1404

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 26, NO. 11, JUNE 1, 2008

A Multiquantum-Dot-Doped Fiber Amplier With Characteristics of Broadband, Flat Gain, and Low Noise
Cheng Cheng
AbstractDoped with PbSe nanocrystals of different sizes, a multiquantum-dot-doped ber amplier (MQDFA) is proposed. Performance of the MQDFA is simulated by solving light-propagation equations and rate equations for a three-level system, and by applying superposed emission-absorption cross sections of the quantum dots (QDs). Pumped with 980 nm, this proposed MQDFA has characteristics of broader band, atter gain, and lower noise than those of the erbium-doped ber ampliers at present. There is evidence to show that two factors, i.e., the equivalent Stokes shifts and the full wave at half maximum of the superposed spectra, dominate the performances of the MQDFA. Also, the equivalent Stokes shift and the FWHM are tunable by adopting the QDs in different sizes and choosing the doped number available. It is expected that such a MQDFA may be able to cover the all waveband in telecommunications if it is optimized on the doped number, sizes, and doped concentrations of the QDs.

I. INTRODUCTION N THE pursuit of increasing the overall capacity of the wavelength-division-multiplexing (WDM) system, research effort has been directed towards developing a highly efcient , broadband ber amplier. Some rare-earth ions, such as , , and ions and/or their combinations are used as dopants in order to increase the transmission capacity of the WDM network. For example, Harun et al. [1] obtained a double-pass, long-wavelength band ( -band) erbium-doped ber amplier (EDFA) with a bandwidth of 32 nm (15681600 nm) and a at gain of 33 dB. This is achieved by combining the bandwidth of conventional band ( -band) and -band. Lu et al. [2] presented a novel structure by combining two pieces of -band and -band dual-core erbium-doped bers. This novel amplier has a at gain of 15 dB over a wavelength range of 105 nm (15151620 nm). The gain variation for the band at gain region (15151555 nm) is 1.3 dB, and for the band at gain region (15621620 nm) is 1.5 dB. The noise gure varies from 4.5 to 4.8 dB over the whole bandwidth. In addition, thulium-doped ber ampliers (TDFAs) combined with the EDFAs dominate in a short-wavelength band ( -band) due to considerable performance of a gain of approximately 25 dB and a noise gure of less than 9 dB over a wide wavelength

Manuscript received September 13, 2007; revised December 5, 2007. This work was supported in part by the National Natural Science Foundation of China (No. 60777023) and in part by the Zhejiang Provincial Natural Science Foundation of China (No. Z407371). The author is with the Department of Applied Physics, Zhejiang University of Technology, Xiaoheshan, Hangzhou 310023, China (e-mail: chengch@zjut. edu.cn). Color versions of one or more of the gures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/JLT.2008.923641

triple-band region of 14581540 nm [3]. Recently, an double-pass EDFA is reported, in which it is with an embedded dispersion-compensating ber (DCF) module for WDM applications [4]. These new approaches may be considered as the best comparing with the many other variations at present. Although the EDFAs appear to offer bandwidths to spare, it is inadequate for future networks. It is believed that there is no enough space left to further develop the ampliers doped with the rare-earth ions due to inherent atomic structures and spectral characteristics. Therefore, a specially designed and engineered dopants that can provide gain bandwidth at a higher level for the ber ampliers have being expected. Note that there has been a rapid development in nanocrystal material science recently. The nanocrystal is a sub-zero-dimensional material that is termed quantum dots (QDs). By virtue of their tunable sizes and shapes, these self-organized or self-assembled nanocrystals show unique characteristics of electrics and optics. For example, Brumer et al. [5] investigated nanocrystals of PbSe core and PbSe/PbS as saturable absorbers in passively -switched near-infrared laser. Their theoretical ts yield a ground-state absorption cross section of m , an excited-state m and an effective absorption cross section of lifetime of s. Chung et al. [6] described the temporally depending uorescence intensity of CdSe nanocrystal quantum dots, which is based on single quantum dot uorescence blinking statistics. Some issues, e.g., the analytical model, the high gain and the gain saturation in the quantum-dot optical amplier were also investigated [7][9]; [10] and [11] reviewed the recent progresses in quantum dots applied to optoelectronic devices, including advanced sensors, sources, and lters. Some nanocrystal QDs, such as CdSe, CdTe, PbS, PbSe, and/or their combinations, possess desirable emission-absorption spectra that cover wavelength regions spanning 4652340 nm [11] and [12]. In the QDs with a strong connement of electrons, small variations in the sizes can be translated into relatively large variations in the allowed energies, transition wavelengths, and spectral bandwidths. This inhomogeneous broadening degrades the lasing emissions of QDs. On the contrary, an optically pumped amplier can benet from this inhomogeneous broadening: broad, spectrally at gain with independent saturation of each channel is desirable for telecommunications. It has been shown that such ampliers can achieve temperature-insensitive, low-noise, and low-crosstalk amplication [13]. We presented a novel ber amplier, namely a quantum-dotdoped ber amplier (QDFA) in [14]. The PbSe QD with a diameter of 5.5 nm is adopted as a dopant, employing its favorable characteristics, e.g., its full-width at half-maximum (FWHM)

