[Ack! Somehow the final draft of this post did not get saved (or posted) last night!]
After a weekend pause, we continue to discuss "Water/Surface Model/Biological Interfaces 1", whatever that means! I took notes on the first two spectroscopy talks of the day and a later one on dewetting.
Today, Stephen Meech (East Anglia, UK) gave a talk on "Time Domain Optical Kerr Effect (OKE) Studies of Aqueous Solvation: Ions to Proteins". This is the experiment that corresponds to measuring polarisability anisotropy correlation functions (which Branka Ladanyi talked about on Day 5). The measurement results from measuring the third-order electric susceptibility (chi^(3)) through a pump-probe experiment (the second-order susceptibility, chi^(2), which plays the starring role in sum-frequency generation, is zero in the bulk medium, so chi^(3) is the lowest-order non-linear effect in the bulk). The basic idea is to get two pump signals to polarize the sample (changing its refractive index), then use the probe signal to pass through this polarized material and measure the transmitted intensity. Meech's slides have all the details (there are many), and the slides will hopefully be on the WatSurf 2013 website within a few days. Essentially, OKE, is a time domain Raman spectroscopy. You get a plot of intensity vs. delay between pump and probe (ps). In a few fs, you get electronic transitions. Within 1/4 ps, you get librations. Then for delay times > ~ 1/4 ps, you see lots of vibrations. In the long time limit, you get diffusive reorientation. If you FT the signal, you get the low wavenumber Raman spectrum (< ~200 cm^-1).
Really, what you measure is the following response function (Branka Ladanyi is the person who has worked out how to read OKE signal from simulations):
R_ijkl^(3) ~ -beta d/dt < Pi_ij(t) Pi_kl(0) >,
where Pi is the sample's polarizability tensor. You can split Pi into molecular and intermolecular part. It's impossible to measure these separately, but you can get some information about the splitting through games with polarization of the pump and probe signal. In water, the molecular polarisability is essentially isotropic, so most of the OKE signal is due to interaction-induced translational dynamics. What do you get?
The FT of the OKE signal shows dynamics of H-bond modes: bending-like motions around 50 cm^-1, stretching-like motions around 150 cm^-1 (really very collective modes), some (weak) libration-like motions around 500 cm^-1. Back in the time domain, there's a long exponential decay tail for delay times between ~0.5ps to 3-4 ps. This is related to the relaxation of the H-bond network accompanying reorientation (his explanation made a connection to the Laage-Hynes jump model).
He did some OKE measurements in solutions of urea (hydrophilic) and TMAO (amphiphilic) to look at relaxation of water in their presence. Work at low mole fractions < 0.02 to avoid OKE signal having a large component due to the osmolytes themselves. Urea seems to change the H-bonding modes significantly, but TMAO barely changes it (even at 4M TMAO). One explanation might be that TMAO is forming micellar structures, and so most waters are really in a bulk-like environment. By looking at average relaxation time of water, and decomposing it as a linear combination of "bulk-like" relaxation and "solvation-shell-like" relaxation, weighted by molar ratio of solute to solvent (very hand-wavy), find that solvation water relaxes 3-6 times more slowly than in bulk water, more so in a hydrophilic solute (urea).
Summary of his conclusions: solutes generally retard dynamics in solvation shell; retardation greatest near hydrophilic solutes; H-bond structures (as seen in bimodal THz repsonse) is persistent, especially for hydrophobic solutes.
Now switch to hydration around peptides: NAGMA (hydrophilic everywhere, hydration shell has ~30-35 water molecules) and NALMA (hydrophobic leucine residue in the middle), hydration shell with 40-45 waters. What do you see? Water structure is disrupted when half water molecules have peptide as nearest neighbour. At high concentration (3M) of NAGMA, find very slow relaxation dynamics (but may be due to peptide itself). At low concentrations, can do the previous separation of bulk- and hydration-layer-like dynamics. Find solvation shell relaxation time of 12 ps for NAGMA, 7 ps for NALMA (were ~4 ps for urea and ~1ps for TMAO). They're 10-20 slower than bulk water. Perhaps due to multiple H-bond sites?
What about larger proteins? BSA, lysozyme, trypsin. For low concetrations of peptide (few % by wt), find relaxation times of 3-4 ps for water in solvation shells around these proteins. Ordering is consistent with higher hydrophilic surface areas leading to more retardation. There may be small populations of waters that are highly retarded, but the experiment is not sensitive to them.
