X-ray microfocussing combined with microfluidics for on-chip X-ray scattering measurements

Abstract

This work describes the fabrication of thin microfluidic devices in Kapton (polyimide). These chips are well-suited to perform X-ray scattering experiments using intense microfocussed beams, as Kapton is both relatively resistant to the high intensities generated by a synchrotron, and almost transparent to X-rays. We show networks of microchannels obtained using laser ablation of Kapton films, and we also present a simple way to perform fusion bonding between two Kapton films. The possibilities offered using such devices are illustrated with X-ray scattering experiments. These experiments demonstrate that structural measurements in the 1 Å–20 nm range can be obtained with spatial resolutions of a few microns in a microchannel.

---

Introduction

Microfluidics is now a well-established tool for studying chemical reactions, performing biological assays, and even answering fundamental questions of physics.^1–5 Microfluidic technologies also require characterization tools compatible with the miniaturization of the devices. Obviously, optical microscopy techniques are well-suited for detecting small amounts of samples in microdevices: fluorescence, IR, and Raman microscopy^6–9 are, for instance, routinely used to image chemical reactions on-chip.

X-ray techniques offer very powerful measurements for elemental (X-ray fluorescence) and structural analysis (X-ray diffraction). In particular, X-ray scattering is a widely used technique to determine the structure of crystallized proteins. Recently, several groups have shown that microfluidics is a very promising tool to perform high throughput screening of protein crystallization conditions.^10,11 For these studies the X-ray diffraction sample analysis was performed off-chip, since the device design would give rise to excessive X-ray attenuation or scattering.¹² Moreover, chips fabricated using materials such as polydimethylsiloxane (PDMS) display rather short life times if intense X-ray beams (such as those generated by a synchrotron) are used. Therefore, performing X-ray diffraction measurements directly on a microfluidic chip is a real challenge.

Among the few papers reporting works combining microfluidics and X-ray diffraction, one can cite Pollack et al. who successfully measured the kinetics of the folding of a specific protein using small angle X-ray scattering (SAXS)^13,14 (see also the related reference on RNA folding¹⁵). In this work, the authors used a pink X-ray beam focussed onto a microfluidic mixer in order to record scattering patterns in the 0.04–0.5 Å⁻¹ range. The size of the focussed beam onto the chip is 10 × 40 µm². To avoid background scattering from the microdevice, they sealed their microfluidic mixer with silicon nitride membranes as windows for the X-ray beam.¹⁶

Greaves and Manz recently reviewed all the problems related to X-ray analysis on microfluidic devices.¹² They also showed successful X-ray fluorescence measurements using on-chip X-ray generation. For X-ray scattering measurements, they reported diffraction data of weak intensities using a chip in polycarbonate of bisphenol, and using a 1 mm wide X-ray beam. The authors also indicated some recommendations in how to perform on-chip X-ray scattering measurements: (i) low absorption materials, (ii) thin chip thicknesses, (iii) high energies for small angles. In the present work, we have fabricated new microdevices in Kapton using laser ablation.¹⁷ Kapton is a commercial polyimide polymer that has a relatively good resistance to the high intensities of X-ray beams generated by synchrotrons, and low absorption for X-rays in the keV energy range. The Kapton films are yellow–orange colored, indicating absorption in a certain range of visible wavelengths, but are sufficiently transparent to visualize fluid flows inside microchannels. Moreover, Kapton is highly resistant to a wide variety of solvents (acetone, benzene, toluene…) which makes it attractive for chemical applications.¹⁷ We show that (i) Kapton is suitable to fabricate complex microfluidic geometries, (ii) these chips resist intense microfocussed X-ray beams and permit X-ray diffraction measurements, (iii) we also show X-ray scattering experiments performed at the European Synchrotron Radiation Facility (ESRF) with a microfocussed beam at an energy of 14 keV. We have demonstrated during these experiments that both large and small angle X-ray scattering can be performed for structural analysis, in the 0.03 < q < 6 Å⁻¹ range, with a spatial resolution of a few microns. The upper limit is determined by the angular acceptance of the two dimensional detector. To illustrate these points, we present diffraction patterns of a complex fluid (wormlike micelles) undergoing a shear-induced transition in a microchannel, as the flow rate is changed. These results also open the possibility to study the rheology of complex fluids with a spectacular spatial resolution.