0733-8724/$25.00 2008 IEEE

CHENG: MULTIQUANTUM-DOT-DOPED FIBER AMPLIFIER

1405

reaches up to 150 nm with its absorption and emission peak merely locating at 1550 and 1630 nm. In a steady-state approximation, propagation equations of the light power with different frequencies are solved by using a numerical method in a single-mode condition. The resulting bandwidth of 3 dB increases to 50 nm, and the noise gure decreases to 3.2 dB with the signal gain of 20 dB [14]. These performances have advantages over the EDFAs at present, generally, with 30-nm bandwidth and noise gure of 3.5 4.5 dB [15] and [16]. However, it is difcult to further widen the bandwidth on such a singly doped QDFA due to its bandwidth restricted by the wavelength spacing of the emission-absorption peak (i.e., the Stokes shift), and also due to the FWHM is not be tunable for the given species of QD in the same size. To increase the bandwidth of the QDFA, one method is to use the dopants of multidoped QDs in different sizes, due to the linear superposition of the cross sections depending on their sizes. Thus, an effective Stokes shift and a broader FWHM can be obtained by overlapping the cross sections of the QDs in different sizes. Although the larger QDs can absorb some short wavelength light emitted by the smaller QDs, this can be accommodated by changing the doped number, the spacing of the spectrum peaks, and the relative concentrations. More important, all of these QDs can be pumped with the same frequency since they all absorb with large cross sections on the short waveband. The ability to pump several different QDs with a single frequency is a major advantage over current natural dopants. Also, it is required to choose other parameters available, such as ber length, pumping wavelength and pumping power, etc., all of which are determining factors on the performance of the ber ampliers. In this paper, we describe the emission-absorption and the superposed cross sections of the PbSe QD. A multidoped QDFA (MQDFA), including double, triplet, and quartet dopings, is simulated by a numerical method. The method solves the light propagation equations and the rate equations for a three-level system, which uses the superposed cross sections. With a pump of 980-nm wavelength, the bandwidth, gain and noise are estimated in the conditions of a single mode and a steady approximation. There is evidence to show that this proposed MQDFA possesses more attractive performance on the bandwidth than either the singly doped QDFA or the conventional triple band if EDFAs. It is possible to cover the the MQDFA can be further optimized on the key parameters, e.g., the doped number, the QD concentrations and the optimal combinations of the QDs in different sizes. II. LEVELS AND CROSS SECTIONS The absorption and emission spectra of the PbSe QD with a diameter of 5.5 nm are measured, e.g., in [14], as shown in Fig. 1. The PbSe-QD specications in different diameters are shown in Table I. For the QDs with the other diameters, the absorption-emission spectra have the similar proles, excepting the peak wavelengths (refer to Table I). The values of the cross section peak can be determined by Beer-Lamberts law [17] (note that there is an erratum in Fig. 1 of [14] due to inattention, in which the right Y-coordinate should be m ). Referred to Fig. 1, electron-hole transitions are shown in Fig. 2(a), in which the absorptions are corresponding to ,

Fig. 1. Emission-absorption spectra of the PbSe QDs in a diameter of 5.5 nm [14].