Now for something completely different: Time-Resolved Fluorescence. You put a fluorophore at a particular place on a molecule. Then, three measurements are possible: a) time-dependent frequency-solvation dynamics, b) t.d. population-reaction dynamics; c) t.d. orientation-rotational dynamics. These tell you something about the environment around the fluorophore, but the caveat is that the fluorophore perturbs that environment. Solvation dynamics follows the usual story: water around unexcited fluorophore at equilibrium; fluorophore excited and acquires a dipole; solvation structure slowly adjusts to new fluorophore state; so energy difference between excited and ground state changes with time. Measure that time-dependent shift in fluorescence frequency: S(t) = nu(t) - nu(infty) / (nu(0) - nu(infty)). Reaction dynamics measures the total emission as the fluorophore is changing, R(t) = int_0^t dt' nu(t'); gives some indication of the dynamics of the fluorophore change. They have lots of optical tricks to measure fluorescence response to probe with fs accuracy and blurred by ~50 fs pulse widths. Since the pulse width is known, they make an attempt at deconvoluting it. Also, they can really measure the *spectrum* of the light at any delay time, not just its intensity.
First experiment: Auramine O (goes bright to dark based on a single bond twist) confined to a 1-10 nm radius sphere of water (inside a reverse micelle). Even in a 10 nm micelle, AuO reaction is slowed down enormously -> AuO is sitting at the micelle interface, not the bulk. As you increase confinement below 3 nm diameter, reaction slows down substantially => all the water in the micelle starts looking like "interfacial" water instead of bulk water. [These reaction dynamics experiments can be fit very well with simple 1D reaction-diffusion model.] Effect of surfactant cation: only charge density (singly-charged vs. multiply-charged cations) makes a difference, the size of the cations is not very important. What about a neutral surfactant ("Igepal 5")? Essentially the same result as for singly-charged cations.
Thomas Elsaesser (Max Born Institute, Berlin, Germany) "Ultrafast Hydration Processes of DNA and Phospholipids". Focus on 2DIR of water around DNA, then energy exchange between DNA and its hydration shell, as well as hydration of reverse micelles.
H-bonding sites in DNA, in reverse order of importance: 1) outer region of phosphates (by far strongest), 2) inner region of phosphate, 3),4),5) typical H-bonding sites in ribose oxygen and carboxyl/amide groups in bases. "Fully hydrated DNA" = about 20-25 water molecules per base pair. Summary of established results about DNA hydration: A) rigid spine of water in minor groove, more flexible hydration in major groove & phosphate groups. B) H-bond lifetimes in first longer longer than in bulk (few ps - 500 ps). C) Residence times of water molecules in first shells > 50 ps. D) Water reorientation: sterically hindered in minor groove, ps timescale in major groove. E) Highly controversial: "biological water" (outrageously rigid water, "100 ps - several ns" lifetime of some waters as measured by solvation dynamics experiments, A. Zewail PNAS 100, 8113 (2003)) -> controversial because slowing down is attributed entirely to water and not to DNA response [perhaps tenable at < 10 ps, but not for long times] and because the probe chromophore affects the water around the DNA.
To test this idea of biological water, use non-invasive probe: vibrational spectra. Typical vibration periods: OH / N-H stretch = 3000 cm^-1 = ~ 0.01 ps; H-bond mode <= 300 cm^-1 = ~0.1-0.5 ps. In a fluctuating electrical environment like that provided by surrounding waters, the energy levels between vibrational modes changes all the time, which leads to dephasing, spectral diffusion ~0.05 ps to several ps. Early MD simulations from Laage & Hynes tracked changes in, say, OH stretch frequency nu_1(t). Oscillates about the average value by about +/- 250 cm^-1, and is well explained by perturbation theory (?). The time-correlation function of nu_1(t) can be measured by 2D-FTIR. But not only do you have this spectral diffusion, you have couplings between different transitions. An excitation at nu_e leads to an emmision at nu_d, where a) if the waiting time is almost zero, nu_d =~= nu_e; b) spectral diffusion at long times leads to nu_d and nu_e being mostly uncorrelated; c) coupling between transitions shows up as "jumps" between values of nu_d and nu_e without much signal at intermediate frequencies.
He then explained how you actually measure 2D-FTIR spectra using Photon Echo Spectroscopy: so many pulses!!! Sequence of three pulses, with delay tau between first two, delay T between next two. Think of two states |i> and |f>. Pulse 1 generates a coherent superposition of 2 quantum states ('coherence' or polarization). Pulse 2 transform coherence into population. Pulse 3 generates coherence in the excited sample after population time T. Coherence radiates signal field.
What are the results? At short T and 31 C, you get a fast spectral diffusion of < 100 fs due to hugh-frequency librations and resonant energy transfer between different molecules. As you lower temperature, you get an overall slowing down of spectral diffusion, spectral diffusion at ~3300 cm^-1 is slower (i.e. longer correlations between excitation and detected frequencies without jumps due to coupled transitions). Vibrational lifetimes: OH stretch 200 fs, OH bend 170 fs, Librations < 100 fs. Energy dissipation ('hot ground state'): 0.7-1ps.
