paper_id
stringlengths
9
16
version
stringclasses
26 values
yymm
stringclasses
311 values
created
timestamp[s]
title
stringlengths
6
335
secondary_subfield
sequencelengths
1
8
abstract
stringlengths
25
3.93k
primary_subfield
stringclasses
124 values
field
stringclasses
20 values
fulltext
stringlengths
0
2.84M
1501.03486
2
1501
2015-01-16T10:15:42
A spin-wave logic gate based on a width-modulated dynamic magnonic crystal
[ "cond-mat.mes-hall" ]
An electric current controlled spin-wave logic gate based on a width-modulated dynamic magnonic crystal is realized. The device utilizes a spin-wave waveguide fabricated from a single-crystal Yttrium Iron Garnet film and two conducting wires attached to the film surface. Application of electric currents to the wires provides a means for dynamic control of the effective geometry of the waveguide and results in a suppression of the magnonic band gap. The performance of the magnonic crystal as an AND logic gate is demonstrated.
cond-mat.mes-hall
cond-mat
A spin-wave logic gate based on a width-modulated dynamic magnonic crystal A.A. Nikitin1,2,3*, A.B. Ustinov1,3, A.A. Semenov1, A.V. Chumak2, A.A. Serga2, V.I. Vasyuchka2, E. Lähderanta4, B.A. Kalinikos1, and B. Hillebrands2 1) Department of Physical Electronics and Technology, St. Petersburg Electrotechnical University, St. Petersburg, 197376 Russia 2) Fachbereich Physik and Landesforschungszentrum OPTIMAS, Technische Universität Kaiserslautern, Kaiserslautern, 67663 Germany 3) Department of Mathematics and Physics, Lappeenranta University of Technology, Lappeenranta, 53850 Finland An electric current controlled spin-wave logic gate based on a width-modulated dynamic magnonic crystal is realized. The device utilizes a spin-wave waveguide fabricated from a single-crystal Yttrium Iron Garnet film and two conducting wires attached to the film surface. Application of electric currents to the wires provides a means for dynamic control of the effective geometry of the waveguide and results in a suppression of the magnonic band gap. The performance of the magnonic crystal as an AND logic gate is demonstrated. * and.a.nikitin@gmail.com In recent years artificially patterned magnetic media, magnonic crystals (MCs), have commanded increased interest (see reviews [1-4] and literature therein). One of the important features of MCs is the presence of band gaps in the spin-wave spectrum, i.e. frequency bands in which the propagation of spin waves is forbidden [5-7]. The dispersion management introduced by an artificial periodicity of the magnetic film allows for the observation of a variety of different linear and nonlinear spin-wave phenomena. Among the formation of band gaps [8-11] and the generation of gap solitons [12, 13]. Magnonic crystals are also promising for practical applications, particularly for phase shifters [14] and generators [15], and they have also been suggested for use as temperature or magnetic field sensors [16, 17]. them are A very fascinating type of MCs is the dynamic magnonic crystal (DMC) [18-20]. This is a wave guiding structure with rapidly switchable periodic properties. Up to now, it has been realized in the form of a YIG film with a meander type conductor placed on its surface. An electric current in the conductor produces a periodic spatial modulation of the bias magnetic field [18]. Due to this property, the dynamic for unique signal processing magnonic crystal allows functions, such as all-linear frequency conversion [19], and signal storage [20]. reversal, time logic functionality can also be realized via control of the spin- wave amplitude in the interferometer arms [23, 26]. The logic gate presented here does not utilize any interferometer circuitry; instead, it consists of one WMDMC, which controls the spin-wave amplitude. This function is realized through control of the sensitivity of the propagating spin wave to the edge modulation of the magnetic-film waveguide structure. A schematic view of the WMDMC is shown in Fig. 1. The device was fabricated using an epitaxial Yttrium Iron Garnet (YIG) film. For the experimental investigation, the YIG-film waveguide was cut from a high-quality single-crystal YIG film of 8.5 µm thickness grown on 500 µm thick Gadolinium Gallium Garnet (GGG) substrate by liquid-phase epitaxy. The film has pinned surface spins that lead to standing spin-wave resonances and appearance of additional dips in the amplitude- frequency characteristic [27]. Periodic sinusoidal width modulation of the YIG-film waveguide was made by chemical etching of the film. The spatial modulation period Tm and the peak-to-peak amplitude 2Am (see Fig. 1) were 400 m and 200 m, respectively (see Fig. 1). The length of the modulated In this Letter, we report on the realization of a width- modulated dynamic magnonic crystal (WMDMC). In the designed structure, in contrast to the aforementioned DMC, electric currents control the effective geometry of the magnetic film waveguide for spin waves propagating therein. Moreover, we demonstrate the application of the WMDMC as a spin- wave logic gate. It should be noted that so far spin-wave logic gates have only been realized using a spin-wave interferometer geometry [21-26]. For example, elements providing controllable phase shifts for spin waves propagating in the arms of a Mach- Zehnder interferometer were included in each arm. The logic operations were based on changing the phase of the spin-wave by π with an external current signal coded as a logical ‘1’. The Fig. 1. Sketch of the width-modulated dynamic magnonic crystal. The ports A and B represent the logical inputs. The currents I1 and I2 supplied to the ports correspond to a logical ‘1’ if their values equal IMBG. Zero currents correspond to a logical ‘0’. The output spin-wave signal represents the logical output. (a) (b) (c) (d) 1570 Oe Hext 1600 Oe Fig. 2. Distributions of the bias magnetic field (top) and transmitted spin-wave power versus frequency for the width-modulated dynamic magnonic crystal (bottom) for the cases when the electric current is not applied to the wires (a); the current IMBG is applied to only one of the wires (b,c); and the current IMBG is applied to both wires (d). The value of the current IMBG = 1500 mA. the conditions for excitation of area was 4 mm, while the width of the unmodulated section of the YIG-film waveguide was 1.5 mm. A spatially uniform bias magnetic field was applied across the YIG waveguide in order to provide surface magnetostatic spin waves [28]. These waves were exited and detected by the microwave strip-line antennas placed at equal distances from both ends of the modulated YIG-film area 10 mm away from each other. Two gold wires of 50 m in diameter were placed on the YIG-film surface along the modulated edges, as shown in Fig. 1. These wires were used for the dynamic control of the magnonic crystal band gap by supplying the electric currents. Let us consider qualitatively the formation of a band gap in the width-modulated magnonic crystal. Generally, a wave vector of traveling spin waves exited by the input antenna in the YIG film has two components, which are in-plane. The first component is longitudinal along the direction from the input to the output antenna while the second component is transverse. As it is physically clear, the value of the transverse component is dependent on the width of the waveguide. This effect leads to the periodic change in the wave guiding properties of the width-modulated YIG film and provides the magnonic band gap (MBG) in the amplitude vs. frequency characteristic of the investigated MC. The number of MBGs is defined by the type of the width modulation. For example, in the case of rectangular modulation there are several MBGs. As soon as the width modulation is sinusoidal, as is the case in the investigated structure (Fig. 1), there is only one MBG in the power vs. frequency characteristic (see Fig. 2(a)) [29]. the borders of The physical mechanism underlying the control of the magnonic crystal band gap by the electric currents applied to the wires can be understood as follows. Positive currents I1 and I2 applied to both wires create an additional negative Oersted field with respect to the applied static bias field. Therefore, the spatial distribution of the resulting magnetic field has two minima at the width modulation. The distributions of the bias magnetic field are shown in Fig. 2 as color maps. Provided the depth of the minima is sufficiently large, they screen the YIG-film modulated edges due to a decrease in the magnetic field inside the YIG waveguide. In other words, the existence of these minima cause the width modulation of the waveguide to become “invisible” to spin waves. This effect leads to the disappearance of the band gap at f0 = 6.6 GHz, as is shown in Fig. 2(d). We will now present the dynamic properties of the device structure realized upon application of the controlling dc electric currents. The experiments were carried out in the following manner. A microwave signal applied to the input antenna excited a surface magnetostatic spin wave. After passing through the magnonic crystal the spin wave was detected by the output antenna (Fig. 1). The power of the received spin waves was measured as a function of their carrier frequency. A typical dependence measured for H = 1600 Oe with no current in the edge wires is shown in Fig. 2(a). As can be seen in the graph, the center frequency of the band gap was f0 = 6.6 GHz. The dependence of the power of the output signal vs. frequency obtained under two equal controlling currents IMBG of 1500 mA is shown in Fig. 2(d). Note that the data clearly demonstrate (in comparison to Fig. 2(a)) a suppression of the band gap at 6.6 GHz. If the dc current was applied to one of the wires then only one of the modulated edges was deactivated. In this case, the input spin wave was still scattered on its path through the waveguide. In this situation, the band gap appears as shown in Figs 2(b, c). The difference between Fig. 2(b) and Fig. 2(c) could be explained by a possible asymmetry in the distribution of the spin-wave amplitude along the cross-section of the YIG-film waveguide. In this case, there is a difference in the sensitivity of the propagating wave to the boundary conditions on the different waveguide edges. Dependences of the spin-wave transmission as a function of the currents I1 and I2 measured at the frequency f0 = 6.6 GHz is shown in Fig. 3. As is visible, an increase in both currents leads to a substantial increase in the spin-wave transmission due to a reduction in the influence of the modulated edges. At the same time, there is a relatively weak influence with only one dc current applied. 6.66.76.80.1110Magnonic crystal band gap I1=I2= 0 mA Power (a. u.)Frequency (GHz) f0=6.6 GHz6.66.76.80.1110 Magnonic crystal band gap I1= IMBG I2= 0 mAFrequency (GHz)Power (a. u.) 6.66.76.80.1110 Magnonic crystal band gap I2= IMBG I1= 0 mAFrequency (GHz) Power (a. u.)6.66.76.80.1110 I1=I2=IMBG Power (a. u.)Frequency (GHz) Fig. 3 Dependence of the spin wave transmission on the value of the dc current at frequency f0=6.6 GHz. Squares: both currents I1 and I2 are changed; triangles: the current I1 is changed and I2=0; circles: I1=0 and I2 is changed. A possible application of the WMDMC for realization of an AND logic gate will now be explored. The operating principle of the AND logic gate is based on controlling the bias magnetic field distribution along the sinusoidal borders of the YIG-film waveguide through a change in the electric currents I1 and I2. The logic ‘0’ is represented by I = 0 mA and a logic ‘1’ – by I = IMBG = 1500 mA, which is enough to "switch off" the width modulation on one side of the WMDMC. The microwave pulses at the input antenna represent clock pulses. The microwave signal at the output antenna serves as the logic signal output. The operational frequency corresponds to the central frequency of the band gap, for example f0 = 6.6 GHz. Therefore, the presence of the band gap (i.e., low power level at the output) represents a logic ‘0’ and the absence of the band gap (i.e., high power at the output, Pout = P1) represents a logic ‘1’. traces of signal demonstrating the performance of the logic gate are shown in Fig. 4. It is observed that the logic ‘0’ appears at the output port in three situations: for two logic ‘0s’ applied to the input ports A and B (Fig. 4(a)) and for the combinations of the logic ‘0’ and the logic ‘1’ applied to the inputs (Fig. 4(b, c)). In the case where a logic ‘1’ is applied simultaneously to both inputs, the signal at the output port corresponds to a logic ‘1’. Typical oscilloscope the output In conclusion, width-modulated magnonic crystals may find a variety of applications. For example, they can be employed for the development of dynamic magnonic crystals and spin-wave logic gates, as demonstrated in this paper. Another example could be microwave notch filters with a stop- band frequency corresponding to the magnonic band gap. Advantages of the dynamic magnonic crystal presented above in comparison to the ones previously developed [18-20] are the potentially for miniaturization. Indeed, the current control is provided here by short conductors, which have an inductance much less than the meander type wire reported in [18]. The use of nanostructured width-modulated waveguides made of Permalloy films, as in Ref. [9], will allow for a significant reduction of the size of the dynamic magnonic crystal and the logic circuit presented here. fast performance and the possibility Fig. 4 Waveforms of the output signal of the AND logic gate. Microwave signal frequency f0=6.6 GHz. Currents I1 and I2 serve as logic inputs; logic ‘0’ corresponds to zero current, logic ‘1’- to IMBG=1500 mA. High level of the microwave output signal corresponds to a logic ‘1’, and low level to a logic ‘0’. The work was supported in part by the Russian Science Foundation (Grant 14-12-01296), the Ministry of Education and Science of Russian Federation, the Academy of Finland, EU-FET grant the Deutsche Forschungsgemeinschaft. InSpin 612759, and by 1. A. A. Serga, A. V. Chumak, and B. Hillebrands, J. Phys. D: Appl. Phys. 43, 264002 (2010). 2. V. V. Kruglyak, S. O. Demokritov, and D. Grundler, J. Phys. D: Appl. Phys. 43, 264001 (2010). 3. B. Lenk, H. Ulrichs, F. Garbs, and M. Münzenberg, Phys. Rep. 507, 107 (2011). 4. M. Krawczyk and D. Grundler, J. Phys.: Condens. Matter 26, 123202 (2014). 5. J. W. Kos, M. Krawczyk, Yu. S. Dadoenkova, N. N. Dadoenkova, and I. L. Lyubchanskii, J. Appl. Phys. 115, 174311 (2014). 6. H. T. Nguyen and M. G. Cottam, J. Appl. Phys. 111, 07D122 (2012). 7. M. Mruczkiewicz, M. Krawczyk, V. K. Sakharov, Yu. V. Khivintsev, Yu. A. Filimonov, and S. A. Nikitov, J. Appl. Phys. 113, 093908 (2013). 8. K.-S. Lee, D.-S. Han, and S.-K. Kim, Phys. Rev. Lett. 102, 127202 (2009). 9. A. V. Chumak et al., Appl. Phys. Lett. 95, 262508 (2009). 10. M. Arikan, Y. Au, G. Vasile, S. Ingvarsson, and V. V. Kruglyak, J. Phys. D: Appl. Phys. 46, 135003 (2013). 11. A. B. Ustinov and B. A. Kalinikos, Tech. Phys. Lett. 40, 568 (2014). 12. A. B. Ustinov, B. A. Kalinikos, V. E. Demidov, and S. O. Demokritov, Phys. Rev. B 81, 180406(R) (2010). 13. S. V. Grishin, Y. P. Sharaevskii, S. A. Nikitov, E. N. Beginin, and S. E. Sheshukova, IEEE Trans. Mag. 47, 3716 (2011). 14. Y. Zhu, K. H. Chi, and C. S. Tsai, Appl. Phys. Lett. 105, 022411 (2014). 0300600900120015000.000.250.500.751.00 I1=I2= I I1= I, I2= 0 I1= 0, I2= INormalized transmission (a. u.)Current I (mA) 15. S. V. Grishin, E. N. Beginin, M. A. Morozova, Yu. P. Sharaevskii, and S. A. Nikitov, J. Appl. Phys. 115, 053908 (2014). 16. M. Inoue, A. Baryshev, H. Takagi, P. B. Lim, K. Hatafuku, J. Noda, and K. Togo, Appl. Phys. Lett. 98, 132511 (2011). 17. R. G. Kryshtal and A. V. Medved, Appl. Phys. Lett. 100, 192410 (2012). 18. A. V. Chumak, T. Neumann, A. A. Serga, B. Hillebrands, and M. P. Kostylev J. Phys. D: Appl. Phys. 42, 205005 (2009). 19. A. V. Chumak, V. S. Tiberkevich, A. D. Karenowska, A. A. Serga, J. F. Gregg, A. N. Slavin, B. Hillebrands, Nat. Commun. 1, 141 (2010). 20. A. D. Karenowska, J. F. Gregg, V. S. Tiberkevich, A. N. Slavin, A. V. Chumak, A. A. Serga, and B. Hillebrands, Phys. Rev. Lett. 108, 015505 (2012). 21. A. B. Ustinov and B. A. Kalinikos, Tech. Phys. Lett. 27, 403 (2001). 22. M. P. Kostylev, A. A. Serga, T. Schneider, B. Leven, and B. Hillebrands, Appl. Phys. Lett. 87, 153501 (2005). 23. T. Schneider, A. A. Serga, B. Leven, B. Hillebrands, R. L. Stamps, and M. P. Kostylev, Appl. Phys. Lett. 92, 022505 (2008). 24. A. Khitun, M. Bao, and K. L. Wang, J. Phys. D: Appl. Phys. 43, 264005 (2010). 25. Y. Au, M. Dvornik, O. Dmytriiev, and V. V. Kruglyak, Appl. Phys. Lett. 100, 172408 (2012). 26. A. V. Chumak, A. A. Serga, and B. Hillebrands, Nat. Commun. 5, 4700 (2014). 27. B. A. Kalinikos, A. N. Slavin, J. Phys. C: Solid State Phys. 19, 7013 (1986). 28. D. D. Stancil and A. Prabhakar, Spin Waves: Theory and Applications, Springer, 2009. 29. F. Ciubotaru, A. V. Chumak, N. Yu. Grigoryeva, A. A. Serga, and B. Hillebrands, J. Phys. D: Appl. Phys. 45, 255002 (2012).