, , and , etc., respectively, wavelength ranging 800 1800 nm and peaking around 1550 nm, , wavelength and the emission is corresponding to ranging 1450 1800 nm and peaking around 1630 nm. Considering Figs. 1 and 2(a), a three-level system is assigned, as shown in Fig. 2(b), in which level 3 includes a level group and level 2 includes two-ne levels corresponding to both the . In the emission peak and the rst absorption peak MQDFA, the pumping is usually located at a short wavelength (e.g., 980 nm) due to big absorption cross sections. Absorbing , the QDs are excited pumping energy in the cross sections to populate the upper level 2 and 3. The population of level 2 comes back to the ground state by two channels, i.e., one is by the emission in a probability being proportional to the cross section , and another is by a nonradiative transition in . In contrast, the population of level 3 arising a probability from the excitation of the ground state cannot come straight back to the ground state following an emission due to a parity selection rule, referred to Fig. 1 showing a single-emission peak . Therefore, it is reasonable that the population of level 3 de-excites to level 2 by the nonradiative transition in a probability , and then to the ground state by the other emission/nonradiative transition. Note that some researches report that (i.e., between level 3 and 2) is with a very big probability as an intraband transition, or level 3 has a very short uorescence lifetime ( ps [18] and [19]). So, the population of level 3 transits immediately down to level 2, resulting in the three-level system being degraded to a two-level system. In the multidoped case, the averaged cross sections are dened as follows: (1) is the doped number involved in the simulation. As an where example, Figs. 3 and 4 show the absorption-emission cross sections in the triplet doping case, in which the peak-peak spacing is nm corresponding to a size distribution

1406

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 26, NO. 11, JUNE 1, 2008

TABLE I PbSe-QD SPECIFICATIONS [12] AND [20]

Fig. 2. Level diagram of the PbSe QD. (Bold solid line) Emission, (dotted line) absorptions, and (thin solid line) nonradiative transitions.

Fig. 4. Emission cross sections (for nm averaged cross section in the triplet doping case.

D = 5:5

6 (5  10%)) and the

Fig. 3. Absorption cross sections (for nm averaged cross section in the triplet doping case.

D = 5:5

6 (5  10%)) and the

nm , and the black solid lines corresponding to the averaged cross sections . Notice that the averaged cross sections are already attened by the superposition, either for the absorption or for the emission. For the emission, its equivalent FWHM is widened to 300 nm, 2 times broad compared with the single QD case. For the absorption, the rst

absorption peak disappears already, showing a at prole. In the cases of the double and quartet dopings, the proles of the overlapping spectra are similar. Obviously, such a broad FWHM is favorable to from a broadband exciton and emission. Then, the FWHM of the overlapping spectra can be regarded as the rst key determining the bandwidth of the MQDFA. The three-level system is employed in this paper. The light propagating in the ber is dispersed as beams with different frequencies. Supposing the homogenous medium or no dispersive effect involved, the rate equations, including for the signal power in wavelength , for the pumping power in wavelength , for the noise and for the level population and , can be given, as shown in Appendix. In models that have amplied-spontaneous emission (ASE), both the forward and backward directions should be included. The number of modes in (A1) in Appendix is 2 for the noise, as in the case of the ber supporting only the two polarization states of the lowest order optical mode. The scattering arising from such a dopant in big sizes (compared with atomic sizes of the rare-earth ions) can be

