Energy consumption in optical modulators for interconnects

Ginzton Laboratory, Stanford University, 348 Via Pueblo Mall, Stanford, California 94305-4088, USA.
Optics Express (Impact Factor: 3.49). 03/2012; 20 Suppl 2(6):A293-308. DOI: 10.1364/OE.20.00A293
Source: PubMed


We analyze energy consumption in optical modulators operated in depletion and intended for low-power interconnect applications. We include dynamic dissipation from charging modulator capacitance and net energy consumption from absorption and photocurrent, both in reverse and small forward bias. We show that dynamic dissipation can be independent of static bias, though only with specific kinds of bias circuits. We derive simple expressions for the effects of photocurrent on energy consumption, valid in both reverse and small forward bias. Though electroabsorption modulators with large reverse bias have substantial energy penalties from photocurrent dissipation, we argue that modulator diodes with thin depletion regions and operating in small reverse and/or forward bias could have little or no such photocurrent energy penalty, even conceivably being more energy-efficient than an ideal loss-less modulator.

Download full-text


Available from: David A. B. Miller, Oct 07, 2015
27 Reads
  • Source
    • "The low swing voltages and small device size allow for device switching energies in the tens of fJ per bit. The overall link energy of systems based on these QCSE modulators is thus expected to be quite attractive [2], [21], [38]. Given the potential for full CMOS compatibility of these devices, such modulators may prove suitable for highly parallelized free-space optical interconnects between silicon chips. "
    [Show abstract] [Hide abstract]
    ABSTRACT: We demonstrate the first vertical-incidence Ge/SiGe quantum well reflection modulators fabricated entirely on standard silicon substrates. These modulators could help enable massively parallel, free-space optical interconnects to silicon chips. An asymmetric Fabry-Perot resonant cavity is formed around the quantum well region by alkaline etching the backside of the Si substrate to leave suspended SiGe membranes, upon which high-index-contrast Bragg mirrors are deposited. Electroabsorption and electrorefraction both contribute to the reflectance modulation. The devices exhibit greater than 10 dB extinction ratio with low insertion loss of 1.3 dB. High-speed modulation with a 3 dB bandwidth of 4 GHz is demonstrated. The moderate-Q cavity (Q ~ 600) yields an operating bandwidth of more than 1 nm and permits operation without active thermal stabilization.
    Journal of Lightwave Technology 12/2013; 31(24):3995-4003. DOI:10.1109/JLT.2013.2279174 · 2.97 Impact Factor
  • Source
    • "The on-chip insertion loss is 12.5 dB, which includes 8 dB caused by the MMI splitter and combiner and the intrinsic silicon waveguide and 4.5 dB caused by the doped region. Although the distance from the ridge to the heavily doped region is reduced from 1 m in the previous device [23] to 0.85 m in the present device, no additional optical loss is observed, which is consistent with [29]. We fabricated the same device without the terminator to measure the electrical S parameters. "
    [Show abstract] [Hide abstract]
    ABSTRACT: We demonstrate a 40 Gbit/s silicon Mach-Zehnder optical modulator driven by a differential voltage of 0.36 Vpp. The energy efficiency is as low as 32.4 fJ/bit which is near the power efficiency of the ring modulator. We analyze the relationship between the electrical bandwidth and the electro-optical (EO) bandwidth based on the electrical S parameter measurement. Because of the nonlinear response, the electro-optical bandwidths in the small-signal tests are is slightly different when the modulator is biased at different transmission points. But the EO response is much different when the optical phase change is large enough to cover the nonlinear and linear regions at the same time. The nonlinearity can greatly improve the EO response in large-signal test. In our experiment, the rise/fall (20%-80%) time decreases from 13 ps to 10 ps as the driving amplitude increases from 5 V to 6 V under the same reverse bias of 3 V.
    Journal of Lightwave Technology 07/2013; 31(14):2434-2440. DOI:10.1109/JLT.2013.2262522 · 2.97 Impact Factor
  • Source
    • "Note that the result is independent of the generator resistance. According to Miller [22], the electrical energy required for charging a capacitor, which is also given by , can be recovered by using a by-pass capacitor. This formula gives the upper limit of switching energy per bit irrespective of data coding scheme; for a Non-Return-to-Zero (NRZ) sequence with roughly equal bits of 1's and 0's, the average energy per bit will only be half of the value given by Eq. 2. Since the energy per bit is independent of the capacitor shape to the first order (refer to Appendix I for proof), we assume a cubic shape cavity with edge length L which supports only a single resonant mode. "
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we quantitatively analyzed the trade-off between energy per bit for switching and modulation bandwidth of classical electro-optic modulators. A formally simple energy-bandwidth limit (Eq. 10) is derived for electro-optic modulators based on intra-cavity index modulation. To overcome this limit, we propose a dual cavity modulator device which uses a coupling modulation scheme operating at high bandwidth (> 200 GHz) not limited by cavity photon lifetime and simultaneously features an ultra-low switching energy of 0.26 aJ, representing over three orders of magnitude energy consumption reduction compared to state-of-the-art electro-optic modulators.
    Journal of Lightwave Technology 04/2013; 31(24). DOI:10.1109/JLT.2013.2280820 · 2.97 Impact Factor
Show more