Microfabrication of Kapton chips

Polyimide-based microfluidic devices have been recently microfabricated using transfer and lamination techniques by Metz et al.¹⁸ Polyimides are commonly used in microelectronics due to their thermal, electrical and mechanical properties (dielectric and passivation layers for semiconductors, protective and insulating films…). In the present work, microfluidic chips in commercial polyimide films (Kapton KJ, Dupont¹⁷) were fabricated using laser ablation techniques.

Laser ablation of polymers was first reported in 1982 using an excimer laser.^19,20 For highly absorbing polymers, a UV laser beam induces a photochemical decomposition known as photoablation. With increasing absorbance of the polymer, the development of thermal artifacts should diminish. Another advantage of laser ablation is that ejected material carries out a significant proportion of the initial energy thus limiting any thermal damage inside the target. Nowadays, high quality laser ablation of polyimide is mainly achieved using UV lasers operating at wavelengths of 355 or 248 nm. For instance, laser ablation at 355 nm is one of the key technologies used for via drilling of polyimide dielectric layers in printed-circuit-boards, in multi-chip-module components or in micro-electro-mechanical systems.^21–23 This technology has been used for several years for microdrilling of polyimide inkjet printer nozzle arrays.^24,25 Nevertheless, Yung et al. have demonstrated that melting occurs during polyimide ablation at 355 nm, which indicates that a photothermal mechanism contributes to the ablation process.²⁶ In addition, previous XPS analysis of polyimide has shown that the C content increases significantly during UV irradiation (355 or 248 nm), while the O content and the N content decreases, due to surface carbonization.²² Recent work indicates that the use of femtosecond radiation enables both thermal and mechanical side-effects in solid targets to be minimized.^27–31 Adhi et al. explain that in ultra-short pulse laser ablation, the incident energy is deposited in the material in a timeframe which is shorter than the relaxation time of the material. Furthermore, due to high laser intensity and multiphoton absorption, femtosecond radiation leads to a higher probability of processing materials with sub-band gap photon energies compared to conventional IR or UV lasers.³¹ Thus, we expect a high machining quality of polyimide using a femtosecond laser beam.

Kapton films of 10 × 10 cm² (Dupont) of thickness 75 µm were machined by laser ablation. Different complex patterns of microchannels have been manufactured. We used a diode pumped femtosecond laser (Amplitude Systemes, s-Pulse model) for polyimide engraving. A pulse duration for 400 fs at wavelength 1030 nm was used giving a maximum pulse energy of 75 µJ at 10 kHz. The laser is coupled with a beam shaping and delivering setup which includes a beam expander, a beam deflector device (scanner) and a 100 mm f-theta lens. The sample holder is mounted on a high precision XYZ motorized stage assembly. Machining occurs in air at ambient temperature. This technique requires no surface preparation before laser machining. The width and depth of the engraved channel are measured using a laser scanning confocal microscope. Complex patterns could be performed by combining both laser hatch scanning and motorized stage translations. The motorized stage translation is used along the length axis of the channel during which hatch scanning is used across the width axis of the channel (perpendicular to the former). The channel cross sectional profile, for example square convex or concave, is achieved by adjusting scan or translation speed. Ablated thickness is about 6 µm per step. Thus, a 25 µm deep channel requires four steps. The effective engraving speed for a 100 µm wide and 30 µm deep channel is about 50 µm s⁻¹. The high repetition rate of the laser (10 kHz) requires a high accuracy on synchronization between the scanning and laser trigger. Any uncontrolled pulsed release on the target may induce either overmachining or undesirable damage to the target. The pulse energy is set at a low level (10 µJ) in order to reduce both thermal and mechanical artifacts during laser machining. Use of a low pulse energy gives low etch rates which enables accurate control of the channel depth. For the same reason, we used a spot diameter (15 µm) smaller than the channel width (>100 µm). Furthermore, this allows a better control of the sidewall angle (aspect ratio) in the channel.

We fabricated several complex patterns for which the roughness of the obtained microchannels did not exceed 1 µm (see Fig. 1 and 2). The main advantages of femtosecond laser ablation for this application are versatility and the possibility to micro machine 3D structures. Various topographies (depth, width) could easily be achieved by varying translation speed and number of steps. Because of low machining speeds and wide patterns, the main drawback is a long process time.

