittivity for electric field, and ε in ( t ) is the complex electric field of total input.
E(x,t) and Einj(x,t) are the slowly varying field amplitudes.
Then, the rate equation for the total complex electric field ε (t ) in the FP SOA can be written as [7][8]
dE 1 1
= (G − γ ) E + E inj Cos (φ − φ inj ) (3)
dt 2 τ RT

244 Proc. SPIE Vol. 4646

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/28/2013 Terms of Use: http://spiedl.org/terms


The first term on the right hand side of (3) describes the change of the electric field due to internal amplification and
attenuation. The second term in (3) can be easily understood with a propagating wave picture: the injected field adds Einj
to the amplitude of the intracavity propagating field every time the intracavity wave reflects from the irradiated facet at a
time interval of τ RT . The time and space dependence of E, Einj, φ , φ inj has been dropped for simplicity. Since the
optical amplification is based on stimulated emission in device, the phase difference between the intracavity field and the
injected field in (3) is 0 ( φ - φ inj =0).
In order to get the explicit photon density rate equation for FP SOAs, it is convenient to express (3) in term of
intracavity photon density NP and injected photon density N Pinj . To this end, the expression for the intracavity electric
field distribution in ref [9] is used. Also, the field amplitudes E and Einj are normalized in such a way that
N P = V P EE * and N inj = V P E inj E inj* , V p = V / Γ being the volume of the optical mode, V the active volume.
P

Then, integrating expression (3) over the whole device volume yields

(1 + G s )(1 − ) Rb G s
R t R b G s Cos 2θ 0 + (1 − R t R b G s ) Sin 2θ 0
dN P kL N Pinj
= (G − γ ) N P + + R sp
dt (1 − R t R b G s ) 2 + 4 R t R b G s Sin 2θ 0 τ RT (4)

1 + Gs Tt N Pin
= (G − γ ) N P + + R sp ( for resonance frequency : θ 0 = 0 )
1− R t R b G s τ RT
where Rt and Rb are the reflectivity of the top and bottom mirrors, respectively. Gs= exp(goL) is the single-pass intensity
gain, which is determined by the material properties and the device bias condition. T t = t t2 is the intensity transmittivity
of the top mirror, N Pin is the total input photon density. The round-trip phase delay is 2θ 0 = 2kL where k = 2πn g / λ
is the wave vector. In (4), the spontaneous emission rate Rsp has been added manually, which may be done since the
spontaneous emission field has no correlation the other fields. It can be modeled by Rsp=nspG, where nsp is the Fermi
inversion factor [10].
The second term on the right-hand side in expression (4) is due to the injected photons. The factor is front of
[ ]
NPinj is the result of the cross terms ε inj (t ) ∗ ε ∗ (t ) in the integration. Thus the coupling term is not determined by the
injected complex electric field ε inj (t ) alone, but the interference between the injected complex electric field ε inj (t ) and
the intracavity complex electric field ε (t ) . As a result, the coupling efficiency for the injection photons is not only a
function of the top mirror reflectivity Rt, but also of the reflectivity of bottom mirror Rb, and the device bias condition Gs.
Expression (4) also incorporates the Fabry-Perot filter effect in FP SOAs through the Sin and Cos factors. The amplitude
of ε (t ) is typically much larger than that of ε inj (t ) , due to the high Q cavity. This is the reason that the term for injection
photons in ref.[6] has underestimated the photon density coupled into the cavity and underestimated the gain, causing its
failure in experimental validation. We would like to point out here that same analysis could be applied directly to the
similar devices with resonant cavities, such as resonance enhanced photodetectors.
Usually, we are interested in the peak gain at resonant wavelength, and so, only the input with resonant
wavelength is considered in the rest of this paper. Also, we neglect light from spontaneous emission, which is incoherent
to the input light and in general, treated as noise in FP SOAs.
With the steady state condition d N P = 0 , the expression for the total intracavity photon density NP can be
dt
obtained from (4)
1+ Gs T t N Pin 1 (5)
NP =
τ RT (
1 − R t R b G s γ 1 − g

) 
 g th 
Here we have used the definitions of γ = v g (α m + α int ) = Γv g g th , with g th the material gain at the threshold.
According to the Fabry-Perot cavity theory, the total output power for the cavity can be described as
Ptotal = v g α m hνN P [11]. Furthermore, considering the fact that photons escape from the cavity through both top and