0908.4485
2
0908
2010-03-15T10:35:23
Discontinuous Euler instability in nanoelectromechanical systems
[ "cond-mat.mes-hall" ]
We investigate nanoelectromechanical systems near mechanical instabilities. We show that quite generally, the interaction between the electronic and the vibronic degrees of freedom can be accounted for essentially exactly when the instability is continuous. We apply our general framework to the Euler buckling instability and find that the interaction between electronic and vibronic degrees of freedom qualitatively affects the mechanical instability, turning it into a discontinuous one in close analogy with tricritical points in the Landau theory of phase transitions.
cond-mat.mes-hall
cond-mat
Discontinuous Euler instability in nanoelectromechanical systems Guillaume Weick,1, 2 Fabio Pistolesi,3, 4 Eros Mariani,1, 5 and Felix von Oppen1 1Dahlem Center for Complex Quantum Systems and Fachbereich Physik, Freie Universitat Berlin, D-14195 Berlin, Germany 2IPCMS (UMR 7504), CNRS and Universit´e de Strasbourg, F-67034 Strasbourg, France 3CPMOH (UMR 5798), CNRS and Universit´e de Bordeaux I, F-33405 Talence, France 4LPMMC (UMR 5493), CNRS and Universit´e Joseph Fourier, F-38042 Grenoble, France 5School of Physics, University of Exeter, Stocker Road, Exeter, EX4 4QL, UK (Dated: November 1, 2018) We investigate nanoelectromechanical systems near mechanical instabilities. We show that quite generally, the interaction between the electronic and the vibronic degrees of freedom can be ac- counted for essentially exactly when the instability is continuous. We apply our general framework to the Euler buckling instability and find that the interaction between electronic and vibronic de- grees of freedom qualitatively affects the mechanical instability, turning it into a discontinuous one in close analogy with tricritical points in the Landau theory of phase transitions. PACS numbers: 73.63.-b, 85.85.+j, 63.22.Gh I. INTRODUTION The buckling of an elastic rod by a longitudinal com- pression force F applied to its two ends constitutes the paradigm of a mechanical instability, called buckling instability.1 It was first studied by Euler in 1744 while in- vestigating the maximal load that a column can sustain.2 As long as F stays below a critical force Fc, the rod re- mains straight, while for F > Fc it buckles, as sketched in Fig. 1a-b. The transition between the two states is con- tinuous and the frequency of the fundamental bending mode vanishes at the instability. There has been much recent interest in exploring buck- ling instabilities in nanomechanical systems. In the quest to understand the remarkable mechanical prop- erties of nanotubes,3 -- 5 there have been observations of compressive buckling instabilities in this system.6 The Euler buckling instability has been observed in SiO2 nanobeams and shown to obey continuum elasticity theory.7 There are also close relations with the recently observed wrinkling8 and possibly with the rippling9 of suspended graphene samples. Theoretical works have studied the quantum properties of nanobeams near the Euler instability,10 -- 13 proposing this system to explore zero-point fluctuations of a mechanical mode11 or to serve as a mechanical qubit.13 In this work, we study the interaction of current flow with the vibrational motion near such continuous me- chanical instabilities which constitutes a fundamental issue of nanoelectromechanics.14 Remarkably, we find that under quite general conditions, this problem ad- mits an essentially exact solution due to the continu- ity of the instability and the consequent vanishing of the vibronic frequency at the transition ("critical slow- ing down"). In fact, the vanishing of the frequency implies that the mechanical motion becomes slow com- pared to the electronic dynamics and an appropriate non- equilibrium Born-Oppenheimer (NEBO) approximation becomes asymptotically exact near the transition. Here, FIG. 1: (color online) Sketch of a nanobeam (a) in the flat state and (b) the buckled state with two equivalent metastable positions of the rod (solid and dashed lines). An equivalent circuit of the embedded SET is shown in (c). we illustrate our general framework by applying it to the nanoelectromechanics of the Euler instability. We find that the interplay of electronic transport and the mechanical instability causes significant qualitative changes both in the nature of the buckling and in the transport properties. In leading order, the NEBO ap- proximation yields a current-induced conservative force acting on the vibronic mode. At this order, our principal conclusion is that the coupling to the electronic dynamics can change the nature of the buckling instability from a continuous to a discontinuous transition which is closely analogous to tricritical behavior in the Landau theory of phase transitions. Including in addition the fluctua- tions of the current-induced force as well as the corre- sponding dissipation leads to Langevin dynamics of the vibrational mode which becomes important in the vicin- ity of the discontinuous transition. Employing the same NEBO limit to deduce the electronic current, we find that the buckling instability induces a current blockade over a wide range of parameters. This is a manifestation of the Franck-Condon blockade15 -- 17 whenever the buckling instability remains continuous but is caused by a novel tricritical blockade when the instability is discontinuous. The emergence of a current blockade in the buckled state suggests that our setup could, in principle, serve as a mechanically-controlled switching device. II. MODEL Close to the Euler instability, the frequency of the fun- damental bending mode of the beam approaches zero, while all higher modes have a finite frequency.1 This al- lows us to retain only the fundamental mode of amplitude X (see Fig. 1a-b) and following previous studies,10 -- 12 we reduce the vibrational Hamiltonian to18 Hvib = P 2 2m + mω2 2 X 2 + α 4 X 4, (1) which is closely analogous to the Landau theory of contin- uous phase transitions. In Eq. (1), P is the momentum conjugate to X and m denotes an effective mass. The mode frequency ω2 ∼ 1 − F/Fc changes sign when F reaches a critical force Fc. Global stability then requires a quartic term with α > 0. Thus, for F < Fc (ω2 > 0), X = 0 is the only stable minimum and the beam remains straight. For F > Fc (ω2 < 0), the beam buckles into one of the two minima at ±X+ = ±(cid:112)−mω2/α. The vibronic mode of the nanobeam interacts with an embedded metallic single-electron transistor (SET), con- sisting of a small metallic island coupled to source, drain, and gate electrode (Fig. 1c). We assume that the SET operates in the Coulomb blockade regime19 and that bias and gate voltage are tuned to the vicinity of the conduct- ing region between SET states with, say, zero and one excess electron. The electronic degrees of freedom cou- ple to the vibronic motion through the occupation n of excess electrons on the metallic island. Specifically, we assume that the electron-vibron coupling does not break the underlying parity symmetry of the vibronic dynam- ics under X → −X. This follows naturally when the coupling emerges from the electron-phonon coupling in- trinsic to the nanobeam20 and implies that the coupling depends only on even powers of the vibronic mode coor- dinate X. The dominant coupling is quadratic in X, with a coupling constant g > 0.20 When there is a signif- icant contribution to the electron-vibron coupling origi- nating from the electrostatic dot-gate interaction, we en- vision a symmetric gate setup consistent with Eq. (2). In the presence of the vibronic dynamics X(t), the elec- tronic occupation n(X, t) of the island is described by the Boltzmann-Langevin equation21 = {n, Hvib} + Γ+(1 − n) − Γ−n + δJ+ − δJ−. (3) dn dt This equation assumes that the bias is large compared to temperature so that tunneling is effectively unidi- rectional and the relevant tunneling rates Γ± for tun- neling onto (+) and off (−) the island are given by Γ± = R−1(V /2 ± ¯Vg)Θ(V /2 ± ¯Vg). Here, R denotes the tunneling resistances (R (cid:29) h/e2) between island and leads, V is the bias voltage, Θ(x) denotes the Heaviside 2 step function, and we set  = e = 1. Since both gate voltage (via the capacitances C and Cg in Fig. 1c) and vi- bronic deformations couple to the excess charge n on the island, the effective gate voltage ¯Vg = Vg − gX 2/2 com- bines the gate potential Vg (measured from the degen- eracy point between the states with zero and one excess electron) and the vibron-induced shift of the electronic energy described by Hc. The stochastic Poisson nature of electronic tunneling is accounted for by including the Langevin sources δJ± with correlators (cid:104)δJ+(t)δJ+(t(cid:48))(cid:105) = Γ+(1 − n)δ(t − t(cid:48)) and (cid:104)δJ−(t)δJ−(t(cid:48))(cid:105) = Γ−nδ(t − t(cid:48)). The vibronic dynamics enters Eq. (3) through the Pois- son bracket {n, Hvib}. III. STABILITY ANALYSIS We are now in a position to investigate the influence of the electronic dynamics on the vibronic motion. Near the instability, the vibrational dynamics becomes slow compared to the electronic tunneling dynamics. As has recently been shown,22,23 the effect of the current on the vibrational motion can then be described within a NEBO approximation in which the vibrational motion is subject to a current-induced force −gXn(X, t) originat- ing in the electron-vibron interaction (2). This current- induced force involves both a time-averaged and conser- vative force as well as fluctuating and frictional forces, resulting in Langevin dynamics of the vibronic degree of freedom. In lowest order, the Langevin dynamics only involves the conservative force which emerges from the average occupation n0(X) in the absence of fluctuations (δJ± (cid:39) 0) and vibronic dynamics ({n, Hvib} (cid:39) 0). In this limit, Eq. (3) reduces to the usual rate equation of a metallic SET so that19 1, 1 2 0, vg(x) > v/2, + vg(x) v , −v/2 (cid:54) vg(x) (cid:54) v/2, vg(x) < −v/2 (4) with v > 0 and vg(x) = vg − x2/2. Here and below, we employ dimensionless variables by introducing char- acteristic scales E0 = g2/α of energy, l0 = (cid:112)g/α of length, and ω0 = (cid:112)g/m of frequency (or time t) from a comparison of the quartic vibron potential in Hvib and the electron-vibron coupling Hc. Specifically, we intro- duce the reduced variables x = X/l0, p = P/mω0l0, τ = ω0t, v = V /E0, vg = Vg/E0, and r = Rω0/E0. In terms of these variables, we can also write Hvib + Hc = E0[p2/2 + (− + n)x2/2 + x4/4] in terms of a reduced compressional force  = −mω2/g. The current-induced force −xn0(x) has dramatic ef- fects on the Euler instability, as follows from a stability analysis of the vibrational motion. The (meta)stable po- sitions of the nanobeam are obtained by setting the effec- tive force feff (x) = x − x3 − xn0(x) to zero. Our results Hc = g 2 X 2 n, (2) n0(x) =  3 FIG. 2: (color online) (Meta)stable (solid blue lines) and unstable (dashed red lines) positions of the nanobeam vs. scaled force  for (a) vg < v/2, v < 1/2, (b) vg < v/2, v > 1/2, (c) vg > v/2, v < 1/2 (for + > 1; a similar plot holds for + < 1), (d) vg > v/2, v > 1/2, as indicated in the vg -- v plane in (e). The dotted blue line is the result without electron-vibron coupling. + =  − 1, and x2− = ( − )/(1 − 1/2v). Grey indicates conducting regions. Notation: ± = 2vg ± v,  = 1/2 + vg/v, x2 + = , x2 are summarized in the stability diagrams in Fig. 2. The most striking results of this analysis are: (i) The current flow renormalizes the critical force required for buckling towards larger values. (ii) At low biases, the buckled state can appear via a discontinuous transition. These results can be understood most directly in terms of the potential veff (x) associated with feff (x). Focusing on the current-carrying region (shown in grey in Fig. 2 and delineated by max{0, −} < x2 < + with ± = 2vg ± v), we find veff (x) = 1 2 − + v + 2vg 2v x2 + 1 4 1 − 1 2v x4. (5) (cid:18) (cid:19) (cid:18) (cid:19) The quadratic term shows that the current indeed stabi- lizes the unbuckled state, renormalizing the critical force to  = 1/2 + vg/v when − < 0 < + (Fig. 2a-b). Re- markably, however, the current-induced contribution to the quartic term is negative at small x2 and thus desta- bilizes the unbuckled state. According to Eq. (5), the quartic term in the current-induced potential becomes in- creasingly significant as the bias voltage v decreases and we find that the overall prefactor of the quartic term be- comes negative when v < 1/2.24 It is important to note that this does not imply a globally unstable potential since the current-induced force contributes only for small x2. A sign reversal of the quartic term is also a famil- iar occurrence in the Landau theory of tricritical points which connect between second- and first-order transition lines.25 In close analogy, the sign reversal of the quartic term in the effective potential (5) signals a discontinuous Euler instability which reverts to a continuous transition at biases v > 1/2 where the prefactor of the quartic term remains positive. Specifically, when v > 1/2 (Fig. 2b,d), the current- induced potential renormalizes the parameters of the vi- bronic Hamiltonian but leaves the quartic term posi- tive. This modifies how the position of the minimum depends on the applied force in the conducting region max{0, −} < x2 < +, but the Euler instability remains continuous. When v < 1/2, the equilibrium position at finite x becomes unstable within the entire current- carrying region. This leads to a discontinuous Euler tran- sition when − < 0 < + (Fig. 2a) and to multistability in the region − < x2 < + when − > 0 (Fig. 2c).26 At the level of the stability analysis, we can also ob- tain the current I by evaluating the rate-equation result19 RI(x)/V = 1/4 − [vg(x)/v]2 at the position of the most stable minimum. Corresponding results in the vg -- v plane are shown in Fig. 3a-f for various values of the applied force . By comparison with the Coulomb di- amond in the absence of the electron-vibron coupling (dotted lines in Fig. 3), we see that the Euler instability leads to a current blockade over a significant parameter range. For v > 1/2, the blockade is a manifestation of the Franck-Condon blockade,15,17 caused by the induced linear electron-vibron coupling when expanding Eq. (2) about the buckled state. In contrast, for v < 1/2, the current blockade is a direct consequence of the discontinuous Euler instabil- FIG. 3: (color online) Conductance G = RI/V in the vg -- v plane for applied force (a)  (cid:54) 0, (b)  = 0.25, (c,g)  = 0.5, (d)  = 0.75, (e)  = 1, (f,h)  = 1.25, within (a-f) stability analysis and (g-h) full Langevin dynamics (r = γe = T = 0.01). Color scale: G = 0 → 1/4 from dark blue to white. Dotted lines delineate the Coulomb diamond for g = 0. 