Fig. 1 (a) Example of a microdevice obtained by laser ablation in Kapton. (b) The network of microchannels of this device allows one to generate droplets using the flow focussing geometry described in ref. 32. The downstream windings allow a rapid mixing of the reactants contained in the droplet to perform chemical analysis.^33,34 The width of the channel is 300 µm. Inset: details of the pinch-off for droplet formation. The size of the small channel is 80 µm wide and 200 µm long. (c) Typical depth profile of a microchannel, and (d) its roughness measured by a profilometer. The roughness does not exceed 2 µm.

Fig. 2 (a), (b) Examples of SEM pictures of microchannels, the width of the channels are 300 µm. (c) Pressure drop ΔP in a microchannel vs. flow rate Q. The microchannel is 300 µm wide, 25 µm high and 12.4 cm long. The continuous line is the best linear fit of the data ΔP = 0.0476Q + 0.0389 (R² = 0.9991). The error bars are smaller than the symbol size.

After the fabrication of microchannel networks, one has to seal the device using another Kapton film. To perform such bonding, we used Kapton KJ, a specific polyimide film which presents adhesive properties when the temperature is increased above 300 °C.¹⁷ Different protocols were tested to obtain suitable bonding. We retained the following process: the two films are cleaned using isopropanol and dried with nitrogen, then sandwiched between two films of Teflon (thickness 100 µm, Dupont), and finally between two glass slides. The Teflon films do not adhere to the Kapton films when temperature is increased. We then apply a load using steel weights (≈2.5 kg) giving an approximate pressure of 10 kPa on a 5 × 5 cm² chip. Temperature ramps are controlled using a standard oven (Nabertherm). Temperature is first increased from room temperature to 300 °C in 1 h, and is then maintained at 300 °C for 20 min. Finally, the system is allowed to relax to room temperature. We encountered specific problems when channels with low aspect ratio (e.g. 75 × 1000 µm²) have been tested. In these cases, deflections of the films have been observed. We believe that commercial bonding tools, such as those used in standard cleanrooms, may overcome these difficulties, since pressure and temperature can be controlled more precisely. However, good bonding between the two films can be obtained with a simple fusion bonding protocol in a standard oven.

Finally, we used standard connections (Nanoports, Upchurch Scientific) for fluid handling (see Fig. 1(a)). The resulting chips consist of thin flexible films of Kapton, with a thickness of approximately 150 µm. Even if the optical properties of Kapton films are not optimal, the systems are sufficiently transparent to monitor the presence of fluid flows inside the microchannels using visible light techniques. To use such devices for standard microfluidic experiments, one has to know if the bonding can withstand high pressures and if the microchannels deform under pressure. Fig. 2(c) displays a ΔPvs. Q curve, where Q is the imposed flow rate, and ΔP the pressure drop measured between the inlet and the outlet of a microdevice. To perform these measurements, we used a Kapton microdevice (300 µm wide and 25 µm deep, ≈13 cm long) with two inlets and one outlet, forming a so-called T-junction. We inject in the first inlet water at various flow rates using a syringe pump (Braintree BS-8000). The outlet of the device is at atmospheric pressure. A pressure sensor (Sensortechnics, ASDX100D44D) connected to the second inlet allows us to measure the pressure drop between the T-junction and the end of the microchannel (length 12.4 cm). Fig. 2(c) displays the data obtained. The linear relation between ΔP and Q indicates that the hydrodynamic resistance of the chip is constant over the applied pressure, and therefore that the microchannel does not deform significantly. This result is consistent with similar measurements performed by Metz et al. on polyimide-based microdevices manufactured by transfer and lamination techniques.¹⁸

X-ray microfocussing scattering experiments

A series of experiments was performed at the ESRF, at the microfocussing endstation ID18F³⁵ (see Fig. 3).

Fig. 3 Schematic setup of ID18F used for X-ray scattering measurements. The microdevice is mounted in the focal spot generated by the Compound Refractive Lens (CRL) and can be scanned in two axes orthogonally to the direction of the beam (see www.esrf.fr\UsersAndScience\Experiments\Imaging\ID18F\ for more details).

The experimental station uses aluminium compound refractive lenses³⁶ to produce a monochromatic X-ray microprobe beam of dimensions as low as 5 × 1 µm² (horizontal × vertical) at energies which can be varied between 10 and 30 keV. The flux in the monochromatic (ΔE/E ≈ 10⁻⁴) microbeam is typically of the order of 10² photons s⁻¹. The sample (in this case the microfluidic cell) can be scanned orthogonally through the focussed X-ray beam with resolutions of 0.1 µm vertically and 1 µm horizontally. This allows the cell to be accurately positioned relative to the X-ray beam and permits the study of X-ray interactions at different positions in the device. Diffraction patterns are collected using a cooled CCD detector (MAR Research, 16 bit, 2048 × 2048 pixels, pixel size 64.276 µm). A small beamstop absorbs the transmitted beam and prevents the saturation of the detector. Several experiments were performed using different Kapton chips. All the microdevices resisted prolonged exposure (>1 h) to the intense microfocussed X-ray beam with no observable degradation (no leaks, no change of refractive index, no burn stains…).