CHENG: MULTIQUANTUM-DOT-DOPED FIBER AMPLIFIER

1407

TABLE II PARAMETERS OF THE MQDFA USED IN THIS PAPER

TABLE III COMPUTATIONAL RESULTS OF THE MQDFA

neglected due to Rayleigh scattering ( , and so low doped-volume-ratio of , where the doped concentration m as described below). A detailed description of the equations can be referred to [14], not being discussed here. Two approximations/conditions are adopted in our modeling, i.e., 1) a weak-guidance approximation, and 2) a single-mode condition by choosing a suitable refractive ( , where are the redifference fractive indexes of core and cladding) and employing the small ber-core radius related with the normalized signal frequency for single modes. By scanning signal wavelength from 1400 through 1800 nm nm, and locating the pumping wavelength spacing at 980 nm, one can obtain the signal gains, bandwidths and noise gures of the MQDFA as a function of wavelength . Only the pumping forward direction is implemented, i.e., in (A1). Once the initial conditions for the pumping power , the pumping wavelength , the signal power , and boundary conditions for the frequencies are specied and , (A1)(A3) are integrated over space , at frequency , and time . In the steady-state case, the left of the (A1) equaling to zero, then the differential equations are reduced to the algebraic equations. Table II lists the parameter values required in the computational implementation, in which the ber parameters (the core radius , the refractive-index and the excess ber loss ) are chosen to be difference the same with the usual bers, e.g., fabricated by Corning, Ltd. The transition probability of the level is ps , different experimental measured in [18] and [19]. As to , 0.3 and 0.138 s, respecdata are reported, e.g., tively, in [18], [20], and [21]. Here, it is acceptable to choose a middling s among these data.

III. RESULTS AND DISCUSSION To obtain a desirable gain bandwidth of the MQDFA, it is necessary to choose a suitable QD-concentration . Researches show that the signal gain is proportional to , while the bandwidth is inversely proportional to . In the current work, the total QD concentration is chosen as m , supposing each being the same and in a uniform spacial distribution. On the other hand, some different peak(e.g., 25, 50, 100, 150, and 200 nm, etc.) peak spacing are employed in our implementation. Among these peak-peak spacing, optionally chosen nm is available, coinciding with the peak-wavelength tolerance in Table II. With and , an acceptable gain ( 20 dB) these values of the and a broad bandwidth ( 100 nm) can be obtained at a berby numerically solving (A1)(A6). Table III shows length the implemented results in different doping cases, in which the is dened as the bandwidth of 3 dB as usual, and is dened as the at bandwidth of 1 dB. An optimized EDFA is also listed in the last row in Table III to be compared with the MQDFA, where the EDFA is doped with the concentration of m m , with an inputted signal power of 30 dBm and a double pumping pass in the wavelength 1499 nm [15]. As for the noise gure determined by (A7) and (A8) in the Appendix, their values are listed in Table III corresponding to the wavelength with the maximum gain. From Table III, the bandwidth of the multidopings has an obvious advantage over the single doping, reaching to as much as 122, 200, and 277 nm for the doped number , 3, and 4, respectively. Note that the values of the at bandwidth depending on are proportional to the equivalent Stokes shift , e.g., 75, 120, and 57 nm for 185, 207, and , 3, and 4. Here, the maximum 190 nm in the case of at bandwidth is nm corresponding to the maximum

1408

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 26, NO. 11, JUNE 1, 2008

nm (for ). Big Stokes shifts are advantageous to produce more exciton of the QD around the emission-peak wavelengths due to both big excitations and small absorptions despite the cross section tops being attened by the superposition. Therefore, it is important to seek the maximum . In fact, the equivalent Stokes shift is dependent on both the doped number and the peak-peak spacing related with the size distributions (or ). Different combinaand possess different . Thus, it is diftions of cult to obtain a maximum if only increases (e.g., here) while (or ) is kept unaltered. In this nm is due to corresponding paper, so chosen nm , howto the size distribution ever, such a 100 nm is not likely to be suitable for the other size distributions. If one optimizes the parameters together with , (or ), and , the optimally doped number may 2 or 4, etc. Therefore, a useful criterion used in the be of the superposing spectra, but not just MQDFA should be the doped number . Based on the discussion above, the equivalent Stokes shift can be regarded as the second key determining the bandwidth of the MQDFA. As a broadband amplier of the doped QD, a broad size distribution of the QD is preferred due to the broad emitted-absorption spectra and obvious Stokes shift in the multidoped case, despite a sharp size distribution being preferred in some of photoelectronic devices. is required as the doped From Table III, the longer length number increases; this is because the averaged cross secdue to a tions dened in (1) become smaller increased with constant cross section for each QD. Note that the ber lengths of the MQDFA available are always much shorter than that of the EDFA, this is because the cross sections of the QD are even ion. to 5 orders of magnitude larger than that of the On the other hand, the gain of the MQDFA barely reaches to 20 dB in the conditions listed in Table II, being smaller compared with the EDFA. The reasons can be ascribed to the fact s of the PbSe QD, while that so much short lifetime it is as long as 10 ms for ion, resulting in no enough population to be accumulated in the upper level. As an example, for , the signal gain and the noise gure at the ber-length are shown in Fig. 5, where the region between two vertical lines are the at waveband of 1 dB, reaching to 120 nm and extending to the -band connected with the -band. The noise gure stays a constant 3.3 dB, showing a at prole in the interested wavelength region. Fig. 6 shows the signal gain and the at bandwidth as a function of ber length, in which 1595 nm is with the maximum gain. From Fig. 6, the gain increases with ber length (refer to Fig. 6). Note that the gain and the bandwidth are conned each other; namely, the higher the gain, the narrower the bandwidth. Dening a factor to describe gain-bandwidth characteristics of the MQDFA, the value comes near to a constant ( 500, except from the rst point) among the involved ber lengths. Thereafter, the relation between and can be probed. In this way, a desirable MQDFA with either broadband or high gain according the constant value can be designed to meet different requirements for telecommunications. The MQDFA suggested in this paper is un-optimized. An optimal MQDFA can be obtained by optimizing its key parameters. However, such an optimization is beyond this paper due to