Proc. SPIE Vol. 4646 245

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/28/2013 Terms of Use: http://spiedl.org/terms


bottom mirrors, the expressions of optical gain in FP SOA for operation in reflection GR and transmission mode GT can
be shown as following
2
 
 
αm (1 + G s ) T t
=  
Tt
GR

 (1 − R t ) +
Rt
(1 − R b )
α m + α int 2 1 − ( )
R t R b G s (1 − g / g th ) 

(6)
 Rb 
Tb αm (1 + G s ) T t
GT =
(1 − R b ) +
Rb
(1 − R t )
(
α m + α int 2 1 − R t R b G s (1 − g / g th ) )
Rt
So far, we did not include the contribution of the electric field reflected directly from the top mirror (reflection mode
only). By taking it into account, (6) can be rewritten as
2
 
 
αm (1 + G s ) T t
=  Rt 
Tt

GR

 (1 − R t ) +
Rt
(1 − R b )
α m + α int 2 1 − ( )
R t R b G s (1 − g / g th ) 

(7)
 Rb 
Tb αm (1 + G s ) T t
GT =
(1 − R b ) +
Rb
(1 − R t )
(
α m + α int 2 1 − R t R b G s (1 − g / g th ) )
Rt
In the expression for GR, the electric field that is reflected from the top mirror is represented by the second term. The
minus sign in front of Rt results from the π phase difference due to the reflection of the input field. This result is
consistent with the gain expressions obtained from the Fabry-Perot resonator approach. The gain saturation can also be
predicted by expression (7) because the single-pass intensity gain Gs is coupled to the carrier and injected photon
densities. This issue will be discussed in detail in a future paper.

3. EXPERIMENT

3.1 Comparison method


Usually, the VC SOAs have exactly the same device structure as vertical-cavity surface emitting lasers
(VCSELs). The only difference is the operation condition for VC SOAs, which requires that the bias current is lower
than the threshold current to avoid lasing. Under such a condition, the gain coefficient g as a function of carrier density
Ne is approximately linear [12]. Furthermore, the single-pass intensity gain Gs has the value of 1 R t R b when the
device is biased at its threshold. So with carrier density Ne , Gs can be written as
N e − N tr
 1  N th − N tr
(8)
G s(N ) =  
e
 RtRb 
 
Here Nth and Ntr are the threshold and transparency carrier density, respectively. In order to compare the theoretical
prediction and VC SOA experimental measurements, we further assume that the carrier density Ne is proportional to the
bias current Ibias. This assumption will overestimate Ne, especially near the threshold current, because spontaneous
emission, Auger recombination and gain saturation become important processes in the cavity and reduce the carrier
density Ne. Under such circumstances, the optical gain is overestimated and theoretical predictions deviate from the
measurements very near the threshold. Nevertheless, this assumption provides a reasonable way to check the theory.
With this assumption, (8) can be rewritten as
I − I tr
 1  I th − I tr
(9)
G s(I ) =  
 RtRb 
 

246 Proc. SPIE Vol. 4646

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/28/2013 Terms of Use: http://spiedl.org/terms


3.2 Experimental setup
The experimental setup is shown in Fig.1. A tunable DBR laser (TuiOptics DC100) is used as input(signal)
light source. And its wavelength is measured by wavelength meter after 50/50 beam splitter #1. The spatial filter serves
two purposes. The first one is to clean up the spatial profile of the input beam to increase the coupling efficiency. The
second (and more important) one is to ensure that the beam incident on VCSOA and the one reflected from it are
collinear within micron accuracy. Two polarizers are used to control the input light polarization and the intensity, The
input and output (amplified) power are measured by power meters A and B, respectively.

Figure 1: Experimental setup

In our experiment, an 850nm VCSEL with 15um aperture size and 1 − λ cavity fabricated as a discrete light
transmitter has been used as a VC SOA. It is operated in the reflection mode as shown in Fig.1. This device has high
reflectivity DBR structures (Top mirror: Rt=99.47%, Tt=0.37%, At=0.16% and bottom mirror: Rb=99.85%, Tb=0.0 %,
Ab=0.15%). Here At and Ab are the top and bottom mirror absorption, respectively. As a light transmitter, its threshold
current is 2.65mA and I tr I th ~ 0 . 35 . The L-I curve and spatial mode profile are shown in Fig.2 and Fig.3. Fig.3
shows the device is operated at its fundamental spatial mode.