Different experimental configurations were tested, and demonstrated that both small and large angle scattering data can be obtained simply by varying the distance between the microdevice and the detector. Therefore, one can readily measure diffraction patterns in the range 0.03 < q < 6 Å⁻¹, and thus access sizes ranging from crystal structures (1 Å) to supramolecular assemblies (≈20 nm). In the following paragraphs, we illustrate these experiments with a specific system: wormlike micelles under shear. The energy of the X-ray beam was set to 14 keV (wavelength 0.88 Å). A beam size of 10 × 6 µm² (horizontal × vertical) with a depth of field of several centimeters was used. In this geometry, the accessible wave vectors range between 0.03 and 3.5 Å⁻¹.

Wormlike micelles consist of very long cylindrical aggregates of self-assembled surfactant molecules that mimic polymer solutions, but can dynamically break and recombine.^37,38 The linear and non-linear rheology of these systems have been widely studied (see for instance ref. 39, 40 and references therein). In the non-linear regime and using classical rheological experiments, one generally observes the behaviour schematically displayed in Fig. 4.

Fig. 4 (a) Schematic flow curve observed for wormlike micelles in a classical rheological experiment. The first branch () corresponds to isotropic highly viscous micelles, whereas the second branch () corresponds to a homogeneous low viscous state of aligned micelles. Along the plateau, one observes generally the coexistence between the two states and therefore a banded velocity profile as displayed in (b). The shear rate indicated by the rheometer is in this case , where e is the size of the gap of the rheological cell. Along the plateau, the proportion 1 − α of the highly sheared band progressively invades the gap, as the total shear rate is increased.

(i) At low rates of shear, the wormlike micelles behave as a weak shear thinning isotropic solution. (ii) Above a critical shear rate and at a given shear stress σ*, one observes a plateau in the flow curve (shear stress vs. shear rate). This plateau is associated with the nucleation and growth of a low viscous and highly sheared band, suggesting a strong alignment of the wormlike micelles with the flow. The flow profile displays therefore two differently sheared bands (see Fig. 4(b)). As the shear rate is increased, this new organization fills the flow at a constant stress up to a second critical shear rate . (iii) Above , one recovers a second branch of increasing stress, and the flow is homogeneous again. This behaviour has been observed in many different systems, but several fundamental questions remain, and a profound understanding of these phenomena is still lacking.^38–45 We believe that microfluidics, providing full control of hydrodynamics at small length scales, may answer some of these questions, and offer new perspectives for the rheology of complex fluids. In the present work, we have studied a specific wormlike micellar system: cetyl trimethyl ammonium bromide (CTAB) in D₂O at a temperature T = 40 °C and at a surfactant concentration of 20 wt%. This system has been studied by several groups,^40,42 and different techniques (neutron scattering, nuclear magnetic resonance) have shown that the shear-induced structure is a nematic phase.

The wormlike micelles are injected at a controlled flow rate Q using a syringe pump (Harvard apparatus) into a Kapton chip. The device consists of a straight channel of width 300 µm and depth 40 µm. The temperature of the system is controlled and regulated to 40 °C using several adhesive heaters (Minco) to control the temperature of the chip, of the tubing, and of the syringe. All the temperatures are controlled with feedback temperature measurements with Pt100 sensors (Minco). The X-ray beam was focussed in the middle of the microchannel at a distance of ≈1 cm from the inlet.