4 ity. We have seen above that in this regime, the buckled state becomes unstable throughout the entire current- carrying region. As a result, the current-induced force will always drive the system out of the current-carrying region, explaining the current blockade. An intriguing feature of this novel tricritical current blockade is the curved boundary of the apparent Coulomb-blockade dia- mond (Fig. 3), a behavior which is actually observed in nanoelectromechanical systems. Planck equation (6). Numerical results for the scaled lin- ear conductance G = RI/V are shown in Fig. 3g-h, using the same parameters as in Fig. 3c,f. We observe that the fluctuations reduce the size of the blockaded region and blur the edges of the conducting regions as the system can explore more conducting states in phase space. Nev- ertheless, the conclusions of the stability analysis clearly remain valid qualitatively. IV. LANGEVIN DYNAMICS V. CONCLUSION To investigate the robustness of the stability analysis against fluctuations, we turn to the complete vibronic Langevin dynamics x + γ(x) x = feff(x) + ξ(τ ). The fluctuating force ξ(τ ) is generated by fluctuations of the electronic occupation and the frictional force −γ(x) x by the delayed response of the electrons to the vibronic dynamics. To compute γ(x) and ξ(τ ), we solve Eq. (3) including the vibronic dynamics and the Langevin sources. Writing separate equations for average and fluctuations of the occupation by setting n = ¯n + δn, we see that the leading correction to ¯n arises from the Poisson bracket, yielding ¯n = n0 − X∂X n0. At the same time, the fluctuations δn obey the correla- tor (cid:104)δn(t)δn(t(cid:48))(cid:105) = Insert- ing these results into the expression for the current- induced force −gXn and employing reduced units, we find (cid:104)ξ(τ )ξ(τ(cid:48))(cid:105) = D(x)δ(τ−τ(cid:48)) with diffusion and damp- ing coefficients D(x) = 2rx2n0(1 − n0)/v and γ(x) = −rx∂xn0/v, respectively. Finally, we can pass from the Langevin to the equivalent Fokker-Planck equation22,23 for the probability P(x, p, τ ) that the nanobeam is at po- sition x and momentum p at time τ , Γ++Γ− n0(1 − n0)δ(t − t(cid:48)). 2 1 Γ++Γ− We have presented a general approach to the interplay between continuous mechanical instabilities and current flow in nanoelectromechanical systems, and have applied our general framework to the Euler buckling instabil- ity. The current flow modifies the nature of the buck- ling instability from a continuous to a tricritical transi- tion. Likewise, the instability induces a novel tricritical current blockade at low bias. Our nonequilibrium Born- Oppenheimer approach generalizes not only to other con- tinuous mechanical instabilities, but also to other systems such as semiconductor quantum dots or single-molecule junctions with a discrete electronic spectrum, to other types of electron-vibron coupling,28 and to further trans- port characteristics (e.g., current noise). Our proposed setup can be realized experimentally by clamping, e.g., a suspended carbon nanotube and apply- ing a force to atomic precision either using a break junc- tion or an atomic force microscope. Indeed, several recent experiments show that the electron-vibron coupling is surprisingly strong in suspended carbon nanotube quan- tum dots.4,5,16 ∂τP = −p∂xP − feff∂pP + γ∂p(pP) + pP. ∂2 (6) Acknowledgments D 2 The current I =(cid:82) dxdpPst(x, p)I(x) is now obtained Note that the diffusion and damping coefficients are non- vanishing only in the conducting region.27 from the stationary solution ∂τPst = 0 of the Fokker- We acknowledge financial support through Sfb 658 of the DFG (GW, EM, FvO) and ANR contract JCJC- 036 NEMESIS (FP). FvO enjoyed the hospitality of the KITP (NSF PHY05-51164). 1 L. D. Landau and E. M. Lifshitz, Theory of Elasticity (Pergamon Press, Oxford, 1970). 2 L. Euler, in Leonhard Euler's Elastic Curves, translated and annotated by W. A. Oldfather, C. A. Ellis, and D. M. Brown, reprinted from ISIS, No. 58 XX(1), 1744 (Saint Catherine Press, Bruges). 3 P. Poncharal et al., Science 283, 1513 (1999). 4 G. A. Steele et al., Science 325, 1103 (2009). 5 B. Lassagne et al., Science 325, 1107 (2009). 6 M. R. Falvo et al., Nature 389, 582 (1997). 7 S. M. Carr and M. N. Wybourne, Appl. Phys. Lett. 82, 709 (2003). 9 J. C. Meyer et al., Nature 446, 60 (2007). 10 S. M. Carr, W. E. Lawrence, and M. N. Wybourne, Phys. Rev. B 64, 220101(R) (2001). 11 P. Werner and W. Zwerger, Europhys. Lett. 65, 158 (2004). 12 V. Peano and M. Thorwart, New J. Phys. 8, 21 (2006). 13 S. Savel'ev, X. Hu, and F. Nori, New J. Phys. 8, 105 (2006). 14 H. G. Craighead, Science 290, 1532 (2000); M. L. Roukes, Phys. World 14, 25 (2001). 15 J. Koch and F. von Oppen, Phys. Rev. Lett. 94, 206804 (2005). 16 R. Leturcq et al., Nature Phys. 5, 327 (2009). 17 F. Pistolesi and S. Labarthe, Phys. Rev. B 76, 165317 8 W. Bao et al., Nature Nanotech. 4, 562 (2009). (2007). 18 The rotation of the plane of the buckled nanobeam is as- sumed massive due to clamped boundary conditions. 19 See, e.g., Chap. 3 in T. Dittrich et al., Quantum Transport and Dissipation (Wiley-VCH, Weinheim, 1998). 20 E. Mariani and F. von Oppen, Phys. Rev. B 80, 155411 (2009). 21 See Ya. M. Blanter and M. Buttiker, Phys. Rep. 336, 1 (2000) for a review of the Boltzmann-Langevin method. 22 Ya. M. Blanter, O. Usmani, and Yu. V. Nazarov, Phys. Rev. Lett. 93, 136802 (2004); ibid. 94, 049904(E) (2005). 23 D. Mozyrsky, M. B. Hastings, and I. Martin, Phys. Rev. B 73, 035104 (2006). 24 We note that the singularity at small v is cut off for bias voltages of the order of temperature or level broadening. 5 25 See, e.g., P. M. Chaikin and T. C. Lubensky, Principles of Condensed Matter Physics (Cambridge University Press, Cambridge, 1995). 26 In a quite different context, a discontinuous Euler insta- bility has also been predicted in: S. Savel'ev and F. Nori, Phys. Rev. B 70, 214415 (2004). 27 In some cases, a stable numerical solution of Eq. (6) re- quires a small extrinsic damping γe and temperature T . 28 E.g., a small symmetry-breaking coupling linear in X leads to a tricritical point in an external field, a purely linear coupling to a 2nd order transition in a field, G. Weick et al., unpublished.
1007.3144
1
1007
2010-07-19T13:24:01
Polarization dependence of coherent phonon generation and detection in highly-aligned single-walled carbon nanotubes
[ "cond-mat.mes-hall" ]
We have investigated the polarization dependence of the generation and detection of radial breathing mode (RBM) coherent phonons (CP) in highly-aligned single-walled carbon nanotubes. Using polarization-dependent pump-probe differential-transmission spectroscopy, we measured RBM CPs as a function of angle for two different geometries. In Type I geometry, the pump and probe polarizations were fixed, and the sample orientation was rotated, whereas, in Type II geometry, the probe polarization and sample orientation were fixed, and the pump polarization was rotated. In both geometries, we observed a very nearly complete quenching of the RBM CPs when the pump polarization was perpendicular to the nanotubes. For both Type I and II geometries, we have developed a microscopic theoretical model to simulate CP generation and detection as a function of polarization angle and found that the CP signal decreases as the angle goes from 0 degrees (parallel to the tube) to 90 degrees (perpendicular to the tube). We compare theory with experiment in detail for RBM CPs created by pumping at the E44 optical transition in an ensemble of single-walled carbon nanotubes with a diameter distribution centered around 3 nm, taking into account realistic band structure and imperfect nanotube alignment in the sample.
cond-mat.mes-hall
cond-mat
Polarization dependence of coherent phonon generation and detection in highly-aligned single-walled carbon nanotubes L. G. Booshehri,1, 2 C. L. Pint,2, 3, 4 G. D. Sanders,5 L. Ren,1, 2 C. Sun,1, 2 E. H. H´aroz,1, 2 J.-H. Kim,6 K.-J. Yee,6 Y.-S. Lim,7 R. H. Hauge,2, 4 C. J. Stanton,5 and J. Kono1, 2, 3, ∗ 1Department of Electrical and Computer Engineering, Rice University, Houston, Texas 77005, USA 2The Richard E. Smalley Institute for Nanoscale Science and Technology, Rice University, Houston, Texas 77005 3Department of Physics and Astronomy, Rice University, Houston, Texas 77005, USA 4Department of Chemistry, Rice University, Houston, Texas 77005, USA 5Department of Physics, University of Florida, Box 118440, Gainesville, Florida 32611-8440 6Department of Physics, Chungnam National University, Daejeon, 305-764, Republic of Korea 7Department of Applied Physics, Konkuk University, Chungju, Chungbuk, 380-701, Republic of Korea (Dated: January 5, 2018) We have investigated the polarization dependence of the generation and detection of radial breath- ing mode (RBM) coherent phonons (CP) in highly-aligned single-walled carbon nanotubes. Using polarization-dependent pump-probe differential-transmission spectroscopy, we measured RBM CPs as a function of angle for two different geometries. In Type I geometry, the pump and probe po- larizations were fixed, and the sample orientation was rotated, whereas, in Type II geometry, the probe polarization and sample orientation were fixed, and the pump polarization was rotated. In both geometries, we observed a very nearly complete quenching of the RBM CPs when the pump polarization was perpendicular to the nanotubes. For both Type I and II geometries, we have de- veloped a microscopic theoretical model to simulate CP generation and detection as a function of polarization angle and found that the CP signal decreases as the angle goes from 0◦ (parallel to the tube) to 90◦ (perpendicular to the tube). We compare theory with experiment in detail for RBM CPs created by pumping at the E44 optical transition in an ensemble of single-walled carbon nanotubes with a diameter distribution centered around 3 nm, taking into account realistic band structure and imperfect nanotube alignment in the sample. PACS numbers: 78.67.Ch, 63.22.+m, 73.22.-f, 78.67.-n I. INTRODUCTION The one-dimensionality of single-walled carbon nan- otubes (SWNTs) is attractive from both fundamental and applied points of view, where the 1D confinement of electrons and phonons results in unique anisotropic electric, magnetic, mechanical, and optical properties.1,2 Individualized SWNTs, both single-tube and in ensem- ble samples, have shown anisotropy with polarized Ra- man scattering and optical absorption measurements where maximum signals result when the nanotube axis is aligned parallel to the polarization of incident light.3–8 Additionally, due to strong anisotropic magnetic sus- ceptibilities, both semiconducting and metallic SWNTs align well within an external magnetic field, and with the added properties of the Aharonov-Bohm effect, the elec- tronic band structure of SWNTs respond anisotropically with the strength of a tube-threading magnetic flux.9–16 Such optical and magnetic anisotropy is also expected with bulk samples, but detailed measurements showing extreme anisotropy have been lacking.17 Here we investigate anisotropic optical and vibrational properties of a bulk film of highly-aligned SWNTs us- ing polarization-dependent coherent phonon (CP) spec- troscopy. CP spectroscopy is an ultrafast pump-probe technique that complements CW Raman spectroscopy, and although both techniques provide information about electron-phonon coupling, CP spectroscopy avoids the common disadvantages of Raman spectroscopy that in- clude detection of Rayleigh scattering and photolumi- nescence and the broadening and blending of peak features.18 Such advantages are useful when investigat- ing SWNTs, where a large majority of samples include ensembles dispersed in various environments and their optical properties are obscured by the collection of vary- ing species of nanotubes. Recent CP studies on SWNTs have produced direct observation of CP oscillations of both the radial breath- ing mode (RBM) and G-band phonons, in addition to their phase information and dephasing times.18–24 Fur- thermore, when pulse-shaping techniques are combined with CP spectroscopy, predesigned trains of femtosec- ond pulses selectively excite RBM CPs of a specific chirality, avoiding inhomogeneous broadening from the ensemble.23 However, it is important to note that pre- vious CP studies investigated randomly-aligned SWNT samples, and as the quasi-1D nature of SWNTs leads to optical anisotropy that is dominant in the polarization dependence of PL, absorption, and Raman scattering, CP studies on aligned SWNTs are necessary. Below, we present results of a detailed experimental and theoretical study of CPs in highly-aligned SWNT thin films and found a very strong polarization anisotropy of the RBM as a function of angle. In particular, we ob- served a very nearly complete quenching of the RBM when the optical polarization is perpendicular to the e c n a b r o s b A e c n a b r o s b A 3.0 2.5 2.0 1.5 1.0 0.5 1.0 0.8 0.6 0.4 0.2 0.0 (a) A// A⊥ 1.6 2.4 2.0 Energy (eV) 2.8 3.2 (b) 1 2 A// A⊥ 4 3 Energy (meV) 5 6 7 2 ! Sapphire Type I V V (a) SWNTs H pump probe (b) SWNTs H ! FIG. 1: (color online) Absorption spectra of aligned single- walled carbon nanotubes in the (a) near-infrared to visible and (b) far-infrared (or terahertz) ranges for parallel (Ak) and perpendicular (A⊥) polarization. tubes. We have developed a theoretical model to un- derstand this extreme anisotropy, including band struc- ture, interband optical transition elements, ultrafast car- rier and phonon dynamics, and imperfect nanotube align- ment in the sample. Fitting our results to theory, we also calculated the nematic order parameter, S,15–17 i.e., the degree of alignment of SWNTs in the sample. II. SAMPLES AND EXPERIMENTAL