3).

Fig. 5. Signal gain and the noise gure as a function of wavelength (for

N =

3).

Fig. 6. Signal gain and the bandwidth as a function of ber length (for

N =

too much work involved in the optimization depending on an algorithm and an inverse process that have to be described. As an initial estimation, the at bandwidth of 1 dB can arrive to 200 300 nm with the gain of 20 30 dB on the triple band. This simulation is required to be veried by experiments. However, there are difculties in laboratory technologies, including doping QDs into ber cores according to the given concentration ratio, making a uniform distribution in hollow bers in m magnitude, and solidifying the colloid dopants in cores, all of which demand further efforts in the future. No doubt, developments of the MQDFA depend closely on the quality of quantum dots to be enhanced. It is even expected that the MQDFA will become a potential competitor among a variety of ber ampliers in the future if great efforts can be made by developing technologies, e.g., a coating technology, although the QD has some disadvantages in the current days in respect of, e.g., stability, consistency, and shelf lifetime, as presented in [24].

CHENG: MULTIQUANTUM-DOT-DOPED FIBER AMPLIFIER

1409

IV. CONCLUSION The MQDFA proposed in this paper has observable advantage in the bandwidth, when comparison is made with either the singly doped QDFA or the conventional EDFAs. A at bandwidth ( 1 dB) of 120 nm with a signal gain of 20 dB and a noise gure of 3.3 dB can be reached in the case of triplet doping with a peak-peak spacing of 100 nm. Among the keys inuencing the bandwidth of MQDFA are the equivalent Stokes shift and the FWHM of the superposed cross sections, which are tunable by using the dopants in different sizes and the different doped number in the MQDFA. It is expected to cover the triple band if the MQDFA can be further optimized. In respect of the gain and noise, however, it shows no desirable performance as compared with the singly doped QDFA. APPENDIX [14], [15], [22], AND [23] The equation for the optical power in frequency as follows: is given

is from level 2 to 1; the subscript denotes the pump; and the population of the level 3. The rst term on the right hand 2 of (A2) is the loss of the absorption including the level 1 3 corresponding to the cross sections of the and and 1 , respectively. The second term is from the emission with the cross section , the third term is caused by the nonradiative transition from level 2, and the last term is from the transition . of level 3 with the cross section The rate equation for the population of level 2 is given by

(A3) where is the nonradiative transition probability from level 3 to 2. Followed by the conservation law, , where is the total QD concentration. With a steady-state approximation and from (A3), one has

