news & views magnitude higher conversion efficiency than a flat unstructured target of the same material. At 1 J laser pulse energy, roughly 1015 X-ray photons were generated per pulse, which is comparable to the output of a standard X-ray tube in 1 s, now compressed into less than 10 ps — bright enough to take a snapshot image of an exploding object or, as shown by Hollinger et al. for demonstration purposes, of a wasp’s knee. To put this number into another perspective, 1015 photons is about 100 times more than the output of a modern synchrotron radiation beamline in 1 s, and also 100 times more than the number of photons in an X-ray free-electron laser pulse. This is with the decisive difference that the photons produced by the plasma are emitted into a large solid angle, while the emission of a synchrotron or free-electron laser is highly directional such that it can be focused onto a tiny spot of less than 1 μm diameter. The maximum fluence recorded by the photodiodes in the experiment by Hollinger et al. was thus up to 15 μJ cm–2, which is roughly 1011 photons per cm2, as compared with a fluence of 1 mJ μm–2, which is ten orders of magnitude higher, reached in the focus of a free-electron laser beam. The X-rays emitted from the plasma could, potentially, also be collimated and focused, if required
by the experiment, thus gaining back a few orders of magnitude. Clearly, as demonstrated in an impressive way by Hollinger et al., the schemes for generating X-rays in plasmas created from nanostructured surfaces have significantly improved in recent years, opening up a great potential for building compact sources of high-brightness picosecond X-ray pulses. While by no means comparable to acceleratorbased sources in terms of brightness, these relatively compact and affordable table-top sources can be used in a variety of imaging applications, such as flash radiography, as well as in a much larger context of time-resolved diffraction and spectroscopy, for example, to investigate picosecond dynamics in liquid and solid samples. While further modelling and tailoring of the plasma properties may be able to squeeze out a few more per cent in conversion efficiency, a more promising next step towards achieving even higher X-ray fluences would be to work out ways to make the X-ray emission more directional such that less photons are ‘wasted’ because they do not reach the sample. Another intriguing possibility might be to use more than one laser pulse to modify the temporal structure
of the X-ray emission, for example, to produce shorter X-ray pulses, even if it comes at the cost of sacrificing some of the hard-fought for photons. Either way, whether it is for further source development or for concrete imaging or spectroscopy applications, the latest jump in conversion efficiency has made picosecond X-ray sources based on plasmas generated from nanostructured surfaces a much more versatile tool that is bound to see more widespread uses in the near future. ❐ Daniel Rolles
J. R. Macdonald Laboratory, Department of Physics, Kansas State University, Manhattan, KS, USA. e-mail:
[email protected]
Published online: 26 January 2018 https://doi.org/10.1038/s41566-018-0092-9 References
1. McNeil, B. W. J. & Thompson, N. R. Nat. Photon. 4, 814–821 (2010). 2. Waldrop, M. Nature 505, 604–606 (2014). 3. Popmintchev, T., Chen, M.-C., Arpin, P., Murnane, M. M. & Kapteyn, H. C. Nat. Photon. 4, 822–832 (2010). 4. Teichmann, S. M., Silva, F., Cousin, S. L., Hemmers, M. & Biegert, J. Nat. Commun. 7, 11493 (2016). 5. Corkum, P. B. & Krausz, F. Nat. Phys. 3, 381–387 (2007). 6. Rudenko, A. et al. Nature 546, 129–132 (2017). 7. Chapman, H. N. et al. Nature 470, 73–77 (2011). 8. Hollinger, R. et al. Optica 4, 1344–1349 (2017). 9. Murnane, M. M. et al. Appl. Phys. Lett. 62, 1068–1070 (1993).
IMAGING
Optical nanoscopy turns coherent The ability to switch fluorophores on and off is key to performing super-resolution nanoscopy. To date, all switching schemes have been based on an incoherent response to the laser field. Now, a nanoscope that uses on–off coherent switching of quantum dots has been demonstrated.
Thomas A. Klar
I
n the 1990s, it was proposed1 and experimentally shown2 that the resolution of fluorescence far-field microscopy is not limited by diffraction. The basic idea behind this super-resolution capability is to shrink the focal point spread function (PSF) by switching off fluorophores in the outer rim of the focal area. This was first achieved by stimulated emission depletion (STED) and since then many other optically driven, spatial switching techniques have been established3. Alternatively, localization of single molecules4 together with stochastic temporal switching5 can also yield super-resolution through schemes referred to as photoactivated localization microscopy (PALM) and stochastic optical reconstruction microscopy (STORM). Both routes, based on either spatial or temporal switching, have transformed
optical microscopy into nanoscopy and the Nobel Prize in Chemistry was awarded for this achievement in 2014. To date, all published methods that switch a fluorophore into an off-state are effectively incoherent, which means that molecular states rapidly lose their phase memory with respect to the switching photons. Now, writing in Nature Photonics, Kaldewey and co-workers report the first demonstration of coherent on–off switching of fluorophores in an optical nanoscope6. Using a scheme called rapid adiabatic passage (RAP)7, they smoothly turn a quantum dot (QD) into an excited on-state8,9 with almost 100% fidelity and, in the outer rim of the PSF, turn it off with the same efficiency and elegancy. The result is the localization of the QD’s emission to a region as small as ~30 nm (λ/31). ‘Elegancy’ is
probably the right term to describe this level of control to excite and de-excite fluorescing species, compared with the incoherent approaches used previously. Some important details, usually unmentioned while explaining STED, highlight the unique aspects and importance of RAP. As traditional STED (Fig. 1a) is based on an incoherent approach, the best depopulation one can achieve is a 50:50 equal population between the two levels S1 and S0*. This is because, according to Einstein, the stimulated emission transition S1 → S0* and the (stimulated) absorption S0* → S1 are of equal probability. Degeneracies of the levels may lead to some other distribution, but nevertheless, a close to 100% depopulation of S1, as required for STED nanoscopy, is not reachable using just the two levels S1 and S0*. So, why does STED
Nature Photonics | VOL 12 | FEBRUARY 2018 | 59–67 | www.nature.com/naturephotonics
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
63
news & views a
b
Vib.