METHODS We investigated CPs through degenerate pump-probe spectroscopy measurements on highly-aligned SWNT thin films transferred onto sapphire substrates. Pat- terned, vertically aligned SWNT arrays grown by chemi- cal vapor deposition were subsequently etched with H2O vapor to allow transffer to sapphire via a dry contact transfer printing technique, which forms horizontally- aligned SWNT thin films.25,26 The resulting film pro- duces aligned nanotubes of the same length, with a di- ameter distribution centered around 3 nm.26 Figure 1(a) shows polarization-dependent absorption spectra of a typical aligned sample in the visible to near-infrared range, while Fig. 1(b) shows absorption spectra in the far-infrared or terahertz (THz) range. The polarization Sapphire pump probe Type II FIG. 2: (color online) Scanning electron microscopy image of aligned single-walled carbon nanotubes, and the two exper- imental configurations employed in the current pump-probe spectroscopy work. (a) Type I: pump and probe polariza- tions are fixed and sample orientation is rotated. (b) Type II: probe and sample orientations are fixed and pump polar- ization is rotated. anisotropy is obvious in both (a) and (b), but extreme in the THz regime (b), as there is virtually zero ab- sorption when the sample is perpendicular to the THz polarization.17 Therefore, with such an aligned sample, our polarization measurements can be extended to in- clude the sample as an additional rotation parameter, compared to our earlier study of CP polarization depen- dence in randomly-aligned SWNTs.21 Using a mode-locked Ti:Sapphire laser with ∼80 fs pulse width that is shorter than the period of excited CP oscillations and ∼40 mW average pump power, the laser was tuned to central wavelength of 850 nm (1.46 eV) to predominantly excite the E44 interband transitions of the SWNTs in the sample. A shaker-delay system and balance detector was employed for fast-scan, real-time observation of CPs. Extraction of CP oscillations from raw pump-probe time-domain signal were performed as previously described.18 As depicted in Fig. 2, two types of polarization measurements were investigated. Type I configuration maintained the same polarization for the 60 50 40 30 20 10 6 − 0 1 × / T T ∆ 0 0.5 (a) Type I q = 0° (b) Type II f = 0° 15° 30° 45° 60° 90° 15° 30° 45° 60° 90° ) s t i . n u b r a ( y t i s n e t n I 1.5 1.0 Time (ps) 2.0 0.5 1.0 1.5 Time (ps) 2.0 3 Type II λ = 850 nm f = 0° 30° 45° 60° 90° FIG. 3: (color online) Experimental differential transmission data in the time domain showing coherent phonon oscillations of the radial breathing mode for different polarization angles in (a) Type I and (b) Type II configurations (see Fig. 1). The traces are vertically offset for clarity. pump and probe, while the alignment axis of the sample was rotated. Type II configuration maintained the same orientation for the probe and sample, while the pump polarization was rotated. For Type II measurements, a half-wave plate provided 90-degree rotation of the pump. All measurements were performed at room temperature. III. EXPERIMENTAL RESULTS Results of our polarization-dependent ultrafast pump- probe differential transmission measurements on the aligned SWNT film for both Type I and Type II ori- entations are shown in Fig. 3. Here, it is seen that the typical differential transmission amplitude of the RBM CP oscillations is on the order of 10−6, with a CP de- cay time of ∼1.5 ps. This short decay time, compared to our earlier studies on individually-suspended SWNTs in solution,18,21,23 is expected with our sample of bun- dled SWNTs. As the polarization angle is rotated, we see a strong polarization anisotropy of the RBM CPs as a function of angle, where the strongest oscillations are observed at 0◦ while the signal is very nearly completely quenched at 90◦ for both Type I and Type II geometries. Figure 4 shows CP spectra for different polarization angles for Type II configuration. To produce these spec- tra in the frequency domain, we calculated the Fourier transform (FT) of the time-domain CP oscillations. The frequencies of the CP oscillations have a wide range, from 50 to 200 cm−1, which is consistent with the large di- ameter distribution of nanotubes within the sample.26 The polarization anisotropy is clearly observed in the CP 0 100 200 Frequency (cm-1) 300 FIG. 4: (color online) Coherent phonon spectra for different polarization angles in Type II configuration obtained through Fourier transform of the time-domain data in Fig. 3(b). The traces are vertically offset for clarity. ) s t i n u . b r a ( y t i s n e n t I t d e a r g e t n I 1.0 0.8 0.6 0.4 0.2 0.0 0 (a) Type I (b) Type II 30 60 q (°) 90 0 30 60 f (°) 90 FIG. 5: (color online) Spectrally integrated coherent phonon intensity as a function of angle, measured in (a) Type I and (b) Type II configurations. spectra. Figures 5(a) and 5(b) plot the integrated CP in- tensity of the FT for all excited nanotubes as a function of θ [in Fig. 5(a)] and φ [in Fig. 5(b)], where θ (φ) is the angle of rotation for the Type I (Type II) configuration. IV. THEORY We have developed a microscopic theory for the gener- ation of coherent phonons in single-walled carbon nan- otubes and their detection in coherent phonon spec- troscopy experiments. Our microscopic theory is de- scribed in detail in Ref. 22, so we only summarize the main points here and indicate how our earlier work has been extended to include polarization-dependent coher- ent phonon spectroscopy. We treat the π and π∗ electronic states in (n,m) car- bon nanotubes using a third-nearest-neighbor extended tight binding (ETB) formalism developed by Porezag et al.27 for carbon compounds. Using a density functional based parametrization, Porezag et al. derived analytic expressions for the Hamiltonian and overlap matrix ele- ments that depend only on the C-C bond lengths. Using the ETB formalism, we obtain tight-binding wave func- tions and electronic energy levels Esµ(k) where s = c, v labels the conduction and valence bands, µ labels the cut- ting lines, and k is the one-dimensional nanotube Bril- louin zone.1 By exploiting the screw symmetry of the nanotube, we can block-diagonalize the electronic Hamil- tonian and overlap matrices into 2×2 blocks, one for each cutting line, as described in Appendix A of Ref. 22. We treat nanotube lattice dynamics using a seven- parameter valence-force-field model described in Ap- pendix B of Ref. 22. Exploiting the nanotube screw sym- metry, the dynamical matrix can be block-diagonalized into 6 × 6 blocks, one for each cutting line, which can be solved for the phonon displacement vectors and dis- persion relations ω2 βν(q). In the the dispersion relations, β = 1,··· , 6 labels the phonon modes, ν is an angu- lar momentum quantum number, and q is the phonon wave vector in the one-dimensional nanotube Brillouin zone. Following the work of Jiang et al.28 and Lobo and Martins,29 we include four types of force field potentials, i.e., the bond stretching, in-plane bond bending, out-of- plane bond bending, and bond twisting potentials. Our force-field potential energies are invariant under rigid ro- tations and translations (force constant sum rule), and, as a result, our phonon model correctly predicts the dis- persion relation for the long-wavelength flexure modes.22 The electron-phonon interaction is treated in a second- quantized formalism where the electron-phonon inter- action matrix elements are evaluated in Appendix C of Ref. 22. In calculating electron-phonon matrix ele- ments, we use 2pz graphene atomic wave functions and screened atomic potentials obtained from an ab initio calculation.28 We obtain equations of motion for coherent phonon amplitudes using the Heisenberg equations as described by Kuznetsov and Stanton.30 We assume that the opti- cal pulse and photoexcited carrier distributions are dis- tributed uniformly over the nanotube. In this case, only the ν = q = 0 phonon modes are excited. To excite RBM coherent phonons, the laser pulse must be short in comparison with the RBM phonon oscillation period. 4 For coherent RBM phonons, the coherent phonon ampli- tude is proportional to the tube diameter D(t). The tube diameter, being proportional to the coherent phonon am- plitude, satisfies a driven oscillator equation ∂2D(t) ∂t2 + ω2D(t) = S(t) (1) where ω is the angular frequency of the ν = q = 0 RBM phonon mode. The driving function S(t) for RBM co- herent phonons is given by S(t) ∝ −Xsµk Msµ(k)(cid:2)fsµ(k, t) − f 0 sµ(k)(cid:3) (2) where Msµ(k) is the driving function kernal for RBM co- herent phonons defined in Eq. (12) of Ref. 22 and fsµ(k, t) and f 0 sµ(k) are the time-dependent and initial equilibrium carrier distribution functions, respectively. From Eq. (2) we see that the driving function depends on the photoexcited carrier distribution functions. We treat photoexcitation of carriers in a Boltzmann equa- tion formalism and obtain the photogeneration rate us- ing Fermi's golden rule. In Ref. 22 we only considered linearly-polarized laser pulses with the electric polariza- tion vector parallel to the nanotube axis. For parallel po- larization, optical transitions only occur between states with the same cutting line index µ, and the resulting photogeneration rate is given by Eq. (13) in Ref. 22. In the present work, we consider linearly-polarized pump and probe beams in which the electric polariza- tion vector makes a finite angle with the nanotube axis. If ǫ is the unit electric polarization vector for the pump, then the photogeneration rate is now given by ∂fsµ(k) ∂t = (cid:12)(cid:12)(cid:12)(cid:12)gen 8π2e2 u(t)  n2 g (ω)2 (cid:18) 2 ×(cid:16)fs′µ′ (k, t) − fsµ(k, t)(cid:17) δ(cid:16)∆Eµµ′ m0(cid:19)Xs′µ′ ǫ · ~P µµ′ ss′ (k) − ω(cid:17) ss′ (k)2 (3) where ∆Eµµ′ ss′ (k) = Esµ(k) − Es′µ′ (k) are the k- dependent transition energies, ω is the pump photon energy, u(t) is the time-dependent energy density of the pump pulse, e is the electronic charge, m0 is the free electron mass, and ng is the index of refraction in the surrounding medium. To account for spectral broaden- ing of the laser pulses, the delta function in Eq. (3) was replaced with a Lorentzian lineshape31 δ(∆E − ω) → Γp/(2π) (∆E − ω)2 + (Γp/2)2 (4) In Eq. (3), ~P µµ′ ss′ (k) are the dipole-allowed optical ma- trix elements between the initial and final states where µ and µ′ are initial and final state cutting lines. For polar- ization parallel to the tube axis (taken to be z), µ′ = µ and z · ~P µµ ss′ (k) with the right hand side being defined in Eq. (14) of Ref. 22. For linear polarization ss′ (k) = P µ parallel to the x axis, µ′ = µ ± 1 and in the notation of Ref. 22, we have ss′ x · ~P µ,µ±1 ×XrJ (k) =  √2m0 1 2 Xr′ C ∗ r′ (s′, µ ± 1, k) Cr(s, µ, k) eiφJ(k,µ) (cid:0)Mx(r′, rJ) ± iMy(r′, rJ)(cid:1) Similarly, for linear polarization along the y axis, we have (5) (a) E44 ) 1 - m c 4 0 1 ( (38,0) E77 ) V e ( y g r e n E 1 0 -1 y · ~P µ,µ±1 ss′ (k) = ∓ i(cid:0)x · ~P µ,µ±1 ss′ (k)(cid:1) (6) -0.4 -0.2 0.0 0.2 0.4 8 6 4 2 0 5 (b) 00 150 300 450 600 900 E44 2 1 0 3 Photon Energy (eV) E44 (d) RBM 10.91 meV Type II (38,0) 0 30 60 90 Angle (deg.) In Eqs. (5) and (6), the atomic dipole matrix element vectors (which can be evaluated analytically) are given by M(r′, rJ) = Z dr ϕ∗ r′ 0(r − Rr′ 0) ∇ ϕrJ(r − RrJ) (7) where the 2pz orbitals ϕrJ are defined in Eq. (C3) of Ref. 22. In our experiments, a probe pulse is used to mea- sure the time-varying absorption coefficient. In our the- ory, the time-varying optical properties measured by the probe pulse are obtained from the imaginary part of the dielectric function. For a linearly polarized probe pulse, we have ) s t i n u . b r a ( r e w o P k ( /T) E44 E77 (c) ) s t i n u . b r a ( r e w o P d e a r g e n t t I 00 150 300 450 600 900 2 1 Pump Energy (eV) 3 ε2(ω, t) = 8π2e2 At(ω)2 (cid:18) 2 m0(cid:19) Xss′µµ′Z dk π ǫ · ~P µµ′ ss′ (k)2 ×(cid:16)fsµ(k, t) − fs′µ′ (k, t)(cid:17) δ(cid:16)∆Eµµ′ ss′ (k) − ω(cid:17) (8) where At = π(dt/2)2 is the cross-sectional area of the tube and dt is the equilibrium nanotube diameter. The dirac delta function is replaced by a broadened Lorentzian of the form shown in Eq. (4). The photon energy of the probe pulse photons is ω, and ǫ is the probe unit polarization vector. In our experiments, the photon energy ω is the same for both pump and probe, while the polarization vector ǫ may be different for pump and probe. To obtain the computed coherent phonon spectrum, we take the power spectrum of the computed differential transmission signal after background subtrac- tion using the Lomb periodogram algorithm described in Ref. 32. V. COMPARISON OF THEORY AND EXPERIMENT To compare experiment with theory, we performed simulations of polarization-dependent CP spectroscopy on a (38,0) zigzag nanotube. This is a mod-2 semicon- ducting tube with a diameter of 3.01 nm. Our sample contains an ensemble of nanotubes with diameters cen- tered around 3 nm,26 and we expect that (38,0) tubes FIG. 6: (color online) (a) Theoretical bandstructure for (38,0) nanotubes showing the strong E44 and E77 transitions for a polarization vector parallel to the tube. The wavevector k is expressed in units of π/T where T = 4.31 A is the length of the translational unit cell. (b) Computed absorption spectra for polarization angles varying from 0◦ (parallel to tube) to 90◦ (perpendicular to tube). (c) Coherent phonon spectra as a function of pump energy and pump polarization angle φ for RBM coherent phonons (ω = 10.91 meV). Coherent phonon spectra are Fourier transforms of computed time dependent differential transmission for Type II pump-probe experiments. (d) Integrated power (black dots) obtained by taking the area under the computed E44 peaks in lower left panel. We fit (red curve) the theoretical calculations to A cosp(φ) where A and p = 4.017 are fitting parameters. will contribute strongly to the CP signal for the ensem- ble since (i) the (38,0) tube has a diameter in the center of the measured diameter distribution and (ii) mod-2 zigzag tubes tend to have very strong intrinsic CP signals.18,22,23 Our theoretical results for the (38,0) nanotube are shown in Fig. 6. Our computed electronic π band struc- ture for the (38,0) tube is shown in Fig. 6(a). The bands are doubly degenerate with two distinct cutting lines cor- responding to each band. The π valence bands have neg- ative energy, and the π∗ conduction bands have posi- tive energy. For z-polarized light, the allowed E44 and E77 transitions (selection rule ∆µ = 0) giving rise to the strongest CP signals are indicated by vertical arrows. The absorption coefficient for (38,0) nanotubes as a function of linearly-polarized photon energy is shown in Fig. 6(b) for polarization angles varying from 0◦ (parallel to tube) to 90◦ (perpendicular to tube). For the (38,0) nanotube, the RBM phonons have a computed energy ω = 10.91 meV and an oscillation period of 379 fs. We simulate the generation of RBM coherent phonons by pumping with a 50 fs pump pulse, which is much shorter than the RBM oscillation period. Polarization-dependent CP spectra for RBM coherent phonons are shown in Fig. 6(c) for pump polarization angles φ ranging from 0◦ to 90◦. In these simulations, the probe polarization is kept fixed