Fig. 5 shows different diffraction patterns measured on this wormlike micellar system at several flow rates. Each diffraction pattern was acquired using exposure times of 100 s, and we have subtracted from each measurement, a pattern corresponding to pure water flowing in the same region of the microchannel. To check the stability of the measurements, and be sure that the system has reached its equilibrium, several patterns are recorded for each applied flow rate. Each flow rate has been maintained for about 20 min. As shown in Fig. 5, the structure of the complex fluid evolves as the flow rate is changed. At a very low flow rate (Q = 1 µL h⁻¹), the diffraction patterns are isotropic rings. The structure factor exhibits a maximum at q = 0.08 Å⁻¹. This peak is associated with spatial correlations between the wormlike micelles at a distance ≈7.8 nm. At higher flow rates (Q > 15 μL h⁻¹), fuzzy peaks appear on the ring, indicating some alignment of the wormlike micelles with the direction of the flow. Note that a small angle can be observed between the direction of this alignment and the direction of the flow. We cannot ensure this is not due to an artifact (imperfect alignment of the microchannel with the vertical direction) or if it corresponds to a physical effect. The contrast of these peaks on the ring increases as the flow rate is increased and for Q > 60 μL h⁻¹, one does not observe the isotropic ring anymore. All the micelles are aligned with the flow. This pattern is a strong indication of a nematic phase.⁴²

Fig. 5 Diffraction patterns measured at various flow rates in the microchannel. (a) Q = 125, (b) 60, (c) 15, and (d) 1 µL h⁻¹. The peaks are located at q = 0.08 Å⁻¹. The direction of the arrows indicates the flow.

The goal of the present paper is not to analyse in depth these data, but rather to demonstrate the possibilities offered by such techniques. A more detailed study concerning the complex rheological behaviour of such fluids in microchannels (flow profiles, X-ray scattering…) is under way. Using the present approach, the structure of complex fluids can be investigated with a high spatial resolution (a few microns) in microfluidic geometries (dimensions ≈ 100 μm), and therefore mapping of the structure of the fluid can be performed in the flow.

Conclusion

In the present work, we used laser ablation techniques to manufacture new microfluidic devices based on Kapton. Networks of microchannels can be obtained with a spatial resolution of 10 μm, and a simple fusion bonding protocol using specific Kapton films allows one to seal the microdevice. These chips are well-suited for X-ray scattering experiments since Kapton is a weak scatterer of X-rays, and can resist the high flux densities of microfocussed beams generated by a synchrotron. Using such chips, we have demonstrated that both large and small angle X-ray scattering can be measured. To illustrate this last point, we measured the structure of a complex fluid undergoing a shear-induced transition in a microchannel as the flow rate is changed.

The main result of the paper is that structural measurements in the range 1 Å–20 nm can be obtained with a spatial resolution of a few microns within a microfluidic device. This opens several possibilities for studies of protein crystallization, chemical analysis, and, as demonstrated in this work, investigation of shear on soft condensed matter.

Acknowledgements

The authors are deeply grateful to the members of the LOF, Rhodia-CNRS laboratory in Pessac, for many discussions, and to the CREMEM (Université Bordeaux 1) for the SEM imaging.