Now add DNA to the system. Use 23-bp DNA with alternative A-T-A-T-... sequences, with Na+ counterions replaced by surfactant molecules (CTMA) (if you just put DNA in 55 M water, the water spectrum completely dominates, so using a surfactant allows you to make thin films of "hydrated" DNA). DNA film is placed in a humidity cell, and which allows control of water content in film. The main change as a function of relative humidity (r.h.) is that the 3000-3500 cm^-1 region of the IR spectrum shoots up (at 0% r.h., there's still signal there owing to N-H stretches in DNA itself). More detailed features: PO_3 symmetric and asymmetric stretches (1000-1300 cm^-1) change. Asymmetric stretch shifts to higher wavenumbers at higher r.h. => water interacts strongly with phosphate group.
At this point, he showed *lots* of 2D-FTIR spectra of DNA with varying r.h. His main conclusions were: there is a very 'rigid' first solvation shell, with a slow resonant energy transfer due to reduced water concentration (10M). The water around the DNA forms an efficient heat sink for excitations in the DNA. In reverse micelles, there are very slow structural fluctuations in the first hydration shells.
Alenka Luzar (Virginia Commonwealth University, USA) talked about "Tunable Hydration at the Nanoscale". Talked a bit about dewetting and summarized the usual argument for critical distance between two plates for dewetting to happen (including pressure and solid-liquid attraction). Emphasized the kinetic bottleneck to dewet a very large gap between two hydrocarbon plates. The free energy barrier to dewet scales as (separation)^2 / cos(theta_contact) => spontaneous dewetting only occurs for molecular distances. Main point: hydrophobic interaction under kinetic control. Interesting factoid: the first simulations of dewetting in proteins weren't the melittin ones by Bruce Berne's group in 2005, but by Huang, Ding, Hua, Yang, Chen, J. Chem. Phys. 121 (4), 1969 (2004). Rationalize critical distance almost quantitatively for melittin dimers between Berne & Rossky (curved vs "flattened") melittin entirely in terms of hemispheric vs flat geometry of surfaces involved. Main point: the modified Kelvin equation works down to molecular length scales.
She then looked at how dewetting is affected by the presence of surface charges (after all, this is how nature does it in protein-protein interactions). What does an electric field do? Well, in the usual two-plate setup to analyse dewetting, it adds a bulk pressure-like energy term eps_0 |E|^2 / 2 to the wet state (electrostriction: an electric field can cause water to fill pores that it wouldn't otherwise enter), which reduces the critical distance to dewetting. This is a macroscopic argument. Microscopically, however, waters at a surface have a nonisotropic orientation distribution, so the reaction to parallel and perpendicular electric fields is different. For example, when you apply an electric field perpendicular to the normal of the plate surfaces, nothing special happens. But when you apply an electric field parallel to the plate normal, the water density becomes asymmetric: waters orient along direction of E, which costs a lot of bonds on one side (the one with higher electrical potential, which, therefore, looks hydrophobic) but not so many on the other side (which, therefore, looks relatively hydrophilic). In fact, the water exerts a *torque* on the walls: water would be happier if it didn't have to make a choice between aligning with the E-field and forming lots of H-bonds, and one way to achieve that is to rotate the *walls* so that the normal is perpendicular to the E-field. The size of this torque is about twice as large as predicted by a simple continuum theory (where the water and solute have different dielectric constants), owing to these microscopic surface effects. This suggest a way to align particles in water. Reaction times to "low" applied fields (0.02 V / A) is on the order of hundreds of ps for plates of a few nanometres in size.
Finally, some publicity for the Liquids 2014 - 9th Liquid Matter Conference:
"The 9th Liquid Matter Conference will take place at the University of Lisbon, Portugal, 21-25 July 2014. Previous conferences in this series were held in Lyon (1990), Firenze (1993), Norwich (1996), Granada (1999), Konstanz (2002), Utrecht (2005), Lund (2008), and Vienna (2011). The conference is organized jointly by the Liquids Section of the Condensed Matter Division of the European Physical Society, the University of Lisbon, and the School of Engineering of the Lisbon Polytechnic Institute (ISEL).
The conference will consist of plenary lectures, topical symposia with keynote lectures and contributed oral presentations, as well as poster sessions. The highlights of the conference will be published in a special issue of the Journal of Physics: Condensed Matter.
Topics
1 Ionic Liquids and Liquid Metals
2. Water and Solutions
3. Liquid Crystals
4. Polymers, Polyelectrolytes, Biopolymers
5. Colloids
6 Films, Foams, Surfactants, Emulsions
7. Confined Fluids, Interfacial Phenomena
8. Supercooled Liquids, Glasses, Gels
9 Driven Systems, Rheology and Nanofluidics
10. Active Matter
11. Biological and Biomimetic Fluids
For all relevant information please visit our website at http://www.fc.ul.pt/en/conferencia/liquids-2014
You can download the First Circular from http://www.fc.ul.pt/sites/default/files/fcul/public/firstcircl.pdf"
No comments:
Post a Comment