Proc. SPIE Vol. 4646 247

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/28/2013 Terms of Use: http://spiedl.org/terms


Fig.2: L-I curve of VCSOA used in experiment

Fig.3: Spatial mode profile of VCSOA

3.3 Experiment results


Experimental data (dots in figure 4) show that the overall steady state gain of our VC SOA could be as high as
16dB with 3uW input intensity. The solid line is obtained by putting (8) and device parameters into (6). It fits the
experimental data very well in contrast to the prediction of the existing rate equations, which is shown as a dashed line in
the same figure. It is evident in figure 4 that prediction of the existing rate equation is several orders of magnitude lower
than the measured value for most bias points. The theoretical prediction based on new photon density rate equation
deviates from the measurements near the threshold, because of the overestimated carrier density Correction to the
deviation near the threshold can be made by taking spontaneous emission and other non-radiation recombination i.e.
Auger recombination into account, which is beyond the scope of this paper. These results verify our gain expressions
and indicate that the assumptions made to compare experiments with theory are reasonable

248 Proc. SPIE Vol. 4646

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/28/2013 Terms of Use: http://spiedl.org/terms


Fig.4: Optical Gain from VCSOA: Dots are experimental data, solid line is the theoretical prediction
from (6) and the dashed line is the prediction from Ref.6

4. CONCLUSION

In this paper, we have reviewed the problems with existing rate equation. We believe the interference between
input and intracavity field has been overlooked in the treatment and caused its failure in device performance prediction.
Based on the electric field rate equation, we have derived a new photon density rate equation for FP SOAs. Our analysis
shows that the coupling term for the input photons in the photon density rate equation is not only a function of the top
mirror reflectivity, but depends also on the reflectivity of the bottom mirror and the bias condition through the single-
pass intensity gain Gs. Expressions for optical gain both for operation in reflection mode and transmission mode are
obtained based on this new rate equation. In order to validate them, two assumptions are proposed to compare our
theoretical expressions with experimental data. Result shows that our gain prediction fits the experimental data very well
and the two assumptions are reasonable.

REFERENCES
[1] D. Wiedenmann, B. Moeller, R. Michalzik, and K.J.Ebeling, “Performance characteristics of vertical-cavity semiconductor laser
amplifiers,” Electron. Lett. 32, 342-343 (1996)
[2] R. Lewen, K. Streubel, A. Karlsson and S. Rapp, “Experimental demonstration of a multifunction long-wavelength vertical-cavity
laser amplifier-detector,” IEEE Photon. Technol. Lett., 10, 1067-1069 (1998)
[3] C. Tombing, T. Saitoh and T. Mukai, “Performance prediction for vertical-cavity semiconductor laser amplifiers,” IEEE J.
Quantum Electron. , 30, 2491-2499 (1994)
[4] M.J. Adams, J. V. Collins, I. D. Henning, “ Analysis of semiconductor laser optical amplifiers” IEE Proceedings, 132, 58-63
(1985)
[5] Daniel T. Cassidy, “ Comparison of rate equation and Fabry-Perot approached to modeling a diode laser” App. Optics 22, 3321-
3326 (1983)
[6] J. Piprek, S. Bjorlin and E. Bowers, “Design and analysis of vertical-cavity semiconductor optical amplifiers,” IEEE J. Quantum
Electron. , 37, 127-134 (2001)

Proc. SPIE Vol. 4646 249

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/28/2013 Terms of Use: http://spiedl.org/terms


[7] R. Lang, “ Injection locking properties of a semiconductor laser,” IEEE J. Quantum Electron. 18, 976-983 (1982)
[8] G. P. Agrawal and N. K. Dutta “Semiconductor lasers” by AT&T, 2nd Edition, 232-238, (1993)
[9] J. T. Verdeyen, “Laser Electronics” by Prentice Hall, 3rd Edition, 148-159 (1995)
[10]. G. P. Agrawal and N. K. Dutta “Semiconductor lasers” by AT&T, 2nd Edition, (1993)
[11] L. A. Coldren and S. W. Corzine, “Diode lasers and photonic integrated circuit” by John Wiley & Sons, Inc. 77-79 (1995)
[12] L. A. Coldren and S. W. Corzine, “Diode lasers and photonic integrated circuit” by John Wiley & Sons, Inc. 465-469 (1995)

250 Proc. SPIE Vol. 4646

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/28/2013 Terms of Use: http://spiedl.org/terms

You might also like