References

1. A. D. Stroock and G. M. Whitesides, Acc. Chem. Res., 2003, 36, 597.
2. R. F. Ismagilov, Angew. Chem., Int. Ed., 2003, 42, 4130.
3. H. A. Stone, A. D. Stroock and A. Ajdari, Annu. Rev. Fluid Mech., 2004, 36, 381.
4. T. Vilkner, D. Janasek and A. Manz, Anal. Chem., 2004, 76, 3373.
5. T. M. Squires and S. R. Quake, Rev. Mod. Phys., 2005, 77, 977.
6. J.-C. Roulet, R. Volkel, H. P. Herzig, E. Verpoorte, N. F. de Rooij and R. Dandliker, Anal. Chem., 2002, 74, 3400.
7. D. Pan, Z. Ganim, J. E. Kim, M. A. Verhoeven, J. Lugtenburg and R. Mathies, J. Am. Chem. Soc., 2002, 124, 4857.
8. P. Fletcher, S. Haswell and X. Zhang, Electrophoresis, 2003, 24, 3239.
9. J.-B. Salmon, A. Ajdari, P. Tabeling, L. Servant, D. Talaga and M. Joanicot, Appl. Phys. Lett., 2005, 86, 094106.
10. C. L. Hansen, E. Skordalakes, J. M. Berger and S. R. Quake, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 16531.
11. B. Zheng, J. D. Tice, L. S. Roach and R. F. Ismagilov, Angew. Chem., Int. Ed., 2004, 43, 2508.
12. E. D. Greaves and A. Manz, Lab Chip, 2005, 5, 382.
13. L. Pollack, M. W. Tate, N. C. Darton, J. B. Knight, S. M. Gruner, W. A. Eaton and R. H. Austin, Proc. Natl. Acad. Sci. U. S. A., 1999, 96, 10115.
14. L. Pollack, M. W. Tate, A. C. Finnefrock, C. Kalidas, S. Trotter, N. C. Darnton, L. Lurio, R. H. Austin, C. A. Batt and S. M. Gruner, Phys. Rev. Lett., 2001, 86, 4962.
15. R. Russell, I. S. Millett, M. W. Tate, L. W. Kwok, B. Nakatani, S. M. Gruner, S. G. J. Mochrie, V. Pande, S. Doniach and D. Herschlag, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 4266.
16. L. Pollack, 2003–2004 Cornell NanoScale Facility Research Accomplishments.
17. Dupont, http://www.dupont.com/kapton/..
18. S. Metz, R. Holzer and P. Renaud, Lab Chip, 2001, 1, 29.
19. R. Srinivasan and V. Mayne-Banton, Appl. Phys. Lett., 1982, 41, 576.
20. Y. Kawamura, K. Toyoda and S. Namba, Appl. Phys. Lett., 1982, 40, 374.
21. A. Beuhler, A. Tungare and J. Savic, Circuit World, 1998, 24, 36.
22. K. C. Yung and D. W. Zeng, Surf. Coat. Technol., 2001, 145, 186.
23. H. Zheng, E. Gan and G. C. Lim, Opt. Lasers Eng., 2001, 36, 355.
24. H. J. Booth, Thin Solid Films, 2004, 453–454.
25. G. Riccardi, M. Cantello, F. Mariotti and P. Giacosa, CIRP Ann., 1998, 47, 145.
26. W. K. C. Yung, J. S. Liu, H. C. Man and T. M. Yue, J. Mater. Process. Technol., 2000, 101, 306.
27. S. Preuss, A. Demchuk and M. Stuke, Appl. Phys. A, 1995, 61, 33.
28. B. C. Stuart, M. D. Feit, S. Herman, A. M. Rubenchik, B. W. Shore and M. D. Perry, J. Opt. Soc. Am. B, 1996, 13, 459.
29. C. Momma, B. N. Chichkov, S. Nolte, F. V. Alvensleben, A. Tünnermann, H. Welling and B. Wellegehausen, Opt. Commun., 1996, 129, 134.
30. F. Korte, S. Nolte, B. N. Chichkov, T. Bauer, G. Kamlage, T. Wagner, C. Fallnich and H. Welling, Appl. Phys. A, 1999, 69, S7.
31. K. P. Adhi, R. L. Owings, T. A. Railbar, W. D. Brown and A. P. Malshe, Appl. Surf. Sci., 2004, 225, 324.
32. S. L. Anna, N. Bontoux and H. A. Stone, Appl. Phys. Lett., 2003, 82, 364.
33. H. Song and R. F. Ismagilov, J. Am. Chem. Soc., 2003, 125, 14613.
34. H. Song, J. D. Tice and R. F. Ismagilov, Angew. Chem., Int. Ed., 2003, 42, 767.
35. A. Somogyi, M. Drakopoulos, L. Vincze, B. Vekemans, C. Camerani, K. Janssens, A. Snigirev and F. Adams, X-Ray Spectrom., 2001, 30, 242.
36. B. Lengeler, C. Schroer, J. Tümmler, B. Benner, M. Richwin, A. Snigirev, I. Snigireva and M. Drakopoulos, J. Synchrotron Radiat., 1999, 6, 1153.
37. M. E. Cates, Macromolecules, 1987, 20, 2289.
38. N. A. Spenley, M. E. Cates and T. C. B. McLeish, Phys. Rev. Lett., 1993, 71, 939.
39. J.-F. Berret, G. Porte and J.-P. Decruppe, Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top., 1997, 55, 1668.
40. E. Fischer and P. T. Callaghan, Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top., 2001, 64, 011501.
41. P. D. Olmsted and P. M. Goldbart, Phys. Rev. A: At., Mol., Opt. Phys., 1990, 41, 4578.
42. E. Cappelaere, J.-F. Berret, J.-P. Decruppe, R. Cressely and P. Lindner, Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top., 1997, 56, 1869.
43. L. Bécu, S. Manneville and A. Colin, Phys. Rev. Lett., 2004, 93, 018301.
44. S. Lerouge, J.-P. Decruppe and C. Humbert, Phys. Rev. Lett., 1998, 81, 5457.
45. J.-B. Salmon, A. Colin, S. Manneville and F. Molino, Phys. Rev. Lett., 2003, 90, 228303.

---