parallel to the tube axis (Type II geometry). As can be seen in the figure, the CP signal is maximum when the pump is polarized paral- lel to the tube. As the pump photon energy is varied, we excite RBM coherent phonons by successively photoex- citing carriers in different bands. The strongest RBM CP signals are obtained by pumping at the E44 and E77 transitions. As the pump polarization angle increases, the RBM CP signal decreases until it is finally quenched when the pump polarization is perpendicular to the tube. By taking the area under the RBM CP signal curves in the vicinity of the computed E44 transition energy near 1.48 eV, we obtain the integrated power as a function of pump polarization angle shown as black dots in Fig. 6(d). The integrated CP power can be well fit with a fitting function of the form A cosp(θ), where A and p = 4.017 are fitting parameters. The polarization dependence of the RBM CP inte- grated power in Fig. 6(d) can be understood by examin- ing Fig. 7. In Fig. 7(b) we plot the coherent RBM diame- ter oscillations as a function of the pump polarization an- gle θ. Fitting the theoretical results to A cosp(θ), we ob- tain p = 2.11 as indicated in the figure. For the RBM di- ameter oscillations we get p ∼ 2 which makes sense. The driving function S(t) is proportional to the photogener- ated carrier density as seen in Eq. (2) and this in turn is proportional to the squared optical matrix element which has a cos2(θ) angular dependence3. In Fig. 7(a) we plot the amplitude of the time-dependent differential trans- mission oscillations measured by the probe beam after background subtraction. For the differential transmission signal, we get p ∼ 2 for Type II and p ∼ 4 for Type I. This also makes sense. In Type II experiments the probe polarization is parallel to the tube axis and the pump po- larization angle θ is varied. In this case the signal should be proportional to the amplitude of the diameter oscil- lations. In Type I experiments on the other hand, the pump and probe polarizations are parallel and we should get an extra factor of cos2(φ) to account for anisotropy of the probe beam absorption. In our CP spectroscopy measurements, we extract the power spectrum of the dif- ferential transmission oscillations. The CP power spec- trum should be proportional to the square of the am- plitude of the differential transmission signal which we anticipate will give us a cos4(φ) polarization dependence 6 (a) 1.64 (b) 3.65 Type I Type II (38,0) E44 2.11 1.0 0.5 0.0 0.04 0.02 ) . b r a ( e d u t i l p m A T T / ) A ( e d u t i l p m A M B R 0.00 -90 -60 RBM 10.91 meV -30 30 0 Angle (degrees) 60 90 FIG. 7: (color online) In (b) we plot the amplitude of RBM coherent phonon oscillations as a function of the pump polar- ization angle (θ or φ) for a (38,0) nanotube photoexcited by a 50 fs pump at the theoretical E44 transition. In (a) we plot the amplitude of the resulting differential transmission signal for Type I and II experiments as a function of angles θ and φ respectively. Our results are fit to A cosp(θ) [A cosp(φ)] where numerical values of the best fit p are indicated in the figure. for Type II CP spectroscopy experiments and a cos8(θ) dependence for Type I CP spectroscopy experiments. The experimental integrated CP power is obtained by taking the Fourier transform power spectrum of the time- dependent differential transmission data shown in Fig. 3 and then computing the area under the E44 peak in the resulting spectrum. The results for our Type I and Type II experiments are shown in Fig. 8, where the experi- mental data points are plotted as downward-pointing red triangles and the corresponding theoretical predictions are shown as upward-pointing black triangles. In over- laying the experimental and theoretical data points, we subtracted a background from the experimental data to obtain complete quenching at 90◦. We then rescaled ex- perimental and theoretical data (both in arbitrary units) so that the integrated power at 0◦ is equal to unity. We then fit experimental and theoretical data to functions of the form A cosp(θ) [A cosp(φ)]. The best fit functions for our experimental and theoretical results are shown in Fig. 8 as solid red and dashed black lines, respectively. For the Type II experiments, where the probe po- larization is fixed parallel to the average tube orienta- tion, we get decent agreement between theory and ex- periment with best fit parameters p = 4.017 for theory and p = 3.630 for experiment. For the Type I exper- iments, where pump and probe polarization vectors are parallel to each other, there is a discrepancy between the- ory and experiment. As can be seen in Fig. 8(a), the best ) s t i n u . b r a ( r e w o P d e t a r g e t n I 1.0 0.8 0.6 0.4 0.2 E44 (a) Type I pump probe e b u t Type II e b u t e b o r p (b) pump 8.308 Theory (38,0) 4.435 Expt 3.630 Expt 4.017 Theory (38,0) 0.0 0 30 60 30 Angle (degrees) 0 60 90 7 1.0 0.8 0.6 0.4 0.2 0.0 0 30 Standard Deviation Angle degrees) 0 10 20 30 (a) Type I A cos8( (b) Type II A cos4( 60 0 30 Angle (degrees) 60 90 ) s t i n u . b r a ( r e w o P d e t a r g e t n I FIG. 8: (color online) Integrated RBM coherent phonon power as a function of (a) θ for Type I and (b) φ for Type II experiments pumping close to the E44 transition. Experi- mental data points are red downward pointing triangles fit by a solid red line. Theoretical data points for a (38,0) nanotube are black upward pointing triangles and are fit by a black dashed line. The fitting functions are of the form A cosp(θ) and A cosp(φ) where the numerical values of the best fit p are indicated in the figure. The experimental curves have under- gone background subtraction to obtain complete quenching at an angle of 90◦ and then rescaled so the integrated power at an angle of 0◦ is equal to unity. fit parameters are p = 8.308 for theory and p = 4.435 for experiment. The experimental fits in Fig. 8 would imply that the CP intensity scales roughly as cos4(θ) indepen- dently of the probe polarization. This seems unlikely given the anisotropy of the absorption coefficient in car- bon nanotubes. In fact, from our theory, we would expect p ∼ 4 for Type II and p ∼ 8 for Type I. We believe that the cause of this discrepancy is most likely due to misalignment effects. Our sample consists of an ensemble of nanotubes lying horizontally on a sap- phire substrate. While the tubes are highly-aligned, they are not perfectly aligned. To consider the effects of mis- alignment on the experimentally-measured integrated CP intensity in a Type I experiment, we assume that the tube alignment angles on the transferred film are described by a Gaussian distribution function with a small standard deviation ∆θ. If the angle between the pump polariza- tion vector and the ensemble averaged tube axis is θ, then the angles ϑ between the pump polarization vector and the axes of each tube in the ensemble are described by a Gaussian distribution P (ϑ, θ, ∆θ) = 1 √2π(∆θ) exp(cid:18)− (ϑ − θ)2 2(∆θ)2 (cid:19) (9) If the integrated RBM coherent phonon power for each nanotube in the ensemble is given by A cosp(ϑ), the en- semble averaged integrated power Icp(θ, ∆θ) is obtained FIG. 9: (color online) Effects of tube misalignment on inte- grated coherent phonon power for an ensemble of nanotubes with tube axis orientation angles following a Gaussian distri- bution. (a) Type I fitting function A cosp(θ) with p = 8 and (b) Type II fitting function A cosp(φ) with p = 4. by taking the ensemble average over ϑ Icp(θ, ∆θ) = A Z ∞ −∞ dϑ P (ϑ, θ, ∆θ) cosp(ϑ) (10) In Eq. (10), we extended the integration limits on ϑ to infinity in the limit of small ∆θ. If p is a positive in- teger, Icp(θ, ∆θ) can be found analytically. Otherwise, it can be evaluated numerically. In the case of Type II experiments, the ensemble averaged integrated power is obtained by replacing θ with φ in Eqs. (9) and (10). The effects of nanotube misalignment on the integrated CP power fitting function for an ensemble of Gaussian misaligned tubes is illustrated in Fig. 9, assuming p = 8 for Type I and p = 4 for Type II experiments. As the standard deviations ∆θ [∆φ] increase, quenching of the CP intensity is reduced and Icp(θ, ∆θ) [Icp(φ, ∆φ)] become flatter. Using this model, we are able to get reasonable fits to the experimentally measured CP intensity in both Type I and Type II experiments. We assume a fitting function of the form A cosp(θ + ∆θ) + B [A cosp(φ + ∆φ) + B], where A and B are background subtraction and rescaling pa- rameters and ∆θ [∆φ] is a random Gaussian distributed misalignment angle (standard deviation ∆θ [∆φ]). In our fitting procedure, we set p = 8 for Type I and p = 4 for Type II and use the same standard deviation for the tube misalignment angles in fitting both Type I and II data. The results of our fitting procedure are shown in Fig. 10, where the best fits for Type I and Type II geometries are shown as dashed black and solid red lines, respectively. Our best fit standard deviation is ∆θ = ∆φ = 18.7◦, which implies that the tube align- ment angles on the sapphire substrate are 0◦ ± 9.35◦. ) s t i n u . b r a ( r e w o P d e t a r g e t n I 1.0 0.8 0.6 0.4 0.2 0.0 0 p = 8 Type I pu m p pro b e e b u t p = 4 Type II pu m p e b o r p e b u t Type I Type II 30 E44 Experiment 90 60 120 Angle (degrees) 150 180 FIG. 10: (color online) Experimental integrated RBM coher- ent phonon power for the E44 transition as a function of θ for Type I (black upward pointing triangles) and φ for Type II (red downward pointing triangles) experiments. Type I exper- iments are fit to A cosp(θ+∆θ)+B where p = 8 (black dashed line) and Type II experiments are fit to A cosp(φ+∆φ)+B for p = 4 (solid red line). A and B are background subtraction and rescaling parameters and the standard deviations for the random tube axis misalignment ∆θ and ∆φ are restricted to be the same for both Type I and II. For small ∆θ (measured in radians) the nematic order parameter is S = exp(−2 (∆θ)2) = 0.81. Although a value of S = 0.81 indicates a strongly aligned sample, it is not perfectly aligned (S = 1). This is unlike previous work in the THz regime with these highly-aligned samples, where the nematic order param- eter was calculated to be exactly S = 1.17 It is clear that there is a wavelength dependence of the nematic order parameter, but we believe this can be explained quali- tatively: any slight misalignment in the sample would go undetected in the THz regime, as the wavelength is much larger than our visible wavelengths that can de- tect the misalignments. This could explain the differ- ence in calculated nematic order parameter between our 8 CP measurements and previous THz measurements, but nonetheless, regardless of the calculated differences with wavelength, our calculated nematic order parameter is still quite large and we can confidently confirm the high degree of alignment of the sample with our CP measure- ments. VI. CONCLUSION In summary, we investigated the polarization anisotropy of coherent phonon dynamics in highly- aligned single-walled carbon nanotubes and measured RBM coherent phonons as a function of polarization angle. We saw a very nearly complete quenching of the RBM for both geometries and extended our results to determine the degree of alignment of the sample. Comparing our results with theory, we performed simulations of polarization-dependent CP spectroscopy on a (38,0) zigzag nanotubes and also found a decrease in CP signal as optical polarization varies from parallel to perpendicular to the nanotube axis. Using those simulated results, we also theoretically determined a cos8(θ) dependence for Type I CP spectroscopy experi- ments and a cos4(φ) polarization dependence for Type II CP spectroscopy experiments. Including misalignment effects to our fitting, we finally determined the nematic order parameter of our sample to be 0.81. Acknowledgments This work was supported by the National Science Foundation under grant numbers DMR-0325474, OISE- 0530220, and DMR-0706313, the Robert A. Welch foun- dation under grant number C-1509, and the Office of Naval Research (ONR) under contract number 00075094. Y.-S. Lim and K.-J. Yee are supported by a Korea Science and Engineering Foundation (KOSEF) grant funded by the Korean Government (Most) (R01-2007-000-20651-0). We acknowledge useful discussions with Andrew Rinzler at the University of Florida. ∗ kono@rice.edu; www.ece.rice.edu/~kono; corresponding author. 1 R. Saito, G. Dresselhaus, and M. S. Dresselhaus, Physi- cal properties of carbon nanotubes (World Scientific, Sin- gapore, 2003). 2 J.-C. Charlier, X. Blase, and S. Roche, Rev. Mod. Phys. 79, 643 (2007). M. S. S. Dantas, M. A. Pimenta, A. M. Rao, R. Saito, C. Liu, and H. M. Cheng, Phys. Rev. Lett. 85, 2617 (2000). 6 Z. M. Li, Z. K. Tang, H. J. Liu, N. Wang, C. T. Chan, R. Saito, S. Okada, G. F. Li, J. S. Chen, N. Nagasawa, et al., Phys. Rev. Lett. 87, 127401 (2001). 7 A. Hartschuh, H. N. Pedrosa, L. Novotny, and T. D. Krauss, Science 301, 1354 (2003). 3 H. H. Gommans, J. W. Alldredge, H. Tashiro, J. Park, J. Magnuson, and A. G. Rinzler, J. Appl. Phys. 88, 2509 (2000). 4 G. S. Duesberg, I. Loa, M. Burghard, K. Syassen, and S. Roth, Phys. Rev. Lett. 85, 5436 (2000). 8 M. F. Islam, D. E. Milkie, C. L. Kane, A. G. Yodh, and J. M. Kikkawa, Phys. Rev. Lett. 93, 037404 (2004). 9 H. Ajiki and T. Ando, J. Phys. Soc. Jpn. 62, 2470 (1993). 10 J. P. Lu, Phys. Rev. Lett. 74, 1123 (1995). 11 M. A. L. Marques, M. d'Avezac, and F. Mauri, Phys. Rev. 5 A. Jorio, G. Dresselhaus, M. S. Dresselhaus, M. Souza, B 73, 125433 (2006). 12 S. Zaric, G. N. Ostojic, J. Kono, J. Shaver, V. C. Moore, R. H. Hauge, R. E. Smalley, and X. Wei, Nano Lett. 4, 2219 (2004). 13 S. Zaric, G. N. Ostojic, J. Kono, J. Shaver, V. C. Moore, M. S. Strano, R. H. Hauge, R. E. Smalley, and X. Wei, Science 304, 1129 (2004). 14 M. F. Islam, D. E. Milkie, O. N. Torrens, A. G. Yodh, and J. M. Kikkawa, Phys. Rev. B 71, 201401 (2005). 15 J. Shaver, A. N. G. Parra-Vasquez, S. Hansel, O. Portugall, C. H. Mielke, M. von Ortenberg, R. H. Hauge, M. Pasquali, and J. Kono, ACS Nano 3, 131 (2009). 16 T. A. Searles, Y. Imanaka, T. Takamasu, H. Ajiki, J. A. Fagan, E. K. Hobbie, and J. Kono, arXiv (2010), 1001.0524v1. 17 L. Ren, C. L. Pint, L. G. Booshehri, W. D. Rice, X. Wang, D. J. Hilton, K. Takeya, I. Kawayama, M. Tonouchi, R. H. Hauge, et al., Nano Lett. 9, 2610 (2009). 18 Y.-S. Lim, K.-J. Yee, J.-H. Kim, E. H. H´aroz, J. Shaver, J. Kono, S. K. Doorn, R. H. Hauge, and R. E. Smalley, Nano Lett. 6, 2696 (2006). 19 A. Gambetta, C. Manzoni, E. Menna, M. Meneghetti, G. Cerullo, G. Lanzani, S. Tretiak, A. Piryatinski, A. Sax- ena, R. L. Martin, et al., Nature Physics 2, 515 (2006). 20 K. Kato, K. Ishioka, M. Kitajima, J. Tang, R. Saito, and H. Petek, Nano Lett. 8, 3102 (2008). 21 J.-H. Kim, J. Park, B. Y. Lee, D. Lee, K.-J. Yee, Y.-S. Lim, L. G. Booshehri, E. H. H´aroz, J. Kono, and S.-H. Baik, J. Appl. Phys. 105, 103506 (2009). 9 22 G. D. Sanders, C. J. Stanton, J.-H. Kim, K.-J. Yee, Y.-S. Lim, E. H. H´aroz, L. G. Booshehri, J. Kono, and R. Saito, Phys. Rev. B 79, 205434 (2009). 23 J.-H. Kim, K.-J. Han, N.-J. Kim, K.-J. Yee, Y.-S. Lim, G. D. Sanders, C. J. Stanton, L. G. Booshehri, E. H. H´aroz, and J. Kono, Phys. Rev. Lett. 102, 037402 (2009). 24 Y.-S. Lim, J.-G. Ahn, J.-H. Kim, K.-J. Yee, T. Joo, S.-H. Baik, E. H. H´aroz, L. G. Booshehri, and J. Kono, ACS Nano (2010), published online on May 13, 2010. 25 C. L. Pint, Y.-Q. Xu, M. Pasquali, and R. H. Hauge, ACS Nano 2, 1871 (2008). 26 C. L. Pint, Y.-Q. Xu, S. Moghazy, T. Cherukuri, N. T. Alvarez, E. H. H´aroz, S. Mahzooni, S. K. Doorn, J. Kono, M. Pasquali, et al., ACS Nano 4, 1131 (2010). 27 D. Porezag, T. Frauenheim, T. Kohler, G. Seifert, and R. Kaschner, Phys. Rev. B 51, 12947 (1995). 28 J.-W. Jiang, H. Tang, B.-S. Wang, and Z.-B. Su, Phys. Rev. B 73, 235434 (2006). 29 C. Lobo and J. L. Martins, Z. Phys. D 39, 159 (1997). 30 A. V. Kuznetsov and C. J. Stanton, Phys. Rev. Lett. 73, 3243 (1994). 31 S. L. Chuang, Physics of Optoelectronic Devices (Wiley, New York, 1995). 32 W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery, Numerical Recipes (Cambridge University Press, New York, 1992).