(A1) where is the signal/pumping beam power and the noise power in the frequency ; each beam is traveling either in the forward or backward direction; is the emission (absorption) cross section; is the normalized transverse mode, only considered a single-mode approximation here; are the populations of the ground state and the upper level of the emission, respectively; is the excess ber loss per length; is the effective noise bandwidth; is the number of modes in the case of the lowest order is the contribution of spontaneous optical mode, and emission. On the right hand of (A1), the rst term is for the emission gain; the second term is for the absorption loss; and the last term is for the excess ber loss including scattering loss. Note that (A1) includes 3 independent equations, corresponding to the pump, signal and noise, respectively. The amplied-spontaneous emission (ASE), both the forward and for noise, backward components are included, i.e., while for the signal and pumping beams. The rate equation for the ground state is given by (A4) Similarly, the ground state can be also determined from (A2) and (A4) with the steady-state approximation

(A5)

From (A1), (A4), and (A5), the optical power in the frequency along the axis can be obtained by a numerical computation employing the given parameters of , , , and . In this paper, are shown in Figs. 1, 3, and 4; the values of , , and are given in the text; and is the transversal model intensity with a fundamental mode and frequency , which is given by (A6) where are the zeroth- and rst-order Bessel functions, respectively, and the normalized frequency (where are the refractive indexes of core and cladding). Coming from matching the solutions at (for ), the variables and have the approximation and under the weakly guided approximation. For the noise gure NF, there is a formula with a little difference in form from ones given in [14]. Here, labeled with the noise-power , effective noise-bandwidth and signal gain , the noise gure NF is dened as follows: (A7)

(A2) where corresponds to the shortest absorption (emission) wavelength; corresponds to the longest wavelengths of the absorption/emission; ; is the Planck constant; is the nonradiative transition probability

1410

JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 26, NO. 11, JUNE 1, 2008

where the factor

of spontaneous emission is dened by

(A8)

REFERENCES
[1] S. W. Harun, N. Tamchek, P. Poopalan, and H. Ahmad, Double-pass L-band EDFA with enhanced noise gure characteristics, IEEE Photon. Technol. Lett., vol. 15, no. 8, pp. 10551057, Aug. 2003. [2] Y. B. Lu, P. L. Chu, A. Alphones, and P. Shum, A 105-nm ultrawide-band gain-attened amplier combining C - and L-band dualcore EDFAs in a parallel conguration, IEEE Photon. Technol. Lett., vol. 16, no. 7, pp. 16401642, Jul. 2004. [3] T. Sakamoto, S. Aozasa, and M. Yamada et al., High-gain hybrid amplier consisting of cascaded uoride-based TDFA and silica-based EDFA in 14581540 nm wavelength region, Electron. Lett., vol. 39, no. 7, pp. 597599, Apr. 2003. [4] J. B. Rosolem, A. A. Juriollo, and R. Arradi et al., S C L tripleband double-pass EDFA with an embedded DCF module for CWDM applications, J. Lightw. Technol., vol. 24, no. 10, pp. 36913697, Oct. 2006. [5] M. Brumer, M. Sirota, and A. Kigel et al., Nanocrystals of PbSe core, core/shell as saturable absorbers in PbSe/PbS, and PbSe/PbSe S passively Q-switched near-infrared lasers, Appl. Opt., vol. 45, no. 28, Oct. 2006. [6] I. Chung, J. B. Witkoskie, and J. Cao et al., Description of the uorescence intensity time trace of collections of CdSe nanocrystal quantum dots based on single quantum dot uorescence blinking statistics, Phys. Rev. E, vol. 73, no. 1, pp. 011106011112, Jan. 2006. [7] O. Qasaimeh, An analytical model for quantum dot semiconductor optical ampliers, Opt. Commun., vol. 222, no. 16, pp. 277287, Jul. 2003. [8] A. V. Uskov, E. P. OReilly, and M. Laemmlin et al., On gain saturation in quantum dot semiconductor optical ampliers, Opt. Commun., vol. 248, no. 13, pp. 211219, Apr. 2005. [9] Z. Bakonyi, H. Su, and G. Onishchukov et al., High-gain quantum-dot semiconductor optical amplier for 1300 nm, IEEE J. Quant. Electron., vol. 39, no. 11, pp. 14091414, Nov. 2003. [10] E. H. Sargent, Infrared quantum dots, Adv. Mater., vol. 17, no. 5, pp. 515522, 2005. [11] M. Sugawara, N. Hatori, and M. Ishida et al., Recent progress in selfassembled quantum-dot optical devices for optical telecommunication: Temperature-insensitive 10 Gb/s directly modulated lasers and 40 Gb/s signal-regenerative ampliers, J. Phys. D, Appl. Phys., vol. 38, no. 13, pp. 21262134, Jul. 2005. [12] C. B. Murray, S. Sun, and W. Gaschler et al., Colloidal synthesis of nanocrystals and nanocrystal superlattices, IBM J. Res. Dev. vol. 45, no. 1, pp. 4756, Jan. 2001 [Online]. Available: http://www.evidenttech.com