d
+
S0*
ISTED
S0
σ+
X
f σ+
Ifluo
+
Fluo.
RAP ex.
RAP depl.
RAP
X
Ifluo
e
fss
σ–
c
STED
S1
Fluo.
Ex.
STED
Vib.
IRAP, depl.
S0
Fig. 1 | Comparison of STED with RAP super-resolution nanoscopy. a, Energy levels and transitions involved in STED nanoscopy: excitation beam (Ex.; blue), vibronic non-radiative transitions (Vib.; dotted black), STED transition (STED; red) and fluorescence (Fluo.; orange). b, Intensity of the fluorescence (Ifluo) as a function of the strength of the STED beam (ISTED). c, Spatial profiles of the excitation beam (blue) and STED beam (red) involved to reduce the PSF of the fluorescence (orange). d, The three energy levels involved in RAP nanoscopy: the ground state S0, and an excitonic state Xσ–, which couples to a Xσ+ state by fine-structure splitting (fss). Up- and down-chirped laser pulses (RAP ex. and RAP depl., respectively) are used to coherently switch the populations of the states. The fluorescence (orange) is right circularly polarized. e, The right circularly polarized fluorescence intensity is suppressed rapidly with applied RAP depletion power IRAP, depl.. f, An up-chirped excitation pulse (left) populates the upper state with 100% efficiency via RAP. A doughnut-shaped down-chirped pulse (right) switches the excited state off. Both are left circularly polarized, while the detector registers only right circularly polarized fluorescence (orange), which originates from the shrunk, sub-diffraction area as indicated.
nanoscopy work at all? The answer is because the level S0* has a drain via a fast vibronic spontaneous transition S0* → S0. In essence, the red STED beam keeps the incoherent 50:50 population distribution between S1 and S0*, while S0* is continuously drained and hence S1 is emptied. As a side effect though, potentially harmful vibronic energy is dumped into the (mostly) organic molecules used in STED nanoscopy. The depopulation of the S1 level is a highly nonlinear function of the STED intensity as sketched in Fig. 1b. In contrast, RAP-based fluorescence suppression6 is fundamentally different to classical STED1,2. In RAP, a two-level system can be inverted from the ground state S0 (no exciton in a QD) to the upper state X with 100% efficiency (Fig. 1d). If excitation was incoherent, only 50% inversion could be reached, but RAP is coherent and therefore allows for 100% inversion, provided that the applied field strength is sufficiently strong7,9. Using the language of coherent quantum optics, Kaldewey et al. speak of pulse area instead of field strength when describing the excitation pulses, but the reader only familiar with common terms in microscopy should not be confused or worried. It suffices to simply think of the square root of applied intensity as a synonym for pulse area. More commonly known than RAP may be the 64
closely related phenomenon of coherent Rabi oscillations. In a two-level system, a spectrally sharp laser beam that exactly matches the energy gap between the two levels can induce coherent oscillations in the system between the upper and the lower levels. If the laser beam is stopped at exactly the right moment, when the system is in the upper level, the system is excited with 100% fidelity. In turn, if a system starts in the upper state and the Rabi oscillations are stopped in the lower state, the system is turned off with 100% fidelity. This capability makes it sound as if coherent Rabi oscillations would be the perfect switch for STED-inspired coherent nanoscopy and, in fact, this question was asked in the early days of STED. Unfortunately, driving Rabi oscillations with spectrally sharp laser pulses that perfectly match the energy difference in a two-level system turned out to be a Sisyphean task. As a way out of this dilemma, chirped ultrafast laser pulses (pulses that have an upwards or downwards spectral shift during their duration) have proven to be robust and fault-tolerant for coherent excitation7. They can drive exactly half a cycle of a Rabi oscillation, which means they switch a two-level system from the ground (lower) state to the on-state but do not bring it back down again. This is a
perfect on-switch. A second chirped laser pulse then induces exactly the opposite, namely driving the excited system by exactly half a Rabi oscillation down to the ground state, manifesting the perfect off-switch. To render RAP nanoscopy practical, Kaldewey et al. had to remove some further technical difficulties6. The first difficulty was that RAP nanoscopy only works at very low temperatures for the QD system not to lose coherence by exciton–phonon scattering. For this reason, the experiment was carried out at a temperature of 4 K and it was imperative for the RAP process to not introduce phonons. This was solved by choosing the right sign of the chirp. By using an up-chirped pulse for excitation and a down-chirped pulse for depletion, dumping of phonons is avoided10. The spectral chirp is indicated in Fig. 1d by the rainbow