1607.04802
2
1607
2016-09-07T10:45:25
Heat production and error probability relation in Landauer reset at effective temperature
[ "cond-mat.mes-hall", "cond-mat.stat-mech" ]
The erasure of a classical bit of information is a dissipative process. The minimum heat produced during this operation has been theorized by Rolf Landauer in 1961 to be equal to $k_B T \ln 2$ and takes the name of Landauer limit, Landauer reset or Landauer principle. Despite its fundamental importance, the Landauer limit remained untested experimentally for more than fifty years until recently when it has been tested using colloidal particles and magnetic dots. Experimental measurements on different devices, like micro-mechanical systems or nano-electronic devices are still missing. Here we show the results obtained in performing the Landauer reset operation in a micro-mechanical system, operated at an effective temperature. The measured heat exchange is in accordance with the theory reaching values close to the expected limit. The data obtained for the heat production is then correlated to the probability of error in accomplishing the reset operation.
cond-mat.mes-hall
cond-mat
Heat production and error probability relation in Landauer reset at effective temperature Igor Neri1,2,*,+ and Miquel L ´opez-Su´arez1,!,+ 1NiPS Laboratory, Dipartimento di Fisica e Geologia, Universit`a degli Studi di Perugia, 06123 Perugia, Italy 2INFN Sezione di Perugia, via Pascoli, 06123 Perugia, Italy *igor.neri@nipslab.org !miquel.lopez@nipslab.org +these authors contributed equally to this work ABSTRACT The erasure of a classical bit of information is a dissipative process. The minimum heat produced during this operation has been theorized by Rolf Landauer in 1961 to be equal to kBT ln 2 and takes the name of Landauer limit, Landauer reset or Landauer principle. Despite its fundamental importance, the Landauer limit remained untested experimentally for more than fifty years until recently when it has been tested using colloidal particles and magnetic dots. Experimental measurements on different devices, like micro-mechanical systems or nano-electronic devices are still missing. Here we show the results obtained in performing the Landauer reset operation in a micro-mechanical system, operated at an effective temperature. The measured heat exchange is in accordance with the theory reaching values close to the expected limit. The data obtained for the heat production is then correlated to the probability of error in accomplishing the reset operation. Introduction The minimum energy required to reset one bit of information represents one of the fundamental limits of computation arising when one bit of information is erased or destroyed. Starting with a system encoding two possible states with the same probability and finishing the procedure with only one possible state, the variation of entropy in the system is equal to the difference of entropy between the final state and the initial one, D S = kB ln 1 − kB ln 2 = −kB ln 2. On average this reduction of entropy has to be accompanied with an increase of heat in the surrounding environment in order to not violate the second law of thermodynamics, QL ≥ kBT ln 2. This limit takes the name of Landauer limit and was theorized by Rolf Landauer in the 60's1, and remained untested for over fifty years. The development of computational methods allowed to evaluate tiny amounts of heat exchanged and recent technical advances in micro- and nano-fabrication made possible to recently test the validity of the Landauer principle. In particular, the first experimental verification of the Landauer principle has been carried out by B´erut et al. using a colloidal particle trapped in optical tweezers2. More recently Jun et al. presented similar results considering a colloidal particle in a feedback trap3. Finally, the last experimental verification of the Landauer limit has been performed by Hong et al. in a nanomagnetic system4. In this case information is encoded in the magnetization state of the system while an external magnetic field is used in order to flip between two preferred magnetization states. The evaluation of the heat produced during the reset operation was demonstrated to be compatible with the Landauer limit. Even if there is no doubt nowadays on the validity of the Landauer principle, the test on micro-mechanical systems is still missing. We have recently shown that it is possible to use electro-mechanical devices to accomplish basic5 and complex logic operations6 with an arbitrarily low energy expenditure. Therefore the possibility to use this class of devices for memory storage completes the logic architecture for a complete computing device. In the following we show the measurement on heat production when performing the reset operation in a novel memory unit based on a bistable mechanical cantilever at effective temperature. The results are in good agreement with the Landauer limit. The dependence of the dissipation with the error rate is also investigated showing a trend in accordance with the theory10. Results Bistable micro-mechanical system The mechanical system used to perform the experiment is depicted in Figure 1 (a). A triangular micro-cantilever, 200µm long, is used to encode one bit on information. In order to obtain two stable states two magnets with opposite magnetization are placed on the tip of the cantilever and on a movable stage facing the cantilever. In this way, depending on the distance between the magnets, d, and the relative lateral alignment, D x, it is possible to induce bistability on the system. Figure 1 (b) shows the potential energy as a function of d reconstructed from the probability density function of the position of the cantilever at equilibrium, r (x,d) = A exp(−U(x,d)/kBT ), which implies that U(x,d) = −kBT lnr (x,d) + U0 2. When the magnets are far away the effect of the repulsive force is negligible, the system is then monostable and can be approximated to a linear system. Decreasing the distance the repulsive force between magnets tends to soften the system up to the point where two stable positions appear. The effect of reducing even more d is to enlarge the separation of the rest states and to increase the potential barrier separating these two wells. Eventually, when the distance between the magnets is small enough, the system remains trapped in one well for a period of time larger than the relaxation time of the system. Logic states are encoded in the position of the cantilever tip: logic 0 for x < 0 and logic 1 for x > 0. The proposed system presents intrinsic dissipative processes that depend on the maximum displacement of the cantilever tip6. The minimum heat produced when performing a physical transformation of the system is proportional to x2 max. In our setup it is not possible to reduce the separation between the two potential wells to a value that bounds the heat produced by intrinsic dissipation below the Landauer limit. Increasing the effective temperature increases the value for the Landauer limit making possible to have negligible intrinsic dissipation. A piezoelectric shaker is used to excite the structure with a band limited white Gaussian noise to mimic the effect of an arbitrary temperature. In the present experiment the white noise is limited to 50kHz, well above the resonance frequency of the free cantilever ( f0=5.3kHz). The dependence of the effective temperature with the root-mean-squared voltage supplied to the shaker is reported in Figure 1 (c). The red dot, corresponding to an effective temperature of Teff = 5 × 107 K, highlights the condition considered in the present case. The solid line represent the expected trend where T (cid:181) V 2 7. The effective rms temperature has been estimated computing the power spectral density (PSD) of the system at various piezoelectric noise excitation voltages. The obtained curves have been fitted with Lorenzian curves taking as reference for the calibration the one at room temperature (T =300K). The other curves have been used to extract the only varying parameter Teff, corresponding to the effective temperature of the system under external excitation. Finally two electrostatic probes, placed one on the left and the other on the right of the cantilever, are used to apply a negative and positive forces respectively. When a voltage different to zero is applied on one probe the cantilever feels an attractive electrostatic force toward the probe due to the polarization of the cantilever itself. The voltage on the probes, the distance between the magnets and their time evolution are used to specify the protocols used in order to change the bit stored in the system as described in the following subsection. Reset protocol Varying the magnets distance d the barrier separating the two stable wells varies from the minimum value B0 for d=3.65 a.u. to Bmax for d=2.8 a.u. Applying a voltage on one probe corresponds to apply a force toward the probe itself. Thus a voltage on the left probe corresponds to a negative force and a voltage on the right probe corresponds to a positive force. Details of the force calibration are presented in methods section. In Figure 2 (a) the protocol followed to reset the bit to the state 0 and 1 is presented. The procedure is similar to the ones presented in Refs. 8 and 2. Initially the barrier separating the two stable states is removed moving the magnet away (red curve in the first panel) making the system monostable. Once the barrier is removed we apply a negative (positive) force to reset the bit status to 0 (1) applying a finite voltage VL (VR) on the left (right) probe. This is represented by the magenta curve in Figure 2 (a). Once the force is applied we restore the barrier to its original value. Finally, we remove the lateral force finishing in the original parameters configuration. At the end of the operation, if there are no errors the cantilever position encodes the desired bit of information. In order to be sure to perform the operation starting from both initial states, we mimic a statistical ensemble where the initial probability is 50% to start in the left well and 50% in the right well. A trace of the cantilever tip position, x, is shown in Figure 2 (a) (black line), where the dashed blue lines represent the two stable positions once the barrier is restored. When the barrier is removed the system goes from the local prepared state to an undefined state with a free expansion, the entropy on the system thus increases in a irreversible manner8, 9. This increment is related to uncontrollable transitions from one well to the other once the barrier height is close to kBT . These large excursions of the cantilever can be seen in the time series of position. Figure 2 (b) shows a representation of the time evolution of the potential energy during the set of the bit to the logic state 1. In a first step the barrier is removed allowing the system to oscillate in a monostable potential landscape. Then the potential is slightly tiled and when the barrier is restored the system is confined in the desired state. After these stages the bit is set to the state 1. In order to have a reliable measure of the heat produced during the considered operations, reset protocols are repeated for 800 times in order to have a large enough statistic. Heat vs error probability In the optimal case the initial configuration is a mixed logical 0 and 1 where both states have the same probability while the final configuration is the selected state with a 100% probability. This corresponds to an entropy variation of D S = −kB ln(2) and a minimum heat produced of Q ≥ −TD S = QL. It has been shown that the bound QL = −kBT ln(2) applies only for 11. In symmetric potentials. Considering asymmetries on the system the minimum produced heat can be lowered below QL our setup the system is slightly asymmetric and we have evaluated the variation of entropy from the initial to the final state 2/8 from the probability density function of the tip position, r (x), being D SG = −0.61kB D SS = −0.68kB for the Gibbs and Shannon entropy respectively, both close to −kB ln(2). If we consider the possibility to commit errors during the reset operation the heat produced becomes a function of the probability of success10: Q(Ps) ≥ kBT [ln(2) + Ps ln(Ps) + (1 − Ps)ln(1 − Ps)] (1) where Ps is the probability of success or success rate. When Ps is 0.5 no reset operation is performed and thus there is no minimum heat to be produced during the operation3, 10. In Figure 3 (a) we present the average heat produced for the reset operation as function of the lateral alignment of the counter magnet D x. When the system is aligned closely to perfection (i.e., D x ≈ 0) we estimate a heat production slightly above kBT and below two times QL. Asymmetrizing the potential, by means of setting D x 6= 0, the heat produced tends to decrease reaching values this time below QL. However, in this conditions the error rate in performing the reset operation have a major role, in fact in this configuration the probability of success, Ps, decrease rapidly. This is represented by the color map of dots in Figure 3 (a), where green represents higher success rate while blue represents a higher probability of error. In Figure 3 (b) the success rate of the reset operation is reported as function of the lateral alignment. Solid violet circles represent the overall success rate while red and black symbols represent the error rate for resetting to 1 or to 0 respectively. Circles are used to report the error probability for the same initial and final state while crosses are used for 0 to 1 and 1 to 0 transitions. For instance let us consider the case where D x < 0: the counter magnet is moved towards the right and as a consequence the 0 state is more favorable respect the 1 state. From Figure 3 (b) we can see that for D x < 0 the probability of resetting toward 0 is almost 100% while the probability of resetting toward 1 decreases rapidly reaching values below 50%. The same behavior is present in the case D x > 0, where the counter magnet is moved to the left, where the state 1 is more favorable. We can now correlate the heat produced to the probability of success for resetting to 0 and 1 as presented in Figure 3 (c). Dashed lines represent the Landauer limit for a 100% of success rate (≈ 0.7kBT ). While in both cases the heat produced is above the Landauer limit, in the reset to 0 case the obtained values are very close to QL. As expected, decreasing the success rate the obtained values goes below the Landauer limit for both cases accordingly to Equation 1. As it is well known the adiabatic limit in presence of dissipation mechanisms like viscous damping can be only reached if the operation is performed slowly when these mechanisms are negligible. We increased the protocol time for the reset operation from 0.25s up to 3.5s. The results are presented in Figure 3 (d). Increasing the protocol time decreases the heat production reaching values well below the Landauer limit. Notice that in these cases where Q < QL the Ps is well below 1 since the system has more time to relax and therefore tends to thermalize before the reset operation is correctly performed. In fact for protocols lasting more than 1 s the success rate is below 75%. Discussion We have measured the intrinsic minimum energy dissipation during the reset of one bit of information in a micro-mechanical system. We have considered a completely different physical system respect to the existing literature, i.e., micro-electro- mechanical system. To achieve these results we have performed the experiment at an effective temperature of 5 × 107 K in order to make the intrinsic dissipation of the mechanical structure negligible respect to the thermodynamic contribution. In these conditions we have reached values of heat produces consistent with the Landauer limit approaching it closely. Moreover we presented experimental data relating the minimum heat produced with the probability of success of resetting one bit of information. Nowadays where there is a lot of attention on micro-electro-mechanical systems able to perform computation at arbitrary low energy, the achieved results have a significant importance in the development of new computing paradigms based on systems different from the well established CMOS technology. Methods Setup preparation and calibration The micro-cantilever used is a commercial atomic force microscopy (AFM) probe (NanoWorld PNP-TR-TL12). It is long 200µm, with a nominal stiffness k=0.08Nm−1 and a nominal resonace frequency of 17kHz. A fragment of NdFeB (neodymium) magnet is attached to the cantilever tip with bi-component epoxy resin. To set the magnetization to a known direction the sys- tem is heated up to 670K, above its Curie temperature13 in the presence of a strong external magnetic field with the desired orientation. With this additional mass the resonance frequency decreases to 5.3kHz. The quality factor of the system has been estimated from the power spectral density of the displacement, x, fitted with a Lorenzian curve, giving a value of Qf = 320. The deflection of the cantilever, x, is determined by means of an AFM-like laser optical lever. A small bend of the cantilever provokes the deflection of a laser beam incident to the cantilever tip that can be detected with a two quadrants photo detector. 3/8 The laser beam is focused on the cantilever tip with an optical lens (focal length f =50mm). For small deflections the response of the photo detector remains linear, thus x = rxD VPD, where VPD is the voltage difference generated by the two quadrants of the photo detector. In order to determine rx we look at the frequency response of the system as in6 under thermal excitation. The relation between the measured voltage and the expected displacement gives rx =1.8365 × 10−5m V−1. The system is placed in a vacuum chamber and isolated from seismic vibrations to maximize the signal-to-noise ratio. All measurements were performed at pressure P =4.7 × 10−2 mbar. Effective temperature estimation As the system is modeled as a harmonic oscillator with one degree of freedom, according to the Equipartition Theorem the thermal energy present in the system is simply related to the cantilever fluctuations as 1 2 khx2i. According to Parseval's 0 G(w )2 dw where G(w ) stands for the PSD of the system. This takes the form of a Lorenzian function 2 kBT = 1 theorem hx2i =R G(w ) =s 4kBT Qfkw 0vuut (cid:16)1 − w 2 w 2 0(cid:17)2 1 + 1 Q2 f w 2 w 2 0 where w = 2p only free parameter, T , from the measured PSDs to the expected function G(w ). f . Once the system has been calibrated at room temperature we estimated the effective temperatures fitting the Force calibration A set of two electrodes (see Fig. 1(a)) is used in order to polarize the cantilever producing a bend on the mechanical structure. Electrostatic forces depend on the voltage applied to the electrodes, i.e. VL and VR for the left and right electrode respectively. Since the restoring force of the cantilever can be expressed as Fk = −kx, the relation between applied voltage to the probe and the force acting on the cantilever has been estimated in static conditions assuming Fk = Fel. The relation between the applied voltages, VL and VR, and the electrostatic force Fel is then fitted with a 9th degree polynomial. Heat production evaluation The work performed on the system along a given trajectory x(t) is given by the integral14, 15: W =Z 0 t p M(cid:229) k=1 ¶ U(x,lll ) ¶l k ¶l k ¶ t dt (2) where t p is the protocol time duration, U(x,lll ) is the total potential energy of the system and lll is a vector containing all the M control parameters. In our case we have two controls parameter, the voltage applied to the piezoelectric stage to control the energy barrier, and the electrostatic forces. To obtain the heat produced Q we have to consider the variation in internal energy, D E. Ideally the potential energy at both the bottom wells is the same, however considering asymmetries on the system D E can be different from zero. The evaluation of the total energy variation D E is obtained from the reconstructed potential energy, U, and the variation of the kinetic energy. The latter quantity is however negligible even for the shortest t p, where the total kinetic energy variation is one order of magnitude lower than QL (1.4 × 10−17J versus 5.3 × 10−16 J). Finally the heat produced is obtained from Q = W − D E. References 1. Landauer, R. Irreversibility and heat generation in the computing process. IBM journal of research and development 5, 183 -- 1911961. 2. B´erut, A. et al. Experimental verification of Landauer's principle linking information and thermodynamics. Nature 483, 187 -- 189 (2012). 3. Jun, Y., Gavrilov, M. & Bechhoefer, J. High-precision test of Landauer's principle in a feedback trap. Physical review letters 113, 190601 (2014). 4. Hong, J., Lambson, B., Dhuey, S. & Bokor, J. Experimental test of Landauer's principle in single-bit operations on nanomagnetic memory bits. Science Advances 2 (2016). 5. Lopez-Suarez, M., Neri, I. & Gammaitoni, L. Operating micromechanical logic gates below kBT: Physical vs logical reversibility. In Energy Efficient Electronic Systems (E3S), 2015 Fourth Berkeley Symposium on, 1 -- 2 (IEEE, 2015). 6. Lopez-Suarez, M., Neri, I. & Gammaitoni, L. Sub kBT micro electromechanical irreversible logic gate. Nature Commu- nication 7 12068 (2016). 4/8 ¥ 7. Venstra, W. J., Westra, H. J. & van der Zant, H. S. Stochastic switching of cantilever motion. Nature communications 4 (2013). 8. Bennett, C. H. The thermodynamics of computation -- a review. International Journal of Theoretical Physics 21, 905 -- 940 (1982). 9. Crooks, Gavin E. Entropy production fluctuation theorem and the nonequilibrium work relation for free energy differences. Physical Review E 60.3 2721 (1999). 10. Gammaitoni, L., Chiuchi´u, D., Madami, M. & Carlotti, G. Towards zero-power ICT. Nanotechnology 26, 222001 (2015). 11. Sagawa, Takahiro. Thermodynamic and logical reversibilities revisited. Journal of Statistical Mechanics: Theory and Experiment 2014.3 P03025 (2014) 12. NanoWorld - AFM tip - PNP-TR-TL - Pyrex-Nitride. http://www.nanoworld.com/pyrex-nitride-triangular-silicon-nitride- tipless-cantilever-afm-tip-pnp-tr-tl. Accessed: 2016-05-30. 13. Ma, B. et al. Recent development in bonded NdFeB magnets. Journal of magnetism and magnetic materials 239, 418 -- 423 (2002). 14. Seifert, U. Stochastic thermodynamics, fluctuation theorems and molecular machines. Reports on Progress in Physics 75, 126001 (2012). 15. Douarche, F., Ciliberto, S., Petrosyan, A. & Rabbiosi, I. An experimental test of the Jarzynski equality in a mechanical experiment. EPL (Europhysics Letters) 70, 593 (2005). Acknowledgements The authors gratefully acknowledge useful discussion with L. Gammaitoni. The authors acknowledge financial support from the European Commission (FPVII, Grant agreement no: 318287, LANDAUER and Grant agreement no: 611004, ICT- En- ergy). Author contributions statement M.L.S. and I.N. designed the experiment, performed the measurements, analyzed the measured data and contributed to the writing of the manuscript. Data availability statement The data that support the findings of this study are available from the corresponding author upon request. Additional information Competing financial interests The authors declare no competing financial interests. 5/8 Figure 1. Schematic of the whole system and measurement setup. Lateral view of the whole system and measurement setup. Two magnets with opposite magnetic orientations are used to induce bistability in the system. Two electrodes are used to apply electrostatic forces on the mechanical structure: VL and VR to force the cantilever to bend to the left (negative x) and to the right (positive x) respectively. The magnetic interaction can be engineered by changing geometric parameters such as d and D x. (b) Color-map of the reconstructed potential energy as function of the distance between the magnets, d. The distance is expressed in arbitrary units proportional to the voltage applied to the piezoelectric stage. Decreasing the distance between the magnets the potential energy softens and eventually two stable states appear. (c) Dependence of the effective temperature, Teff, with the root mean square of the white Gaussian voltage applied to the piezoelectric shaker. The red dot represent the condition accounted for the experimental data presented. 6/8 Figure 2. Reset protocol. (a) Protocol used to perform the reset operation. In order to account for all the possible transitions we considered the reset to 1 (first two columns) and reset to 0 (last two columnss) starting from both 0 and 1 states. The first row depict the protocol used for removing the barrier. Once the barrier is removed a lateral force is applied (second row). The resulting displacement of the cantilever tip is represented in the third row. Once all the forces are removed and the barrier is restored the cantilever tip encodes the final state. (b) Schematic time evolution of the potential energy and state of the system for the case presented in the first column of panel (a). The operation starts from a double well potential and in the first step of the protocol the barrier is removed. During the next step the potential is tilted and the barrier is restored to its initial value. Finally, the lateral force is removed recovering the initial condition where the barrier between wells is at its maximum and the electrostatic forces are equal to zero. 7/8 Figure 3. Produced heat and probability of success for the reset operation. (a) Average heat produced during the reset operation as function of the lateral alignment D x. For D x < 0 the counter magnet is moved to the right and the 0 state (x < 0) is favorable. Accordingly, for D x > 0 the 1 state is more favorable. Introducing an asymmetry on the potential Q decreases, which is accounted to the probability of success, Ps, that tends to decrease (Ps is encoded in the color map). (b) Success rate of the reset operation as function of lateral alignment. Solid violet circles represent the overall success rate while black and red symbols account for the success rate resetting to 0 and 1 state respectively. The maximum overall success rate is present when the system is almost symmetric, D x ≈ 0. (c) Relation between success rate and heat dissipated. Red circles correspond to the resetting to 1 case while black ones correspond to the resetting to 0. (d) Dependence of Q with the protocol time duration, t p. As t p is increased the effects of frictional phenomena becomes negligible and the produced heat should approach the thermodynamic limit. However, for large t p the reset operation fails giving a wrong logic output. In these cases, where the error probability is high, the produced heat is clearly below the Landauer limit. Inset shows the obtained relation between error probability (1-Ps) and produced heat. The data are compatible with the minimum energy required for a given error probability as predicted by Eq. 1, represented by dashed line. 8/8
1109.0619
1
1109
2011-09-03T13:19:56
Magnetic Fields Effects on the Electronic Conduction Properties of Molecular Ring Structures
[ "cond-mat.mes-hall" ]
"While mesoscopic conducting loops are sensitive to external magnetic fields, as seen by observation(...TRUNCATED)
cond-mat.mes-hall
cond-mat
"Magnetic Fields Effects on the Electronic Conduction Properties of \nMolecular Ring Structures \n \(...TRUNCATED)
1601.03861
1
1601
2016-01-15T10:13:31
Decoherence and Decay of Two-level Systems due to Non-equilibrium Quasiparticles
[ "cond-mat.mes-hall", "cond-mat.supr-con" ]
"It is frequently observed that even at very low temperatures the number of quasiparticles in superc(...TRUNCATED)
cond-mat.mes-hall
cond-mat
"Decoherence and Decay of Two-level Systems due to Non-equilibrium Quasiparticles\n\n1Institut fur T(...TRUNCATED)
1205.2029
1
1205
2012-05-09T16:40:11
Mode bifurcation on the rythmic motion of a micro-droplet under stationary DC electric field
[ "cond-mat.mes-hall", "physics.bio-ph" ]
"Accompanied by the development of microfabrication techniques, such as MEMS and micro-TAS, there ha(...TRUNCATED)
cond-mat.mes-hall
cond-mat
"Mode bifurcation on the rythmic motion of a micro-droplet under stationary DC\n\nelectric field\n\n(...TRUNCATED)
1811.03229
6
1811
2019-07-02T08:23:13
Temperature dependence of side-jump spin Hall conductivity
[ "cond-mat.mes-hall" ]
"In the conventional paradigm of the spin Hall effect, the side-jump conductivity due to electron-ph(...TRUNCATED)
cond-mat.mes-hall
cond-mat
"Temperature dependence of side-jump spin Hall conductivity\n\nCong Xiao,1, ∗ Yi Liu,2 Zhe Yuan,2,(...TRUNCATED)
1106.3697
1
1106
2011-06-19T00:12:16
"Electronic charge and spin density distribution in a quantum ring with spin-orbit and Coulomb inter(...TRUNCATED)
[ "cond-mat.mes-hall" ]
"Charge and spin density distributions are studied within a nano-ring structure endowed with Rashba (...TRUNCATED)
cond-mat.mes-hall
cond-mat
"\nElectronic charge and spin density distribution in a quantum ring\n\nwith spin-orbit and Coulomb (...TRUNCATED)
1601.07524
2
1601
2016-06-14T11:58:55
Surface plasmon polaritons in topological Weyl semimetals
[ "cond-mat.mes-hall", "cond-mat.str-el" ]
"We consider theoretically surface plasmon polaritons in Weyl semimetals. These materials contain pa(...TRUNCATED)
cond-mat.mes-hall
cond-mat
"a\n\nSurface plasmon polaritons in topological Weyl semimetals\n\nJohannes Hofmann1, 2 and Sankar D(...TRUNCATED)