0 0

[13] F. Wise, Quantum dots call shots, SPIE OE Mag., pp. 2427, Oct. 2002. [14] C. Cheng and H. Zhang, Characteristics of bandwidth, gain and noise of a PbSe quantum dot-doped ber amplier, Opt. Commun., vol. 277, no. 2, pp. 372378, Sep. 2007. [15] C. Cheng and M. Xiao, Optimization of a dual pumped L-band erbium-doped ber amplier by genetic algorithm, IEEE J. Lightw. Technol., vol. 24, no. 10, pp. 38243829, Oct. 2006. [16] Y. L. Jin, Q. Y. Dou, and Y. G. Liu et al., Gain-clamped dual-stage L-band EDFA by using backward C-band ASE, Opt. Commun., vol. 266, no. 2, pp. 390392, Oct. 2006. [17] W. W. Yu, L. Qu, W. Guo, and X. G. Peng, Experimental determination of the extinction coefcient of CdTe, CdSe, and CdS nanocrystals, Chem. Master., vol. 15, no. 14, pp. 28542860, Jul. 2003. [18] B. L. Wehrenberg, C. Wang, and P. G. Sionnest, Interband and intraband optical studies of PbSe colloidal quantum dots, J. Phys. Chem. B, vol. 106, no. 41, pp. 1063410640, Sep. 2002. [19] J. M. Harbold, H. Du, T. D. Krauss, and K. S. Cho et al., Time-resolved intraband relaxation of strongly conned electrons and holes in colloidal PbSe nanocrystals, Phys. Rev. B, vol. 72, no. 19, pp. 195312195316, 2005. [20] H. Du, C. Chen, and R. Krishnan et al., Optical properties of colloidal PbSe nanocrystals, Nano Lett., vol. 2, no. 11, pp. 13211324, Nov. 2002. [21] M. H. Lee, W. J. Chung, and S. K. Park et al., Structural and optical characterizations of multi-layered and multi-stacked PbSe quantum dots, Nanotechnol., vol. 16, no. 8, pp. 11481152, Aug. 2005. [22] C. R. Giles and E. Desurvire, Modeling Erbium-doped ber ampliers, J. Lightw. Technol., vol. 9, no. 2, pp. 271283, Feb. 1991. [23] L. Kazovsky, S. Benedetto, and A. E. Willner, Optical Fiber Communications Systems. Norwell, MA: Artech House, 1996. [24] J. J. Peterson and T. D. Krauss, Photobrightening and photodarkening in PbS quantum dots, Phys. Chem. Chem. Phys., vol. 8, pp. 38513856, Aug. 2006. Cheng Cheng was born in Zhejiang, China, on November 22, 1953. He received his B.Sc. degree in physics from Zhejiang Normal University, Jinhua, China, in 1982, the M.Sc. degree in physics from the Southwest Institute of Physics, Chengdu, China, in 1985, and the Ph.D. degree in optical engineering from Zhejiang University, Hangzhou, China, in 2002. From 1985 to 2000, he was with the Hangzhou Normal University, working on gas lasers, copper vapor lasers, and lasing kinetics. In 2001, he joined the Zhejiang University of Technology (ZJUT) and transferred to erbium-doped ber ampliers, quantum-dot-doped ber ampliers, quantum-dot lasers, and high-power CO2 lasers. He visited the Royal Institute of Sweden during 19981999 and the University of Arkansas, Fayetteville, in 20042005. He is currently a Professor of physics and the Dean of the Science College, ZJUT. Dr. Cheng is an Associate Chair of the Physics Society of Zhejiang Province and Director of the Optical Society of Zhejiang Province.