colour of the arrows and in Fig. 1f by the melange of red/blue colours for the excitation and the depletion PSFs. The second difficulty was separating the signal from the level-control pulses. In STED microscopy, the strong stimulating pulse and the stimulated photons (red arrow in Fig. 1a and red doughnut in Fig. 1c) are at a different wavelength from the fluorescence forming the image and thus can be removed by simply placing a band-pass filter in front of the detector, allowing only the orange fluorescence photons to pass. As RAP nanoscopy works essentially with only two energetic levels and thus one common wavelength, another trick needed to be found to distinguish between off-switching photons and fluorescence photons. The solution is to use a third level, energetically (almost) degenerate with the upper state populated by RAP. The trick is that there are two excitonic levels. The Xσ– state is only reachable from the ground state by transitions driven by left circularly polarized σ– photons. This state is coupled via fine-structure splitting to the state Xσ+, from where right circularly polarized fluorescence can occur (Fig. 1d). A circular polarizer in front of the detector will now distinguish between the intense RAP beam and the fluorescence from the shrunk effective PSF (orange dot in Fig. 1f). Apart from these differences, RAP and STED nanoscopy essentially work in the same way. The detectable fluorescence is suppressed approximately exponentially by the depleting beam (Fig. 1b,e) and the depleting beam is applied around the excitation beam, for instance in a doughnutlike fashion (Fig. 1c,f). The reported effective PSF for RAP in these initial experiments with QDs is 30 nm full-width at half-maximum (FWHM), an impressive value for a first report. Similar to the genesis of STED nanoscopy, a further decrease of FWHM can likely be expected for RAP nanoscopy,
Nature Photonics | VOL 12 | FEBRUARY 2018 | 59–67 | www.nature.com/naturephotonics
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
news & views possibly reaching the effective size of the exciton in a QD in the near future. As this size is typically in the range of several nanometres, the FWHM might soon be limited by the QD itself. Record measurements might therefore be expected not with QDs but instead with other two-level systems such as colour centres or single atoms. Eventually, one may even optically resolve details of an excitonic wavefunction in a QD. The range of applications of STED and RAP nanoscopy are likely to be complementary. While STED, STORM and all other incoherent nanoscopy3 have been proven to be essential tools for superresolution optical imaging in physiology, biology and medicine, it is unlikely that RAP nanoscopy will play a similar role in these fields. This is because RAP needs to
operate at low temperatures (4 K so far, but with prospects of reaching 50 K (ref. 6)) to preserve coherence. At these temperatures, cryo-electron microscopy plays an important role in physiological applications and gives, of course, far better resolution than STED. The strength of STED lies in low-invasive, roomtemperature imaging of (live) biomedical samples. In contrast, the straight-forward range of applications of RAP nanoscopy lies in the fields of quantum optics and solid-state physics. Controlling the state of a QD coherently with nanometre spatial resolution might become an important tool for fundamental investigations in quantum optics. Research on the physics of single QDs, colour centres or atoms might become easily possible even within a dense assembly of quantum emitters. ❐
Thomas A. Klar
Institute of Applied Physics, Johannes Kepler University Linz, Linz, Austria. e-mail:
[email protected]
Published online: 26 January 2018 https://doi.org/10.1038/s41566-018-0093-8 References
Hell, S. W. & Wichmann, J. Opt. Lett. 19, 780–782 (1994). Klar, T. A. & Hell, S. W. Opt. Lett. 24, 954–956 (1999). Hell, S. W. Science 316, 1153–1158 (2007). Moerner, W. E. & Kador, L. Phys. Rev. Lett. 62, 2535–2538 (1989). 5. Betzig, E. et al. Science 313, 1642–1645 (2006). 6. Kaldewey, T. et al. Nat. Photon. https://doi.org/10.1038/s41566017-0079-y (2018). 7. Melinger, J. S., Gandhi, S. R., Hariharan, A., Tull, J. X. & Warren, W. S. Phys. Rev. Lett. 68, 2000–2003 (1992). 8. Wu, Y. et al. Phys. Rev. Lett. 106, 067401 (2011). 9. Simon, C.-M. et al. Phys. Rev. Lett. 106, 166801 (2011). 10. Lüker, S. et al. Phys. Rev. B 85, 121302 (2012).
1. 2. 3. 4.
OPTOFLUIDICS
Instant trap formation In biomedical and biochemical research, traps in the channels of a microfluidic chip are often used to capture target microparticles or cells of interest. However, the usual form of hydrodynamic traps come with several limitations. First, many microparticles or cells are prone to bypass the trap structures because the hydraulic resistance of the microtraps is larger than that of the free microchannel, resulting in a low trapping efficiency (