arXiv Classifier Data

Usage:

from datasets import load_dataset, DownloadMode
# download from HuggingFace
dataset = load_dataset('mlcore/arxiv-classifier', name=<CONFIG NAME>)
# load from G2
dataset = load_dataset('/share/nikola/arxiv_classifier/data/arxiv-classifier', name=<CONFIG NAME>)

To force the dataset to be re-generated:

dataset = load_dataset('/share/nikola/arxiv_classifier/data/arxiv-classifier', name=<CONFIG NAME>, download_mode=DownloadMode.FORCE_REDOWNLOAD)

See: https://huggingface.co/docs/datasets/v2.20.0/en/package_reference/builder_classes#datasets.DownloadMode

Standardized terminology:

  • Field: Bio/cs/physics
  • Subfield: Subcategories within each
    • Primary subfield (bio., cs.LG): Given primary subfield, you can infer the field
    • Secondary subfields: Includes primary subfield, but also includes any subfields that were tagged in the paper (1-5)

Old terminology to standardized terminology translation:

  • Prime category = Primary subfield
  • Abstract category = secondary subfield
  • Major category = field

Original data: https://www.dropbox.com/scl/fo/wwu0ifghw4sco09g67frb/h?rlkey=6ddg3yab9la3zeddvmnsfktxq&e=1&dl=0

Minor (default): Dataset of papers between 2010 and 2020 (with some pre-2010 papers) with balanced primary subfields.

Major: Dataset of papers between 2010 and 2020 (with some pre-2010 papers) with unbalanced primary subfields to better represent the true distribution of primary categories, which is dominated by a few subfields. Note that the distribution of major subfields is still truncated.

All 2023: All papers published in 2023.

  • Train: papers with date between January and June (inclusive)
  • Test: papers with date between July and December (inclusive)

Primary subfield distribution

Secondary subfield distribution

To generate subfield distribution plots:

python plot_subfield_distributions.py --output_path <path_to_save_plots>

Set up dependencies

Get subfields to ignore and subfield aliases:

# in your conda env
git clone https://github.com/ag2435/arxiv-classifier-next
cd arxiv-classifier-next
conda develop .

Preprocessing

Transform raw data into JSON format:

# major/minor cats data
python preprocess_major_minor.py -d <DATASET NAME> -s <SPLIT> -op <PATH TO SAVE PREPROCESSED DATA>
# all 2023 corpus (full text)
python preprocess_all2023_v2.py

Additional data checks:

python test_paper_id.py

:white_check_mark: Checked that there is no data leakage between train and test splits for each dataset config

:white_check_mark: Checked validity of arXiv identifiers for each paper

Downloads last month
337
Edit dataset card